In the United States Court of Federal Claims
                               Sub-Master Docket No. 17-9001L

                                  (Filed: December 17, 2019)

 **********************************           )
 IN RE UPSTREAM ADDICKS AND                   )      Post-trial decision; government-induced
 BARKER (TEXAS) FLOOD-                        )      flooding on private property; application of
 CONTROL RESERVOIRS                           )      factors identified in Arkansas Game &
                                              )      Fish; liability for a taking of a flowage
 **********************************           )      easement
 THIS DOCUMENT APPLIES TO:                    )
                                              )
 ALL UPSTREAM CASES                           )
                                              )
 **********************************           )


         Daniel H. Charest and E. Lawrence Vincent, Burns Charest LLP, Dallas, Texas, Charles
Irvine, Irvine & Conner PLLC, Houston, Texas, and Edwin Armistead Easterby, Williams Hart
Boundas Easterby, LLP, Houston, Texas, Co-Lead Counsel for Upstream Plaintiffs. With them
at trial were Vuk. S. Vujasinovic, VB Attorneys, PLLC, Houston Texas, Lawrence G. Dunbar,
Dunbar Barder, P.L.L.C., Houston, Texas, Jack E. McGehee, McGehee, Chang, Barnes,
Landgraf, Houston, Texas, Michael J. Dulaney, Sullins, Johnson, Rohrbach & Magers, Houston,
Texas, Lydia A. Wright, Burns Charest LLP, Dallas, Texas, Mary Conner, Irvine & Conner,
LLC, Houston, Texas, Kyril V. Talanov, Houston, Texas, and Hilary S. Greene, Houston, Texas.

       William Shapiro, Trial Attorney, Environmental & Natural Resources Division,
United States Department of Justice, Sacramento, California, for defendant. With him at trial
and on the briefs were Kristine S. Tardiff, Laura W. Duncan, Sarah Izfar, Jessica Held, Bradley
L. Levine, David L. Dain, and Mayte SantaCruz, Trial Attorneys, Environmental & Natural
Resources Division, United States Department of Justice, Washington, D.C. With him on the
brief was Lawrence VanDyke, Deputy Assistant Attorney General, Environmental & Natural
Resources Division, United States Department of Justice, Washington, D.C.

                                   OPINION AND ORDER

LETTOW, Senior Judge.

       This case brings to the court the occasionally recurring question of the extent and the
nature of government-induced flooding on private property necessary to rise to the level of a
Fifth Amendment taking of a flowage eastment. Particularly, this post-trial decision assesses
whether the government may be liable to private property owners in the Houston, Texas
metropolitan area for takings compensation following Tropical Storm Harvey.1 Thirteen
property owners were selected to serve as bellwethers for the hundreds of property owners who
have filed suit raising similar claims against the government.

         After making landfall in August 2017, Tropical Storm Harvey (“Harvey”) doused
Houston with an average of 33.7 inches of rain over a four-day period. Many properties,
including over 150,000 homes, flooded during the storm. Those affected included private
property owners within the Addicks and Barker Reservoirs, west of Houston, upstream of the
federally designed, built, and maintained Addicks and Barker Dams. During Harvey, the
Addicks and Barker Dams collected storm water in their respective reservoirs causing properties
within the reservoir to flood from the impounded water. At issue in this bellwether trial is the
liability of the government under the Tucker Act, 28 U.S.C. § 1491, and the Takings Clause of
the Fifth Amendment of the Constitution, for the damage to thirteen of these properties.2

         The thirteen bellwether properties are representative of the hundreds of owners of
“upstream” properties who brought suit against the United States in this court after Harvey. The
property owners claimed that the United States was liable to them for an uncompensated taking,
that is, the government-controlled inundation of their properties by the impounded floodwater
from Harvey. The first complaint relating to Harvey and the Addicks and Barker Dams was filed
on September 5, 2017. See Y and J Props., Ltd. v. United States, No. 17-1189L. Hundreds of
such cases followed. Using case management techniques comparable to those employed in
multi-district litigation, the Chief Judge of the court issued Management Order No. 1,
consolidating these cases, and all related later-filed cases, within one master docket. See In re
Addicks and Barker (Texas) Flood-Control Reservoirs, No. 17-3000L; Y and J Props., Ltd. v.
United States, 

(2017). The Chief Judge then bifurcated the issues of liability
and damages, initially setting a schedule to deal with liability. See Order Regarding Judicial
Assignment and Scheduling (Nov. 20, 2017), Master Docket No. 17-3000L, ECF No. 70.
Subsequently, the Chief Judge divided the Master Docket into two sub-master dockets, see In re
Addicks and Barker (Texas) Flood-Control Reservoirs v. United States, No. 17-3000L, 

(Fed. Cl. Dec. 5, 2017)—one for downstream properties, In re Downstream Addicks
and Barker (Texas) Flood-Control Reservoirs, Sub-Master Docket No. 17-9002L, and, pertinent
here, one for upstream properties, In re Upstream Addicks and Barker (Texas) Flood-Control
Reservoirs, Sub-Master Docket No. 17-9001L.



       1
         When Harvey first made landfall on the Texas mainland on August 26, 2017, it was
classified as a Category 4 hurricane. See Eric S. Blake & David A. Zelinsky, Nat’l Hurricane
Center, Tropical Cyclone Report: Hurricane Harvey 3 (January 23, 2018), available at
https://www.hsdl.org/?view&did=807581. But “Harvey rapidly weakened over land to a tropical
storm” within the first twelve hours and to a tropical depression by August 30, 2017. 

majority of the five-day downpour that the Houston area experienced coincided with
Tropical Storm Harvey, the opinion will use this designation.
       2
         The named defendant, the United States, is representative of all relevant government
actors, including the United States Army Corps of Engineers (the “Corps”). Thus, references to
the “United States,” the “government,” and the “Corps” all refer to defendant and its collective
entities and actions.
                                                2
       Since the division, proceedings have moved apace in the upstream docket. In the spring
of 2018, thirteen plaintiff properties were designated to serve as bellwethers for the cases.3 In
February 2018, the government filed a motion to dismiss under the Rules of the Court of Federal
Claims (“RCFC”) 12(b)(1) for lack of subject-matter jurisdiction and 12(b)(6) for failure to state
a claim upon which relief can be granted. See Motion to Dismiss (Feb. 16, 2018), Sub-Master
Docket No. 17-9001L, ECF No. 59. Although the court made some preliminary rulings in
addressing that motion, resolution of the government’s motion to dismiss was deferred until trial,
pursuant to the court’s authority under RCFC 12(i), in light of the fact-intensive inquiry this case
required. See In re Upstream Addicks and Barker (Texas) Flood-Control Reservoirs, 138 Fed.
Cl. 658, 672 (2018).

         A ten-day trial was held in Houston, Texas, commencing on May 6, 2019, regarding the
liability of the United States for the thirteen test properties. During the course of trial, on the
afternoon of May 8, 2019, the court conducted a site visit of the dams that included the spillways
and end points of both dams as well as the drainage canals that feed into the reservoirs, as well as
seven of the test properties. Following post-trial briefing, see Plaintiffs’ Post Trial Brief (“Pls.’
Br.”), ECF No. 235; Defendant’s Post Trial Brief (“Def.’s Br.”), ECF No. 242; Plaintiffs’ Post
Trial Brief Reply (“Pls.’ Reply”), ECF No. 246, the court heard closing arguments on September
13, 2019, in Washington, D.C. The issue of liability pertaining to the thirteen test properties and
the government’s motion to dismiss are ready for disposition.

       Overall, based on the facts and circumstances at hand, the government’s motion to
dismiss is denied and the court finds the government to be liable for a taking of a flowage
easement on the properties.

                                             FACTS4

                        A. The Addicks & Barker Flood Control Project

       1. The impetus for flood control measures.

        Buffalo Bayou originates in eastern Waller County and western Harris County and flows
in a generally eastward direction through a circuitous channel approximately 75 miles long.



       3
         Originally, the court directed the parties to come to an agreement on ten test properties.
See Case Mgmt. Order (Feb. 1, 2018) at 2-3, Sub-Master Docket No. 17-9001L, ECF No. 37. At
the behest of the parties, the number was later increased to fourteen properties. See Order
Approving Test Prop. Selection (Mar. 13, 2018), Sub-Master Docket No. 17-9001L, ECF No.
91. One of these fourteen plaintiffs voluntarily dismissed his suit, see Notice of Voluntary
Dismissal (Aug. 24, 2018), Sub-Master Docket No. 17-9001L, ECF No. 136, and therefore
thirteen plaintiff test properties remain.
       4
        This recitation of facts constitutes the court’s principal findings of fact in accord with
RCFC 52(a). Other findings of fact and rulings on questions of mixed fact and law are set out in
the analysis.

                                                 3
Joint Stip. ¶ 84.5 After its confluence with South Mayde Creek in western Harris County, the
bayou winds through downtown Houston, where it converges with the White Oak Bayou and
continues east, eventually reaching the Houston Ship Channel and pouring into San Jacinto Bay,
Galveston Bay, and the Gulf of Mexico. Joint Stip. ¶¶ 84, 85. The city of Houston is situated at
the confluence of the two bayous and at the base of a fan-shaped system of streams that flow
through a flat and “almost featureless plain,” contributing to the creation of a major flood hazard
in the region. JX5 at 4.6 For much of the year, little or no water flows through the narrow
streams in the Buffalo Bayou watershed, but during heavy rainfall the small stream channels
cannot hold the water and “a general overflow along the banks” results. JX5 at 5. The soil
composition in the region and the foliage it supports do not allow much water seepage and result
in exceedingly poor natural percolation and drainage. See 

Close proximity to the Gulf
of Mexico yields regular hurricanes and tropical storms, rendering the region susceptible to
especially heavy rainfall events and attendant flooding. See Tr. at 614:7-23 (Test. of Jeff
Lindner).7 Due to this combination of factors, between 1854 and 1935 six major floods
occurred in the Buffalo Bayou watershed, including the City of Houston. JX5 at 6. Two
particular storms and the flood devastation they created, one in May 1929 and the other in
December 1935, prompted congressional action that led to the construction of the Addicks and
Barker Dams. Joint Stip. ¶ 81.

        In May 1929, a storm (“the 1929 storm”) produced rainfall ranging from six to twelve
inches over the White Oak Bayou and Buffalo Bayou basins, causing flooding in downtown
Houston and resulting in property damage within the city of $1,392,000. Joint Stip. ¶ 82. About
six years later, in December 1935, another storm (“the 1935 storm”) produced three days of
rainfall dropping approximately fifteen inches throughout the basin. Joint Stip. ¶ 83. That storm
generated flooding that killed eight people and caused property damage estimated at $2,528,000.
Joint Stip. ¶ 83.

        The real possibility of a storm even larger than these events raised serious concern. The
Corps noted in its 1940 Definite Project Report about the then-proposed dams that the Buffalo
Bayou watershed was situated “in an area subject to all of the circumstances making possible
large storms.” JX5 at 7.8 That same report stated that had the 1935 storm centered directly over


       5
          On April 23, 2019, the parties entered into stipulations of fact “for the purposes of the
trial of the thirteen claims in the Upstream Sub-Docket only, and [specifying that the
stipulations] are not intended to be binding as to any other claim that falls within Master Docket
No. 17-3000L, the Upstream Sub-Docket No. 17-9001L, or the Downstream Docket No. 17-
9002L, or in any other action or proceeding.” Stipulations of Fact for Trial at 1, ECF No. 211.
The stipulations number 116 and will be cited as “Joint Stip.” followed by the paragraph number.
       6
         Citations to plaintiffs’ exhibits are identified as “PX___,” defendant’s exhibits are
identified as “DX___,” and the parties’ joint exhibits are identified as “JX___.”
       7
        The transcript of the trial will be cited as “Tr. ___:___;” showing the pertinent page and
line number and the name of the pertinent witness.
       8
       The Definite Project Report was issued in 1940 in connection with plans to construct the
Addicks and Barker Dams.
                                              4
the basin—rather than where it did over Westfield, Texas, about eighteen miles from Houston—
it would have resulted in even more severe flooding. See 

there any “evident
meteorological reason why the storm could not have centered over the basin.” 

also
recognized at the time that, due to the topographic and meteorological features of the region, any
flood control system constructed for Houston could be subjected to storms equal to any of record
in Texas, and greatly in excess of any so far experienced over the basin. See 

the Corps observed, had “prevented the occurrence of a storm over the basin much
larger than the 1935 storm.” 

rainfall of record in the United States at the time
Addicks and Barker were constructed occurred only ninety miles northwest of Houston at
Hearne, Texas in 1899—under meteorological conditions that the Corps noted “could be
approximated closely over the Buffalo Bayou watershed.” 

storm generated a
maximum 31.4 inches of rain in a period of three days, with an average depth over an area of
1,000 square miles of 25.8 inches. PX777 at 4. A hydrology report prepared by the Corps in
1938 concluded that the 1899 Hearne storm was the “maximum probable storm” that might arise
over the Buffalo Bayou watershed, also noting that should such a storm occur the average
rainfall would be “almost twice the average of 15 inches that produced the record flood of 1935.”
JX5 at 8. While conceding that the probability of the occurrence of a storm as severe as the
Hearne storm in the Buffalo Bayou basin was “very remote,” the Corps also noted in its Definite
Project Report that “ultimate protection against such a storm is desirable” though perhaps not
feasible in “the initial stage” of flood control construction. 

Without flood control
measures, the Corps predicted “[c]onsiderable overflow” from storms that would produce
“disastrous peak flows.” 

        Another storm that occurred at Taylor, Texas in 1921 (“the Taylor storm”) constituted the
greatest single-day rainfall ever recorded in the United States at that time, producing 23.11
inches in 24 hours. PX777 at 4. The Taylor storm did not cover as large an area as the Hearne
storm, but the rainfall was more intense over a shorter period of time and the Corps noted in the
1940s that such depths of accumulated rain over a basin as small as Buffalo Bayou would be
considerable. 

        Against this background, Congress acted to initiate the implementation of flood control
measures. Pursuant to the River and Harbor Act of 1938, Congress authorized the Corps to
design and build the Addicks and Barker Dams as part of the Buffalo Bayou and Tributaries,
Texas Project. See Pub. L. No. 75-685, 52 Stat. 802 (codified mainly at 33 U.S.C. §§ 540,
558(b), 558(c), 571, 701(k)). The purpose of the project, as defined by the Corps’ Definite
Project Report published on June 1, 1940, was “to provide for complete control of floods on the
Buffalo Bayou watershed and the protection of the city of Houston, Texas, and the Houston Ship
Channel against the estimated probable maximum storm.” JX5 at 3.

        At critical junctures in the ensuing lifespan of the dams, the Corps consistently echoed
that the whole purpose for the construction and operation of the project was to prevent
downstream flooding, especially in downtown Houston. See, e.g., PX59 at 8 (USACE464077)
(“The sole authorized purpose for [the] Addicks and Barker Reservoirs is to reduce potential
flood damage along the downstream reach of Buffalo Bayou.”); PX59 at 21 (USACE464090)
(“The dams are operated strictly to prevent downstream flooding.”); JX110 at 3-1
(USACE016311) (“The existing project, as authorized, provides for flood risk management, the
protection of the City of Houston from flood damages, and the prevention of excessive velocities
and silt deposits in the Houston Ship Channel Turning Basin.”).
                                                   5
       2. Project design.

        The original design of the project consisted of three detention reservoirs, a system of
canals and levees, and channel improvements along Buffalo Bayou below the reservoirs. See
JX5 at 12-13. The three detention reservoirs were to be built on White Oak Bayou and at the
Addicks and Barker watersheds on Buffalo Bayou, which are seventeen miles west of downtown
Houston and upstream of the confluence of Buffalo Bayou and South Mayde Creek. Id.; Joint
Stip. ¶ 100. The Barker Dam would be located on Buffalo Bayou about 1.5 miles above its
confluence with South Mayde Creek and the Addicks Dam would be located on South Mayde
Creek about one mile above its confluence with Buffalo Bayou. See JX110 at 4-1
(USACE016316). North of and adjacent to the Addicks Reservoir watershed lies the 130 square
mile watershed of Cypress Creek, which flows in an eastward direction toward its outlet into the
San Jacinto River. See 

overflow from the Cypress Creek watershed into the
Addicks Reservoir, an upstream levee was to be built. JX5 at 13. Additionally, approximately
7.4 miles of the Buffalo Bayou channel immediately downstream of the Addicks and Barker
Dams was to be rectified and enlarged. JX110 at 3-3 (USACE016313). This channel
rectification and enlargement was completed in 1948. 

the original design, however, were not completed. Tr. 473:8-18 (Test. of
Robert Charles Thomas, III). Neither the detention reservoir on White Oak Bayou nor the
Cypress Creek levee were ever built, nor was a south canal that would divert the surcharge
releases into the Houston Ship Channel. Tr. 191:16 to 193:5 (Thomas). Notably, the failure to
complete the reservoir on White Oak Bayou has apparently had little effect on the Addicks and
Barker Reservoirs, see Tr. 193:19-21 (May 6, 2019) (Thomas), but because the Cypress Creek
levee was never completed, run-off can still flow from the Cypress Creek watershed into the
Addicks watershed during major rain events, increasing the size of the flood pool in the Addicks
reservoir, see Tr. 1539:3-11 (Test. of Richard Long). And, importantly, the deletion of the south
diversion canal was also a major change from the original design because surcharge releases
from the Addicks and Barker Reservoirs now have no place to go except down Buffalo Bayou or
in the reservoirs themselves. Tr. 193:23 to 194:7 (Thomas).

         The completed Addicks and Barker Dams are parallel u-shaped earthen embankments
that rise almost imperceptibly over a distance of miles. See JX15 at 9-10 (noting that the
Addicks and Barker embankments slope at a rate of about two to seven feet per mile). The size
of the reservoir embankments was determined by reference to the rainfall produced by two
previous storms: the Hearne storm and the Taylor storm. The Hearne storm, modified to account
for the rainfall intensity rates in the Taylor storm, was used as “a basis of design,” the so-called
“design storm.” PX777 at 5. The design storm would produce a maximum rainfall depth of 31.4
inches and served as the basis for the specifications of the dams that were ultimately
constructed—meaning that the dams were built to contain the amount of water the design storm
was projected to produce. 

the Barker Dam and Reservoir began in February 1942 and finished in
February 1945. Joint Stip. ¶ 95. The Barker Dam consists of an earthen embankment that
measures approximately 13.6 miles long and rises 36.5 feet above the stream bed at its highest
point. Joint Stip. ¶ 97; JX23 at 3 (USACE318524). Its outlet at the time of completion consisted
of five conduits which were 9-feet wide by 7-feet high and 190.5-feet long. JX23 at 3
(USACE318524). The Corps began construction on the Addicks Dam, located just north of the
                                               6
Barker Dam, in May 1946. Joint Stip. ¶ 90. Completed in December 1948, the Addicks Dam
consists of an earthen embankment that measures approximately 11.6 miles long and rises 48.5
feet above the stream bed at its highest point. Joint Stip. ¶¶ 90, 92; JX23 at 3 (USACE318524).
When completed, its outlet consisted of five conduits which were 8-feet wide by 6-feet high and
252-feet long. JX23 at 3 (USACE318524). The dams were designed to release water through
these outlet conduits, which could be controlled by gating, making it possible to limit discharges
from the reservoirs and thereby reduce downstream flooding. JX22 at 1-2. The original design
of both dams called for four of the five outlet conduits to be uncontrolled. 

embankment design of the Addicks and Barker Dams required the
government to acquire land behind (upstream of) the dams, thus partitioning off the reservoirs
that would hold the water held back by the dams. Behind the Addicks Dam, the United States
acquired all land at and below an elevation of approximately 103 feet, which amounts to about
12,460 acres of property. Joint Stip. ¶¶ 94, 102. For the Barker Dam, the government acquired
all land at and below an elevation of approximately 95 feet, amounting to 12,060 acres. Joint
Stip. ¶¶ 99, 104. The Corps calculated the amount of land it purchased behind each dam by
adding “[three] feet above the pools which would be produced by the 1935 storm transposed
over each watershed.” JX5 at 26. At the time, the property behind the dams was almost
exclusively used for ranching and rice farming. See Tr. 455:4-19 (Thomas). The government
purchased much of it at prices between five to ten dollars per acre, and “[m]ore lands could have
been purchased upstream for reservoir storage at relatively low prices, but urban development
was not anticipated in this baron [sic] prairie land remote from Houston.” JX52 at 17
(USACE015146).

         Significantly, the Corps calculated the amount of land it purchased based on a historical
storm metric (the 1935 storm) that was different—and, notably, smaller—than the design storm
metric (which combined the Hearne and Taylor storms) it used for engineering the dam
embankments. See Tr. 199:12 to 200:25 (Thomas). In other words, the embankment design
storm would generate more water than would the land-acquisition model storm. Put simply, the
dams were designed to contain more water than the acquired land could hold. These differing
metrics were not an oversight; rather, they were driven by a calculated decision. The Corps
noted at the time that storms of intensities similar to the 1935 storm—which was used for
calculating the land acquisition—were “expected” to “occur several times during the lives of
these structures.” JX5 at 26. While observing that the land purchased was inadequate to contain
the pool elevation which would be produced by the embankments’ design storm, it noted that
“[a]lthough the design of the embankments is based upon the design storm rainfall of 31.4
inches, the occurrence of such a storm in the basin [can not] be expected to occur more than once
in the lives of these structures.” 

viewed the Hearne storm as representing the
upper limit of possible storms that could occur in the region, see Tr. 1029:13-20 (Thomas), so
when setting the land acquisition line it looked for a large storm “but not the worst ever, because
[then the Corps] wouldn’t be able to afford all [its] projects,” Tr. 1060:22-25 (Thomas). If,
however, the Corps did not design the project to survive the upper limit storm and such a storm
did occur, it would create the possibility that the dams “could [] fill[] all the way up and
catastrophically fail[].” Tr. 1061:7-10 (Thomas). While the Corps designed the project
consistent with a much larger design storm, it considered it “unnecessary to acquire lands to the
pool elevation which would be produced by the design storm.” JX5 at 26.


                                                7
        Several documents dating from 1938 provide further insight into the Corps’
contemporary understanding of the likely recurrence of a storm akin to the Hearne storm.9 In
one document evaluating an alternative flood control project that was never authorized or built
(the so-called Triple Corridor Plan), the Corps observed that “[t]ransposition of [the Hearne
storm] to the Houston area does not appear to be unreasonable” because such a storm “has
already occurred but a short distance away.” Pls.’ Mot. to Reopen the Trial R. Ex. A at
USACE2019_0000014. The Corps stated that it considered such a storm “likely to occur with a
frequency of once every 50 years.” Id.10 Likewise, the Galveston District engineer’s office, in
another of these 1938 documents, recognized that the primary meteorological criteria required
for the occurrence of significant storms were satisfied in the Houston area, cautioning that “the
susceptibility of the Buffalo Bayou area to a storm as great as the 1899 storm must be considered
in designing any flood control works in the Houston area.” 

at USACE2019_0000252.

        Ultimately, in 1940, after conducting a cost-benefit analysis, see generally JX52, the
Corps concluded that “[a]cquisition to a taking-line, [three] feet above the computed pool
elevations for the 1935 storm centered above each reservoir, [wa]s considered advisable, since
the savings in annual interest would be in excess of the probable damage from storms producing


       9
         These documents did not come to light until several months after trial, through no
apparent fault of either party. When they were discovered, the Department of Justice transmitted
the documents to plaintiffs in the interest of transparency, all the while maintaining that it was
not legally obligated to do so. Upon receipt, plaintiffs moved to reopen the trial record to
include the documents as additional evidence. See Pls.’ Mot. to Reopen the Trial R. to Include
Additional Evidence from Late-Produced Documents, ECF No. 245 (“Pls.’ Mot. to Reopen the
Trial R.”). The government opposed the motion, asserting that these documents had not been
sought during discovery and that they lacked probative value because they concerned a flood
control plan that was never authorized or constructed and were preliminary parts of an iterative
review and analysis process. See Def.’s Opp’n to Pls.’ Mot. to Reopen the Trial R. to Include
Additional Evidence at 4-5, ECF No. 254 (“Def.’s Opp’n to Mot. to Reopen the Trial R.”). The
government also moved contingently, if the court were to admit the documents, that the court
also admit a sworn statement of Mr. Robert Thomas providing “critical context” for the
documents. Def.’s Opp’n to Mot. to Reopen the Trial R. at 2.

        The decision of whether to reopen the record to submit additional proof is within the trial
court’s discretion. See Zenith Radio Corp. v. Hazeltine Research, Inc., 

, 331
(1971); Confederated Tribes of the Warm Springs Reservation of Or. v. United States, 101 Fed.
Appx. 818, 822 (Fed. Cir. 2004). Upon consideration of the probative value of the evidence and
the circumstances of its identification, the court GRANTS plaintiffs’ motion to include the
additional evidence. To prevent any undue prejudice, defendant’s motion to admit the statement
of Mr. Thomas providing context for the documents is also GRANTED.
       10
         As noted in the statement of Mr. Thomas regarding these documents, this 50-year
prediction as used by the Corps at the time does not necessarily refer to the expected return
frequency of the storm, but to the 50-year planning horizon for the project. In other words, it
means that the Corps believed the Hearne storm “was the worst rainfall that could occur during
the 50-year planning horizon for the [Triple Corridor Plan] project.” Def.’s Opp’n to Mot. to
Reopen the Trial R. Ex. A ¶ 11.
                                                8
pools greater than the taking-line limits,” JX5 at 26-27. The decision to acquire less land than
that required to store the runoff contained and controlled by the dams was “considered an
acceptable low-frequency risk because of the relatively remote rural project location.” JX52 at 5
(USACE015134). Thus, the Corps based its land-purchase decision, at least in part, on a
calculation that “the expected damages of inundating pastures and rice fields” would be less than
the cost of buying additional land. Tr. 200:21-24 (Thomas).

        After its initial land purchase based on the 1935 storm calculations, the government made
one last additional purchase at some time before 1945 during the preconstruction planning for the
Addicks Dam. JX22 at 2. By that time, it had become evident that the levee on Cypress Creek
would not be built, and the Corps determined that it would “be more economical to increase the
capacity of Addicks Reservoir to accommodate Cypress Creek overflow and delete the diversion
levee.” 

end, the Corps acquired an additional three to four vertical feet of property in
the Addicks watershed to contain the anticipated overflow from Cypress Creek. See Tr. 474:11
to 475:7 (Thomas).

                        B. Post-Construction Improvements and Operations

       1. Dam modifications, evaluations, and safety reviews.

           a. 1940s, 1950s, and 1960s: Additional gates added to both dams.

        Due to development in the Houston area and opposition concerned with aesthetic effects,
the original plans for the south discharge canal were shelved and the canal was not constructed.
See JX22 at 2; PX42 at 1-2 (USACE541550-1); see also Tr. 208:4-11 (Thomas). This led to
concerns about a potential flood threat in downtown Houston, as the area adjacent to Buffalo
Bayou saw significant urban development during the 1940s and 1950s, and the Bayou could no
longer sustain a large uncontrolled flow without flooding. See JX22 at 2. The original design of
the Addicks and Barker Dams called for the inclusion of five outlet conduits at each dam, with
four of the five allowing uncontrolled water flows. JX22 at 1-2. The fifth conduit was
controlled using a gate included originally for emergency purposes. Tr. 197:20 to 198:6
(Thomas). With four conduits uncontrolled on each reservoir, a combined uncontrolled
discharge of about 15,700 cubic feet per second would flow into Buffalo Bayou. JX44 at 4. To
alleviate what appeared at the time to be temporarily, any possible resulting issues in Buffalo
Bayou, two additional gates were installed on the conduits, marking a total of three gated
conduits out of five in each reservoir. See JX22 at 2-3. By the time these additional gates were
installed in 1949, the combined uncontrolled discharge from the reservoirs in the four total
remaining conduits was 7,900 cubic feet per second, which was considered to be the maximum
channel capacity at the time for Buffalo Bayou. See JX44 at 4; JX16 at 4. Additional
development around Buffalo Bayou led to a lower calculated maximum channel capacity, and as
a result, the two remaining uncontrolled conduits in each reservoir were gated in 1963. See JX22
at 2; JX16 at 4. This gating, while beneficial for the downstream protection of downtown
Houston, also forces the prolonged storage of water in the reservoirs and increases each




                                                 9
reservoir’s pool size. See JX15 at 44; JX16 at 4.11 After all the gates were added, the Corps
calculated that the maximum design spillway in Addicks and Barker was 114.6 feet (Addicks)
and 106.4 feet (Barker), exceeding the government-owned land by 6.6 vertical feet in Addicks
and 8.1 vertical feet in Barker. See JX16 at 1.

            b. 1970s: Land development, outgrants, and increased concerns.

        “Beginning in the late 1970’s, private land developers upstream from the reservoirs
sought permission to extend channel improvements onto [government-owned land].” JX52 at 15
(USACE015144). Around 1981, the Corps began granting several of these easement requests,
limited to a maximum flow capacity. 

upstream tributaries flow into
Addicks and Barker extending onto government-owned land as a means to facilitate the
movement of water off upstream property. Tr. 383:9-15 (Thomas); Tr. at 868:20-23 (Test. of
Johnson-Muic). These easements—known as “outgrants”—while beneficial to the development
of the upstream area generally, had a number of negative consequences for the Addicks and
Barker Reservoirs. These effects included: (1) “increase[d] [] inflow of sediment into the
reservoirs resulting in the loss of flood storage capacity,” JX52 at 15 (USACE015144); (2)
“faster flood runoff into the reservoirs . . . [that] would result in more frequent impoundments,”

(USACE015145); (3) “larger impoundments,” id.; and (4) “increase[d] flood damages
resulting from reservoir impoundments,” 

consequence of the outgrants, the late 1970s and early 1980s saw the beginning of
rapid urbanization in the upstream areas, as the population of the Houston metropolitan area
grew. The character of the land upstream of the Addicks and Barker Dams was beginning to
shift from that of barren grazing lands and rice fields to a more urban-suburban development.
See generally JX52 at 17 (reporting in 1995 how over the last 50 years, “extensive urbanization
has occurred” in the upstream area). Additionally, scientific advances in hydrology and
meteorology prompted the Corps to issue new policies and procedures pertaining to the
determination of spillway capacities. See generally PX34 (referencing updated criteria such as
1966 Engineer Circular No. 1110-2-22).

        These developments, i.e., advances in science, changing meteorological forecasts, general
wear and tear on the reservoirs, and upstream urbanization, collectively raised concerns with the
Corps that flooding beyond the extent of government-owned land was highly probable, if not
inevitable, during a severe storm. In 1973, the Corps lamented that the possibility of flooding
lands in the reservoirs beyond the government-owned land was soon expected to become a
public issue. See PX37 at 1. In a 1973 memorandum from the Corps’ Engineering Division
Chief in the Galveston, Texas area, the Chief noted, “It is suggested that the project engineer
research the background of the existing situation and develop a history and rationale for our
operating concept of imposing flooding on private lands without benefit of flowage easement or
other legal right.” 

Corps inspection report echoed similar thoughts. See PX38 at 5
(USACE233674) (1974 Buffalo Bayou Inspection Report) (“Development of the area will
eventually place the Government in the position of having to flood the area within the reservoir


       11
          The Corps, in a 1960 study, had even recommended not adding additional gates, noting
that “[t]hese gates would only provide a negligible amount of increased flood protection” and
“would have more undesirable effects than benefits.” JX15 at 44.
                                                10
with the accompanying damages in order to protect downstream improvements in the event of a
severe future storm.”).

           c. Late 1970s: Hydrology investigation.

        Based upon the increased scrutiny, the Corps completed an extensive hydrology study on
Addicks and Barker in 1977. See JX23 (1977 Hydrology Report). The study was deemed
necessary “because it [was] apparent that urbanization of the subject watersheds will soon reach
levels in excess of those considered in the original design[,] and updated hydrologic criteria
prescribe more severe design standards than those addressed in the original.” 

(USACE318522). The 1977 report calculated a dramatic increase to the maximum design
spillway, see Tr. 497:15-21 (Thomas), and developed a higher probable maximum precipitation
value, see Tr. 499:8 to 500:2 (Thomas). Both of these circumstances reflect the notion that then-
current calculations (in 1977) for the possible amount of rain in the Addicks and Barker
watershed and the potential size of the reservoir pools in each dam were significantly higher than
originally calculated when the dams were built. The report also revealed serious safety issues
with the dams, putting both upstream and downstream properties at risk. See Tr. 257:21 to 258:1
(Thomas).

           d. 1980s: Embankment strengthening & dam modifications.

         These concerns led the Corps to consider various measures. The first major modification
of the dams came as a result of the permanence of the conduit gates. The Corps noted that one
negative result of the added conduit gates was the prolonged storage of rainfall runoff behind the
dams that resulted in the need for “emergency seepage control measures” at the pervious sections
of the embankments and foundations of the dams. JX44 at 4. This seepage threatened the
stability of the embankments and created a potential for failure of the dams in the event of a high
reservoir pool. JX15 at 44. Emergency modifications to strengthen the embankments of the
reservoirs were completed between 1977 and 1979. JX44 at 4. These improvements, however,
were not the final time the embankments were modified.

        Around the time of the completion of the improvements addressing seepage, the Corps’
concerns about the current embankment heights as not “safe” in both Addicks and Barker grew.
PX42 (1978 Water Control Plan) at 1 (USACE541550). Specifically, the report noted that a
recent study showed that “the spillway design flood would overtop the dam embankment with
possible embankment failure” for both Addicks and Barker. PX42 at 2 (USACE541551)
(Addicks), 2 (USACE541562) (Barker). Additionally, the report noted that scientific advances
showed an even higher increased maximum size for the reservoir pool in each reservoir, both of
which would exceed the reservoirs’ respective capacities. 

a reservoir pool of
118.1 feet was possible under existing conditions in Addicks and a pool of 110.3 feet was
possible under existing conditions in Barker).

        In the 1980s, the Corps considered a number of potential solutions to address these
concerns. Seventeen alternative plans were considered, seven of which were developed more
thoroughly. See PX51 at 5 (USACE013572) (1984 General Design Memorandum). These
options included taking no action, degrading the ends of the dams to reduce maximum flood
pools, diverting water to other channels, and raising and extending the embankments to impound
more water. 

which consisted of increasing the existing spillway capacities and raising

                                                11
low portions of the dam crests, was recommended first, see JX26 (1980 Corps Mem.), and
included a plan to consider the purchase of real estate upstream of the reservoirs, see 

however, deferred any decision to purchase upstream real estate. See PX48 (1980 Corps
Telephone Record).

        Consistent with the original purpose of the project, a 1981 Corps environmental
assessment highlighted the Corps’ focus on alleviating the risk for downstream flooding. See
generally PX87. Emphasizing the great risk downstream to Houston, the report deemphasized
the upstream risk by simply noting that “should this [standard size] storm occur, flooding would
extend beyond the [g]overnment owned land upstream of the embankments.” PX87 at 4
(USACE012909). The report also explained that “[t]he inadequacy of [g]overnment owned land
upstream . . . to contain the water from the [standard size storm] was recognized in the original
design of the reservoirs. However, it was considered at the time to be a limited problem, because
the land’s primary use at that time was for agricultural purposes and any damages . . . would be
infrequent and relatively minor.” 

Corps concluded that “[t]his problem does
not affect the safety of the dams.” 

potential risks upstream and because of the grave consequences
downstream of dam failure, see Tr. 89:22 to 90:8 (Thomas), the Corps adopted a new alternative
plan, Plan V(b), to raise the main embankments and to add additional erosion protections to the
dams’ auxiliary spillways, see PX51 at 8 (USACE013575) (1984 General Design Mem.). This
work was completed in the late 1980s and did not increase the effective storage of the dams. See
PX2284 at 2 (FB0000633) (1989 Mem.).

           e. 1990s: Storms in the area and subsequent evaluations.

         In the period leading up to and during March of 1992, a series of storms resulted in then-
record flood pools in both the Addicks and Barker Reservoirs. See Tr. 363:20 to 364:2
(Thomas). A large portion of Houston suffered major flooding, and public concern for flooding
of privately-owned land inside the reservoirs grew. See JX52 at II-1 (USACE015195) (1995
Reconnaissance Report). As a result, the Corps prepared a special report in May of 1992 to
provide general background and an overview of anticipated flooding damages which could occur
beyond government-owned property in Addicks and Barker. JX44 at 1 (USACE015073). One
conclusion from the report calculated that, “[T]he Possible Maximum Flood would affect over
4,000 structures valued at approximately $725 million and cause damages of $245 million.” 

(USACE015081). The report detailed a number of options to consider as potential solutions
for this upstream flooding problem, see 

(USACE015082-88), and suggested further
evaluation of the options for their economic, environmental, and engineering feasibility, see 

(USACE015088). Concerned for public safety, one potential option included the creation
of a public awareness program, noting that, “In the absence of a public awareness program,
residents are likely to forget or ignore the flood threat. Turnover in home ownership could also
result in a significant proportion of residents being unaware of the risk.” 

1
(USACE015083).

        The year 1995 saw the completion of the Corps’ Addicks and Barker Reconnaissance
Report evaluating the options, ordered after the completion of the 1992 study. See generally
JX52. The area surrounding the government-owned land was now “densely populated” and full
of “residential and commercial urban developments.” 

(USACE015136). After
                                                12
considering the various recommendations, including taking no action, purchasing flowage
easements, land buyouts, channel enlargements, excavating the government-owned land, and/or
adopting a flood warning and evacuation plan, see 

(USACE015136-37), the Corps
decided to take no action upstream, finding “insufficient economic benefits to justify project
modification,” 

(USACE015148).

           f. 1990s and 2000s: Home elevation surveys.

        Throughout the 1990s and 2000s, the Corps completed surveys of capital investments
located within the standard project floodplain for the purpose of determining potential flood-
damage estimates. See JX52 at 4 (USACE015224) (1995 Reconnaissance Report Appendix).
The surveys looked at homes, businesses, and other structures, see Tr. 390:20 to 391:2 (Thomas),
with over 95% of the structures inventoried in each reservoir being residential, see JX52 at 4
(USACE015224). “Information recorded during the field survey included the location of
structures (i.e., street address), ground elevations of structures, the flooding threshold of
individual structures[,] and structure category types.” 

words, the government was
aware of where and at what elevation water could or would enter each property. See Tr. 100:5-
10 (Thomas). One such field study was conducted and completed in July 1994. JX52 at 4
(USACE015224). Another study was completed in the early 2000s, when the Corps hired a
private contractor to do elevation surveys for over 10,000 structures in the potential
impoundment area. See Tr. 100:11-16 (Thomas). The Corps also prepared internal “Reservoir
Structure” maps that depicted the elevations of these surveyed upstream structures. See PX268.
As a result, the government gained an appreciation of the specific risks upstream in Addicks and
Barker associated with a severe storm.

       2. Community engagement about proximity to the dams.

       Long before Harvey occurred, information about the possibility of flooding upstream of
the dams and beyond the borders of the government-owned land was well known to the Corps
and accessible by the public. At a basic level, it could be obtained in publicly available maps.
Moreover, the Corps had discussed upstream flood risks with developers in the 1980s and 1990s.
Harris County began warning the community about flood risks years before Harvey, and Fort
Bend County began including warning language in upstream subdivision plats in the early 1990s.
The Corps also engaged in public outreach efforts to inform the community about the risk of
upstream flooding.

           a. Publicly available maps.

        Several types of publicly accessible maps graphically illustrated the flood threat long
before Harvey occurred. For example, the Houston and Harris County Atlas Key Maps contain
information enabling an astute map reader to understand the elevation of maximum flood pools
at each reservoir, see DX795, and Key Maps are prevalent enough to be referenced by page on
real estate listings in the Houston area, see, e.g., JX76.

        Likewise, the Federal Emergency Management Agency (“FEMA”) prepares Flood
Insurance Rate Maps, showing areas expected to flood during storms of various intensities, based
on data generated by a hydrologic modeling program. See Tr. 1904:12-21 (Test. of Philip
Bedient). The maps denote zones expected to flood during storms with one percent or less
                                              13
annual chance of occurrence, see Tr. 2353:6 to 2354:14 (Test. of Michael Nakagaki), and all the
test properties fell within one of these zones at the time they were acquired by plaintiffs, see
Def.’s Br. at 26-27 (identifying the map flood zone for each test property both at the time it was
acquired and under the current flood zone maps). Additionally, topographic maps, known as
“quadrangle maps,” produced by the United States Geological Survey have identified areas
upstream of the dams as subject to controlled inundation since the early 1970s. Tr. at 2283:10 to
2285:18 (Test. of Leslie Hansmann). These maps have been publicly accessible online since
2010 and were readily available for purchase or at universities and public libraries prior to that
time. Tr. 2285:21 to 2286:7 (Hansmann).

           b. Discussions with developers.

        Throughout the 1980s and 1990s, representatives of the Corps interacted with various
developers of the properties surrounding the reservoirs. See, e.g., PX2284 at 1-3 (FB0000632-
34) (documenting exchanges between the developer of subdivisions upstream of the reservoirs,
noting the possibility of upstream flooding). The Corps compiled a fact sheet, see 

(FB0000633), to share with developers of the adjacent property to ensure that they “were fully
aware of the capabilities of the project and size of the pools that could occur behind [the dams],”
Tr. 1507:5-10 (Long). The fact sheet indicated that the design flood pool boundary exceeded the
government-owned land. See PX2284 at 2 (FB0000633).

           c. Harris County flood risk studies.

        Flooding in the early 1990s resulted in increased public awareness of the flooding
potential, see Tr. at 2397:5 to 2400:18 (Test. of Steven D. Fitzgerald), and prompted several
studies by the Harris County Flood Control District, see generally JX54; JX60. The studies
noted growing concern among “residents, business owners[,] and government representatives . . .
regarding the level of protection [sic] that the reservoirs provide to the property upstream of the
dams,” JX60 at 1 (USACE795732), and concluded that the “primary flood threat” in the area
was “the inability to drain the Addicks and Barker [R]eservoirs in an efficient manner,” JX54 at
2 (USACE686046). One report emphasized that “[t]he maximum flood pool levels of the
Addicks and Barker [R]eservoirs extend far beyond the limits of the government[-]owned land,”
noting that “more than 8,000 acres [are] within the reservoir ‘fringe’ areas between the limits of
the government[-]owned land and the . . . maximum flood pools.” 

also
engaged in public outreach, seeking to educate the public about flood risk through public
meetings, where information about the potential for flooding from the reservoirs was included.
See Tr. 595:16-23 (Lindner).

           d. Subdivision plat warnings.

        The risk of upstream flooding that could be produced by the reservoirs filling to their
maximum pool level was well known to Fort Bend County officials. See Tr. 719:15 to 720:10
(Test. of Mark Vogler). During a meeting with the Corps in the early 1990s, a county engineer
discussed “[t]he issue of intermittent inundation or flooding within the Corps’ Barker Reservoir.”
JX45 at 1 (FB0006378). The Corps informed the engineer “that the Barker Dam was designed
and/or modified to contain 8.7 more feet of water than was purchased by the Corps,” which
could “translate[] into the flooding of approximately 4,769 acres of land, not under jurisdiction
of the Corps of Engineers.” 

a Fort Bend Engineering Department report included a
                                                 14
“notice that this subdivision is subject to controlled inundation from Barker Reservoir.” DX122
at 2 (FB0000611).

       Around this time, Fort Bend County began requiring the addition of warning language on
subdivision plats to inform purchasers about the possibility, already known to developers, of
upstream flooding due to Barker Dam operations. See Tr. 736:9 to 737:15 (Vogler). For
example, the plat for one of the trial properties (Giron) stated that “[t]his subdivision is located
adjacent to the Barker Reservoir and . . . [is] subject to extended controlled inundation under the
management of the U.S. Army Corps of Engineers.” DX557 at 1 (FB0025541).

           e. Public outreach discussing upstream flood risk.

          The Corps began its own public-outreach efforts regarding the dams in the mid-1980s,
directing its focus toward interested communities, businesses, and governmental entities. See Tr.
1498:20 to 1499:4 (Long). Thereafter, the Corps conducted “dozens and dozens” of public
presentations in the greater Houston area, during which it discussed the project’s purpose,
history, operations, and operational limitations. See Tr. 1498:20 to 1500:10 (Long). These
public presentations also “included information on the storage capacity of the projects, and that,
in severe storm events where [the reservoirs] stored large pools, that those pools could exceed
the limits of government-owned land,” Tr. 1501:12-19 (Long), noting that “[w]ater stored behind
. . . the dams [] would result in floodwaters in [] homes,” Tr. 1501:25 to 1502:2 (Long). After a
series of storms caused flooding in the early 1990s, the Corps released a report to publicly
emphasize the “order of magnitude of the anticipated flooding damages which could occur off of
Government property assuming different flood events.” JX44 at 1 (USACE015073). Various
documents published by the Corps during the 1990s identified the possibility of future flooding
and the “potential threat of property damage upstream of the reservoir lands.” E.g., JX52 at 7
(USACE015136).

       The Corps continued to discuss and inform the public about the possibility of upstream
flooding in the decade and a half leading up to Harvey. In the mid-2000s, it created an
emergency coordination team to organize better with local agencies in the event of an emergency
concerning reservoir operations. See Tr. 2406:3 to 2407:9 (Fitzgerald). Members of that team
discussed the possibility of upstream flooding on multiple occasions. See, e.g., Tr. 598:14-21
(Lindner) (discussing drills that were conducted in the event of flooding upstream); DX206 at 2
(USACE467209) (stating that “it is only a matter of time before the reservoirs flood off
government-owned land”). In 2009, the Corps prepared presentation slides for public meetings
which showed upstream flooding during large storms. See PX1597 at 18-29 (USACE755528-
39). Again in 2010, the Corps held a series of public meetings where it showed slides illustrating
the possibility of upstream flooding. See JX94 at 74-77 (USACE594433-36). Similar public
meetings, where like information was displayed, were sponsored by Corps in the following years
leading up to Harvey. See, e.g., Tr. 1558:6-14 (Long); DX238.

       3. Operating conditions and meteorological setting for Harvey.

           a. Standard Operating Procedures & the Water Control Manual.

       The general operations of the Addicks and Barker Reservoirs are governed by a Water
Control Manual issued in 2012 (the “Manual”), see generally JX110, with the Corps serving as
                                                 15
the regulating agency, see 

(USACE016306). The Manual was prepared pursuant to a
Corps regulation entitled “Water Control Management” dated 1982, see Tr. 63:24 to 64:3
(Thomas), and it explains the guiding procedures for how the Addicks and Barker Reservoirs
should be controlled in varying situations. The 2012 Water Control Manual was in effect when
Harvey made landfall. See Tr. 58:24 to 59:5 (Thomas).

        The Addicks and Barker Reservoirs are normally dry, as they do not impound water
except to alleviate flood risk. JX110 at 6-3 (USACE016334). The reservoirs and dams are part
of a flood risk project and do not serve any other main purpose such as navigation or
hydroelectric power. See Tr. 63:7-13 (Thomas). “In keeping with the primary objective of flood
control for Addicks and Barker Reservoirs, the general plan for reservoir regulation will be to
operate the reservoirs in a manner that will utilize to the maximum extent possible, the available
storage to prevent the occurrence of damaging stages on Buffalo Bayou.” JX110 at 7-4
(USACE016338). This plan for storage includes all of the land in the reservoirs behind the
embankments, including land the government does not own. See Tr. 67:12 to 68:3 (Thomas).

        Under normal conditions, the reservoirs “operate with two gates set at one-foot openings
to pass normal low flows . . . limit[ing] the discharge on each reservoir to approximately 100-250
[cubic feet per second].” JX110 at 7-4 (USACE016338). “The gates on both reservoirs will be
closed when 1 inch of rainfall occurs over the watershed below the reservoirs in 24 hours or less,
or when flooding is predicted downstream.” 

are kept closed until the gauging
station on Piney Point Road, about eight-to-ten miles downstream of the outlets, is reading less
than 2,000 cubic feet per second. See Tr. 989:13 to 990:8 (Thomas). At times, the Manual
recommends instances of induced surcharges. See JX110 at 7-5 (USACE016339). An induced
surcharge is “a release made to optimize the available [reservoir] storage and protect the integrity
of the dams.” Tr. 103:6-9 (Thomas). That is, when the reservoir pools rise to a certain elevation,
releases from the reservoir will be made gradually according to the induced surcharge schedule.
JX110 at 7-5 (USACE016339).

           b. The Addicks & Barker Dams and Reservoirs immediately prior to Harvey.

        During Tropical Storm Harvey, the Corps operated the dams consistent with the
instructions of the 2012 Water Control Manual. Joint Stip. ¶ 109. When Harvey hit the Addicks
and Barker watershed, the reservoirs for each were empty. See Tr. 160:21-25 (Thomas). At the
time, the Addicks Reservoir had a storage capacity for the government-owned land of 127,591
acre-feet of water, Joint Stip. ¶ 105, and the Barker Reservoir had a storage capacity for the
government-owned land of 82,921 acre-feet of water, Joint Stip. ¶ 106. The Addicks Reservoir
had a maximum capacity of 199,643 acre-feet of water, see JX118 at E-2 (USACE019883), and
the Barker Reservoir had a maximum capacity of 209,600 acre-feet of water, see JX118 at E-4
(USACE019885), if the reservoirs were to fill to their highest elevations where the reservoirs
meet the natural ground. The Addicks watershed is approximately 50% developed and the
Barker watershed is about 60% developed, most of which is residential or related commercial
and office use. See JX110 at 4-8 (USACE016323). The undeveloped areas of the watershed are
primarily used for pasture land and general mixed agricultural purposes. 

Major storms prior to Harvey.

        Several storms of substantial scope occurred in the region during the decades preceding
Harvey. After listing a series of major storms, an operational assessment report issued by the
Corps in 2009 observed that “had some of these events been centered over Addicks and Barker
Reservoirs or the Upper Buffalo Bayou Watershed, the combined rainfall and runoff could have
resulted in flood pools exceeding the limits of government[-]owned land and possibly exceeding
the capacity of Addicks and Barker Dams.” PX59 at 5 (USACE464074).

        In 1979, Tropical Storm Claudette dropped 43 inches of rain in 24 hours on Alvin,
Texas—50 miles southeast of the reservoirs. 

(USACE464073). The rainfall produced by
Claudette was the highest recorded in the United States during a twenty-four hour period, 

Corps to conclude in 1984 that “[t]he [Projected Maximum Flood] on an empty pool
is considered a probable occurrence when compared with the 1979 Claudette rainfall event,”
JX31 at 2 (USACE487626). The Corps likewise acknowledged in 2009 that “[i]f this event had
occurred over the Addicks and Barker watersheds, their reservoir capacities may have been
exceeded.” PX59 at 4 (USACE464073). The Corps calculated that if Addicks or Barker were to
receive the amount of rain dropped by Claudette, it would take between approximately 53 and 55
days to remove enough water to get it back on government-owned land. PX1597 at 40-42
(USACE755551-53).

       Tropical Storm Allison struck about 50 miles northeast of the Addicks and Barker
watershed in 2001, dropping almost 36 inches of rain in five days, and “could have potentially
exceeded reservoir capacity had the storm event occurred directly over the reservoirs.” PX59 at
5 (USACE464074). In light of Claudette and Allison, the Corps recognized in 2009 that,
although the reservoirs had never previously flooded off government-owned land, “we know it
can and probably will happen at some point in time.” PX1597 at 46 (USACE755557).

       Water in the reservoirs exceeded government-owned land for the first time when the
April 2016 “Tax Day Storm” produced record flood pools. Tr. 166:6-10 (Thomas); see also
DX295 at 10 (DEPO_0053700) (“At its peak Barker Reservoir occupied 102.5% of its
government[-]owned land and 40.5% of its total storage capacity.”). The Tax Day Storm
generated ten-to-sixteen inches of rainfall over a twelve-hour period, see JX134 at VII-4
(USACE869254), and although the flood pools exceeded government-owned land, the reservoir
water did not inundate any structures, see Tr. 978:17-22 (Thomas).12 Nonetheless, in the Corps’



       12
         Even so, at least one of the test properties suffered flooding during the Tax Day Storm,
see Tr. 1764:11-18 (Test. of Elizabeth Burnham), and some streets flooded during the storm, see
JX134 at VII 3-4 (USACE869253-4). This result may have been attributable to local stream
flooding or other local circumstance rather than conditions in the reservoirs themselves.

        Additional uncontrolled flows occurred on the road surfaces, which act as part of the
drainage system in Houston, being specifically built at lower elevations than buildings such that
gravity naturally pulls the water down to the roads where they can push flows downstream. See
Tr. 25:20 to 26:2 (Site Visit, May 8, 2019) (Test. of Captain Charles Ciliske).

                                                17
assessment, for Buffalo Bayou at the time it “may have been the worst storm of record.” JX134
at VII-3 (USACE869253). A year later, Harvey exceeded that record.

                   C. Operation of the Addicks & Barker Dams During Harvey

       1. Tropical Storm Harvey.

        Harvey made landfall along the coast, near Rockport, Texas, around 10:00 p.m. on
August 25, 2017, as a Category 4 hurricane (130 mile per hour winds), Joint Stip. ¶ 107; DX737
at 12 (FEMA078357), but weakened into a Tropical Storm within twelve hours of making
landfall, Joint Stip. ¶ 108. Harvey, however, stalled over the Houston metropolitan area for four
more days, maintaining its intensity, dumping record amounts of water on the area. See Joint
Stip. ¶¶ 108, 113, 115. Over the five days, Harvey dropped an average of more than 43 inches of
rain in a 2,000 square mile area, DX737 at 12 (FEMA078357), becoming the largest storm in the
recorded history of the United States, see Tr. 2030:14-17 (Bedient).13 Within Harris County,
Harvey poured an average of 33.7 inches of rain over a four-day period. DX682 at 5.

       2. Corps’ response to Harvey.

       During Harvey, the Corps operated the Addicks and Barker Dams according to the 2012
Water Control Manual. See Tr. 982:1-3 (Thomas). The gates on all five conduits were closed
for Addicks and Barker at the beginning of the storm, as called for by the Manual. See Tr.
1446:10-15 (Long). On August 25, 2017, the Corps declared a general emergency, which
included a dam safety emergency. Tr. 118:25 to 119:1 (Thomas). As a result, the Corps
followed the Manual’s instructions for the initial emergency levels. Tr. 119:15-17 (Thomas).

        With the forecasted impending rain, the Corps knew that flooding “beyond the
government[-]owned land limits” in Addicks and Barker was imminent. JX146 at 2
(DOJ0008154) (Corps’ internal letter dated August 25, 2017 at 2:26 p.m.). On August 28,
shortly after midnight, the Corps initiated releases of impounded water in both reservoirs,
pursuant to the induced surcharge regulation in the Manual. DX649 at 1. At that point, the flood
pools in the reservoirs had reached at least 101 feet in Addicks and 94.9 feet in Barker. See Tr.
983:11-16 (Thomas). Because these releases were limited, however, by the guidelines in the
Manual to protect downstream Houston from additional flooding, the Corps was unable to
release water fast enough to decrease the pool size given the high rate of incoming water. See
Tr. 991:8-19 (Thomas). During Harvey, uncontrolled releases also flowed around the ends of the
auxiliary spillway at the northeast end of the Addicks Reservoir onto private land. See Tr. 24:8
to 27:12 (Site Visit) (Ciliske); see also PX25 at 1 (USACE016691) (“Uncontrolled release[s]
around the Addicks north[-]end emergency spillway w[ere] observed on August 29, 2017 when
the reservoir reached [] 108 ft.”).14


       13
         Tropical Storm Claudette in 1979 was roughly comparable in total rainfall, and was
more intense but more localized. 

.
       14
          Testimony about the flows at the north-end auxiliary spillways of Addicks was
ostensibly inconsistent due to the imprecise vocabulary used when describing spillway functions,
i.e., whether water flowed over or around at both the northwest and northeast auxiliary spillways
                                                18
        The flood pools in the reservoirs crested at a record pool elevation of 101.6 feet in Barker
and 109.1 feet in Addicks on August 30, 2017. Joint Stip. ¶¶ 110, 111. When the threat
subsided and flood pools dropped to safer levels, the Corps developed a drawdown plan to fully
drain the reservoirs while still maintaining their integrity. See Tr. 992:4 to 993:4 (Thomas). The
plan went into effect on September 3, 2017, and the Corps resumed normal operations on
September 16, 2017. DX649 at 6-7. The reservoir water elevations far exceeded the extent of
government-owned land in both Addicks and Barker. See DX683 at 1. Of the approximately
one million homes in the Harris County, around 154,000 of them flooded from the impounded
water. See Tr. 2451:19-24 (Fitzgerald). The Addicks and Barker flood control project, though,
did prevent an estimated $7 billion in projected losses downstream in Houston. See Tr. 164:24 to
165:8 (Thomas).

                   D. Background of the Thirteen Upstream Test Properties

        The thirteen upstream test properties provide a sampling of the conditions at various
locations inundated by the Addicks and Barker Reservoirs during and after Tropical Storm
Harvey.

       1. Banker residence.

        The land owned by plaintiffs Christina and Todd Banker is a residential property situated
within the Barker reservoir in Katy, Texas at 4614 Kelliwood Manor Lane. Joint Stip. ¶ 4. The
finished first floor of the home is set at a 100.7-foot elevation. Joint Stip. ¶ 9. Flooding within
the home attendant to Harvey reached approximately 1.1 feet. See PX526 at 46. The Bankers
evacuated their home on the morning of August 28 and returned on September 4. Tr. 1709:23 to
1712:20 (Test. of Todd Banker). During that period, flood water was present in the home for
approximately four days. DX608 at 164. In addition to structural damage to the home, much of
the Banker’s personal property was destroyed by the flooding, see, e.g., Tr. 1717:24 to 1718:21
(Banker), and the home was uninhabitable for about seven months while it underwent
remediation, Tr. 1717:12-18 (Banker).

       2. Burnham residence.

        The land owned by plaintiff Elizabeth Burnham is a residential property situated within
the Addicks Reservoir in Houston, Texas at 15626 Four Season Drive. Joint Stip. ¶ 10. The
finished first floor of the home is set at a 105.4-to-105.5-foot elevation. Joint Stip. ¶ 18.
Flooding within the home attendant to Harvey reached approximately four to five feet. See
PX526 at 46; Tr. 1773:21-24 (Test. of Elizabeth Burnham). Flood water was present in the
home for at least seven days. See DX608 at 164; Tr. 1771:20-22, 1773:25 to 1774:5 (Burnham).
Ms. Burnham’s personal property as well as the home suffered substantial damage. See, e.g., Tr.



of Addicks Reservoir. At the northwest auxiliary spillway, there was no flow around or over the
ends, see Tr. 21:18-22 (Site Visit) (Ciliske), however, at the northeast auxiliary spillway, water
flowed around the spillway, see Tr. 24:15-18 (Site Visit) (Ciliske), but not over it, see Tr. 24:8-
14 (Site Visit) (Ciliske). The spillways are intentionally designed to have water flow over them,
but not around. See Tr. 35:19 to 36:4 (Site Visit) (Ciliske).
        .
                                                  19
1775:24 to 1776:21 (Burnham). The home was uninhabitable for a period of months, at which
time Ms. Burnham sold the property “as is.” See Tr. 1780:15 to 1781:18 (Burnham).

       3. Giron residence.

        The land owned by plaintiff Juan Giron is a residential property situated within the
Barker Reservoir in Katy, Texas at 4310 Cassidy Park Lane. Joint Stip. ¶ 19. The finished first
floor of the home is set at a 101.5-foot elevation. Joint Stip. ¶ 23. Flooding within the home
attendant to Harvey reached approximately one foot at the high-water mark. See Tr. 1675:7-11
(Test. of Juan Giron). Flood water was present in the home for approximately five days. See
DX608 at 164. The Giron property suffered substantial damage to the home and much of what
was inside was unsalvageable. See, e.g., Tr. 1678:19-22, 1680:3-5 (Giron). At the time of trial,
Mr. Giron was still living in a trailer parked in his driveway. See Tr. 1646:20-21 (Giron).

       4. Holland residence.

         The land leased by plaintiff Scott Holland is a residential property situated within the
Addicks Reservoir in Houston, Texas at 1923 Wingleaf Drive. Joint Stip. ¶ 24. The finished
first floor of the home is set at an elevation between 107.8 and 107.9 feet. Joint Stip. ¶ 26.
Flooding within the home attendant to Harvey reached approximately 1.5 feet. See PX526 at 46;
Tr. 1845:23 to 1846:1 (Holland). Mr. Holland evacuated his home on August 28, a daunting
process due to sutures in his stomach and chest still healing from a recent kidney surgery. Tr.
1836:7 to 1838:11 (Holland). Flood water was present in the home for about 3.5 days. DX608
at 164. The home suffered severe structural damage and much of Mr. Holland’s personal
property was destroyed by the flooding. See, e.g., Tr. 1842:1 to 1844:4 (Holland). Because the
home was uninhabitable and he was unable to afford repairs, Mr. Holland was forced to move
away from Houston and reside in a small trailer, where he still lived at the time of trial. Tr.
1844:15 to 1845:11 (Holland).

       5. Lakes on Eldridge Homeowners Association.

        Lakes on Eldridge is a homeowners association that acquired its real property from the
developer or builder of the Lakes on Eldridge residential subdivision, a gated community in
Harris County situated within the Addicks Reservoir. Joint Stip. ¶¶ 27, 28. The property at issue
includes a clubhouse and its associated amenities, a swimming pool, tennis court, volleyball
court, and playground. Tr. 1386:3-16 (Test. of Sue Strebel). Flood water rose to about six feet
above the volleyball courts, Tr. 1398:17-22 (Strebel), approximately six inches in the clubhouse,
PX526 at 46, and was present on the property for at least four days, see Tr. 1401:22 to 1402:11
(Strebel). The clubhouse and various amenities required extensive repairs, see, e.g., Tr. 1390:19
to 1391:21 (Strebel), requiring the clubhouse facility and its amenities to close for eight months,
Tr. 1390:5-12 (Strebel).

       6. Micu residence.

        The land owned by plaintiff Christina Micu is a residential property situated within the
Barker Reservoir in Katy, Texas at 6411 Canyon Park Drive. Joint Stip. ¶ 31. The finished first
floor of the home is set at a 99.8-foot elevation. Joint Stip. ¶ 36. Flooding within the home
attendant to Harvey reached approximately two feet. See PX526 at 46. Ms. Micu and most of
                                                20
her family evacuated the home prior to Harvey. Tr. 1296:11-25 (Test. of Christina Micu). Her
husband gained access to the home via kayak on September 2 and she returned on September 5,
Tr. 1298:22 to 1299:1 (Micu), finding mold growth and extensive destruction of personal
property, see, e.g., Tr. 1299:25 to 1300:10 (Micu). Flood water was present in the home for
about ten days. Tr. 1300:24 to 1301:1 (Micu). The Micu family was forced to reside with a
friend and then rent an apartment before moving back into their home a year after Harvey. Tr.
1306:21 to 1305:7 (Micu).

       7. Popovici residence.

        The land owned by plaintiff Catherine Popovici is a residential property situated within
the Barker Reservoir in Katy, Texas at 19927 Parsons Green Court. Joint Stip. ¶ 37. The
finished first floor of the home is set at a 102.2-foot elevation. Joint Stip. ¶ 42. No water entered
inside the home, but it rose to the foundation and was within a couple inches of entering, Tr.
1239:2-5 (Test. of Catherine Popovici), and remained on the property between four and six days,
Tr. 1242:2-12 (Popovici). The flooding around the home prevented ingress or egress and
damaged wooden beams in the structure of the home. Tr. 1243:1-9 (Popovici).

       8. Sidhu residence.

        Plaintiff Kulwant Sidhu is the joint owner of 29 condominium units used as residential
rental properties and situated within the Addicks Reservoir in Houston, Texas at 16111
Aspenglenn Drive. Joint Stip. ¶ 43. The property at issue in Mr. Sidhu’s claim at trial consists
of two of his 29 units: Unit 603 (a first-floor, downstairs unit) and Unit 604 (a second-floor,
upstairs unit directly above Unit 603). Joint Stip. ¶¶ 44, 46. The finished first floor of the
condominium building in which the two units are located is set at an elevation of 107.0 to 107.1
feet. Joint Stip. ¶ 49. No flood water reached the upstairs unit, Joint Stip. ¶ 48, and it was not
damaged, Tr. 1748:5-7 (Test. of Kulwant Sidhu). Flooding within the downstairs unit attendant
to Harvey reached approximately 2.4 feet, PX526 at 46, and remained for about 4.5 days, DX608
at 164. The flood damage required gutting and renovating Unit 603—a process that took nearly
a year, during which time the unit could not be rented. Tr. 1741:23 to 1742:25 (Sidhu).

       9. Soares residence.

        The land owned by plaintiff Elisio Soares is a residential property situated within the
Barker Reservoir in Katy, Texas at 20526 Indian Grove Lane. Joint Stip. ¶ 50. The finished first
floor of the home is set at a 101.1-foot elevation. Joint Stip. ¶ 55. The Soares family was on
vacation when Harvey happened, Tr. 1080:12-16 (Test. of Elisio Soares), and they could only
access their home by kayak when they returned on August 31, Tr. 1080:17 to 1081:10 (Soares).
Flooding within the home attendant to Harvey reached approximately 8.4 inches, PX526 at 46,
and was present in the home for approximately four days, Tr. 1086:22-24 (Soares). In addition
to structural damage to the home, a significant amount of the Soares family’s personal property
was destroyed. See, e.g., Tr. 1091:12 to 1092:4 (Soares). The family was forced to live with
friends for two weeks and then in the upstairs of the home, unable to cook meals, until January of
the following year, and the home could not be repaired until May 2018. Tr. 1092:18 to 1093:23
(Soares).



                                                 21
       10. Stewart residence.

         The land owned by plaintiff Mitchell Stewart is a residential property situated within the
Addicks Reservoir in Houston, Texas at 4719 Eagle Trail Road. Joint Stip. ¶ 56. The finished
first floor of the home is set at a 108.9- to 109.0-foot elevation. 

Flooding within the
home attendant to Harvey reached at least six inches, see PX526 at 46; Tr. 1600:13-15 (Test. of
Mitchell Stewart), and the water was present in the home for approximately four-to-five days,
see Tr. 1607:13-18 (Stewart). The Stewart family suffered damage to much of their personal
property and their home required significant renovations including removing and replacing the
first four feet of the home’s sheetrock. Tr. 1603:25 to 1604:4 (Stewart). The Stewarts were
displaced from their home for about five months. Tr. 1609:17-20 (Stewart).

       11. Turney residence.

         The land owned by plaintiff Robert Turney is a residential property situated within the
Addicks Reservoir in Houston, Texas at 15910 Red Willow Drive. Joint Stip. ¶ 62. The finished
first floor of the home is set at a 104.5- to 104.7-foot elevation. Joint Stip. ¶ 69. Flooding within
the home attendant to Harvey was about five feet, see PX526 at 46, with the high-water mark
reaching about six feet, see Tr. 2134:18-23 (Test. of Robert Turney). Flood water was present in
the home for over a week. See DX608 at 164. The Turney property suffered great damage that
required the entire interior to be gutted and rebuilt. See Tr. 2136:8-11 (Turney).

       12. West Houston Airport.

         The West Houston Airport Corporation is the owner of a commercial property situated
within the Addicks Reservoir in Houston, Texas at 18000 Groschke Road. Joint Stip. ¶ 76. The
finished first floor of the terminal building at the West Houston Airport is set at a 108.6-foot
elevation. Joint Stip. ¶ 78. Flooding within the terminal reached a maximum of about 9.6
inches. See PX526 at 46. The flood water reached the terminal on August 29 and receded from
the terminal by September 1. See Tr. 1885:17 to 1886:13 (Test. of Stacey Lesikar-Martin).
Flood water remained on the runways until September 5, with the water receding from the
property line on approximately September 7. See Tr. 1886:17-24 (Lesikar-Martin). The terminal
itself suffered substantial damage, and damage occurred to numerous aircraft, service apparatus,
and vehicles present at the airport at the time of the flooding. See Tr. 2121:6-25, 2124:8-14
(Test. of Woody Lesikar). The airport was not fully operable for seven-to-ten days’ time, and
repairs on the property took about a year. See Tr. 2121:6 to 2123:8 (Lesikar).

       13. Wind residence.

         The land owned by plaintiffs Kurt and Jean Wind is a residential property situated within
the Addicks Reservoir in Houston, Texas at 5306 Sunbright Court. Joint Stip. ¶ 70. The finished
first floor of the home is set at a 109.2-to-109.3-foot elevation. Joint Stip. ¶ 75. Flooding within
the home attendant to Harvey reached approximately two inches. See PX526 at 46; DX608 at
164; Tr. 1633:12-14 (Test. of Kurt Wind). Flood water was present in the home for about two-
to-three days. See Tr. 1635:10-11 (Wind). In addition to significant structural damage to the
home, much of the Winds’ personal property was destroyed by the flooding. See Tr. 1635:11 to
1636:24 (Wind). Due to the damage and necessary repairs, the home was uninhabitable for 11
months. See Tr. 1637:5-7 (Wind).
                                                 22
                                E. Expert Reports about the Storm

       1.   Dr. Bedient.

        An expert in hydrology, hydraulics, and floodplain analysis, Dr. Philip Bedient testified
on behalf of plaintiffs regarding the cause of flooding at each of the test properties during
Harvey. See generally PX526; Tr. 1934:11-13 (Bedient).15 Dr. Bedient reached his conclusions
by reference to a number of data sources, including slab elevations of the test properties, ground
elevation data, water level gauges for creeks and the reservoir pools, aerial images from the
National Oceanic and Atmospheric Administration, and eyewitness photographs and videos. See
Tr. 1910:1 to 1911:17 (Bedient). Dr. Bedient’s methodology compared the reservoir pool
elevation data obtained from United States Geological Survey gauges to the slab elevation
survey data of each test property, and then confirmed those findings by reference to aerial images
taken on August 30 and LiDAR data (technology for determining ground elevation at numerous
points). See 

that the flooding at each test property was caused by the water
impounded behind the Addicks and Barker Dams. Tr. 1948:3-12 (Bedient); PX526 at 46.

        Dr. Bedient also addressed whether any of the test properties would have flooded without
the dams impounding the rainfall waters or whether any of the properties flooded during Harvey
independently of the reservoir pool levels entering onto the properties. He identified, and
rejected, two other possible causes of the flooding: local drainage systems and riverine flooding.
See generally PX526 at 47-54. His examination of the relevant local drainage systems’
capacities led him to the conclusion that they were capable of adequately handling rainfall much
more intense than Harvey. See PX526 at 49. Likewise, Dr. Bedient concluded that “the riverine
flooding that occurred during Harvey did not cause and would not have caused any of the [t]est
[p]roperties to flood.” PX526 at 54. He reached this conclusion by estimating the flood level
along each of the creeks in the vicinity of the test properties, and then comparing that elevation
to the slab elevation of each test property, except the airport. See Tr. 1984:11-25 (Bedient). He
noted that the airport was a unique situation because of the size of the property but nonetheless
he concluded that because of a high bank attendant to a creek in that location, creek water did not
cause the flooding. See Tr. 1985:18 to 1989:14 (Bedient). He confirmed this observation by
reference to eyewitness testimony and by comparison to the Tax Day flood, which likewise did
not produce airport flooding. 

Dr. Bedient concluded that “[n]one of the test
properties’ structures would have flooded but for the impoundment of rainfall runoff waters
behind Addicks and Barker Dams.” PX526 at 7-8.

       2. Dr. Nairn.

      Dr. Robert Nairn, an expert in coastal and riverine engineering, with a specialty in
numerical modeling across actual and hypothetical conditions, testified on behalf of the
government using a numerical model of the storm and relevant waterflows. See generally



       15
         Dr. Bedient is a professor of engineering at Rice University and teaches and performs
research in hydrology. Tr. 1901:7-9 (Bedient).

                                                23
DX608; see also Tr. 2628:22 to 2629:3 (Test. of Robert Nairn).16 Dr. Nairn looked at the water
surface elevations at each of the test properties, specifically examining the rise and drop over
time. Tr. 2636:14-21 (Nairn). Using various input data sets, including, e.g., topography maps
and a land cover data set, see Tr. 2647:15-20 (Nairn), and taking into account various water
runoff impacts and parameters, see, e.g., Tr. 2648:11-23 (Nairn), Dr. Nairn built a modeling
system that applies mathematical principles of hydrodynamic equations to simulate rainfall and
water movement, see, e.g., Tr. 2655:10-16 (Nairn). Subsequently, after calibrating the model
and relevant coefficients, see Tr. 2656:19 to 2657:20 (Nairn), Dr. Nairn ran the model for each of
the properties and analyzed the outcomes of four different scenarios for Addicks and Barker: (1)
the actual Harvey scenario; (2) the no-dam (“no project”) scenario; (3) the gates-closed scenario;
and (4) the gates-open scenario, see Tr. 2636:14-24 (Nairn); see also DX608 at 92-93. Dr.
Nairn’s main conclusion was that three out of the thirteen test properties (Burnham, Giron, and
Micu) would have flooded even had the Addicks and Barker Dams not been built, due to the
sheer amount of rainfall and riverine flooding, thus positing that the flooding of at least those
three properties was not the result of the government’s actions relating to Addicks and Barker.
See DX608 at iii, 129.

       3. Other testifying experts.

        Several expert witnesses testified at trial regarding the severity of the impact that flood
damage had on the test properties. Dr. Glen Randall Bell, an expert in real estate damage and
economics and real estate valuation (including severity), testified on behalf of plaintiffs about
severity of the impact of the flooding on the valuation of plaintiffs’ properties. See generally
PX660; see also Tr. 1360:5 to 1361:15 (Bell). His testimony and report focused on the severity
of impact rather than on the specific quantification of any lost value. PX660 at 4; Tr. 1362:6-15
(Bell).

        Additionally, Matthew Deal, an expert in real estate market studies and real estate
valuation, conducted a market study appraisal analyzing supply, demand, and prices for specific
property types in the area. See generally PX2205; Tr. 2188:4-15 (Deal). His research concluded
“that properties that were inundated by flood waters suffered significant and immediate
impairment that resulted in precipitous price reductions after flood waters had receded.” PX2205
(report cover letter).

        David Hooper, a microbiologist who is an expert in preparing scopes of work for
property damaged by water, was retained by the government to create models of the scopes of
work that would be required on five of the test properties after the flooding. See generally
DX602-607. His report considered “the overall quality of water; the amount of waters present
within a residence; the duration of flooding; the ability for water to move via capillary action
within a wall; as well as the presence of microbiological growth; and, lastly, application of
industry standards in construction guidelines.” Tr. 2876:2-7 (Hooper). He developed estimates
of the scope of work that would be required under various scenarios, such as if the reservoirs had
not been in place at all, see, e.g., DX602 at 17-18 (considering the Burnham property),



       16
          Dr. Nairn is an engineer who works at Baird & Associates, Oakville, Ontario, Canada,
on river and coastal engineering projects worldwide. Tr. 2625:15 to 2626:1 (Nairn).
                                               24
concluding that the scopes of work would be very similar in the actual Harvey scenario and in
the alternative scenarios, see, e.g., 2922:2-4 (Hooper).

        Andrew Ickert, an expert in hydrology, testified on behalf of the government about the
character of the land in the Addicks and Barker watershed and greater Houston area. See
generally DX600; see also Tr. 3083:21 to 3084:7 (Ickert). Mr. Ickert explained how
development could impact the watershed overall, giving particular regard to increasing runoff
flow rates and size, and the impact of this circumstance on the overall flood pool size. See Tr.
3084:11 to 3085:14 (Ickert).

       William Kappel, an expert in meteorology, was called to testify by the government,
regarding the magnitude of Harvey and the meteorological setting. See generally DX601 at 38-
222; see also Tr. 1156:6-22 (Kappel). Specifically, Mr. Kappel testified about “the methodology
and results of a detailed analysis of the storm precipitation in and around the region of []
Houston[,] Texas during the month of August 2017[, when Harvey hit].” DX601 at 44.

                                STANDARDS FOR DECISION

        Under the Takings Clause of the Fifth Amendment, “private property [shall not] be taken
for public use, without just compensation.” U.S. Const. amend. V. The Takings Clause applies
the fundamental notion that the government cannot “forc[e] some people alone to bear public
burdens which, in all fairness and justice, should be borne by the public as a whole.” Armstrong
v. United States, 

, 49 (1960). Thus, “[w]hen the government physically takes
possession of an interest in property for some public purpose, it has a categorical duty to
compensate the former owner.” Tahoe-Sierra Pres. Council, Inc. v. Tahoe Reg’l Planning
Agency, 

, 322 (2002) (citing United States v. Pewee Coal Co., 

, 115
(1951)).

        In a takings case, the plaintiff must establish two elements to have a viable claim. First,
plaintiff must establish that he or she holds “a property interest for purposes of the Fifth
Amendment.” Caquelin v. United States, 

, 572 (2018), appeal docketed, No.
2019-1385 (Fed. Cir. Jan. 9, 2019) (citing Members of the Peanut Quota Holders Ass’n v. United
States, 

, 1330 (Fed. Cir. 2005)) (additional citations omitted). After identifying a
valid property interest, “the court must determine whether the governmental action at issue
amounted to a compensable taking of that property interest.” American Pelagic Fishing Co.,
L.P. v. United States, 

, 1372 (Fed. Cir. 2004) (citations omitted).

        “[N]o magic formula enables a court to judge, in every case, whether a given government
interference with property is a taking.” Arkansas Game & Fish Comm’n v. United States, 

, 31 (2012). And while there are some bright-line rules, see 

into whether
a taking has occurred ultimately is a question of law based on factual underpinnings, 

, requiring the court to engage in “ad hoc, factual inquiries,” Kaiser Aetna v.
United States, 

, 175 (1979); see also Arkansas Game & 

(“[M]ost takings claims turn on situation-specific factual inquiries.”).

       A government taking can occur in many forms, ranging from the classic example of a
permanent physical occupation of property, see Loretto v. Teleprompter Manhattan CATV Corp.,

, 426 (1982), to regulation which permanently deprives a property owner of all
                                                25
economically viable uses of his or her land, see Lucas v. South Carolina Coastal Council, 

, 1019 (1992), or one that bars most such uses, see Penn Cent. Transp. Co. v. City of
New York, 

, 124 (1978). More specifically, takings can be broken down by their
means, duration, and impact. That is, takings can be either (1) physically or by regulation; (2)
permanent or temporary; and (3) categorical or non-categorical. See 

.

        At the outset, to apply the proper analysis, the court must determine the type of taking
alleged. See American 

. Physical takings, as opposed to those by
regulation, involve physical occupation, and can occur when the “owner [is] deprived of valuable
property rights, even [if] title ha[s] not formally passed.” Caldwell v. United States, 

, 1235 (Fed. Cir. 2004). The distinction between permanent and temporary is a narrower
one, in that on a sufficiently long timeline, every government action could be considered
temporary. See 

. Generally, the word temporary “refers to those
governmental activities which involve an occupancy that is transient and relatively
inconsequential,” Hendler v. United States, 

, 1377 (Fed. Cir. 1991), while
permanent refers to those governmental activities more substantial in nature, though they need
not be “exclusive, or continuous and uninterrupted,” 

the Supreme Court in
Nollan concluded that a “permanent physical occupation” had occurred when the government
essentially took an easement for public access across plaintiffs’ property that granted a
“permanent and continuous right to pass to and fro, so that the real property may continuously
be traversed.” Nollan v. California Coastal Comm’n, 

, 832 (1987). Lastly, a
categorical taking occurs when the government seizes the entirety of a landowner’s property, see

, whereas a non-categorical taking occurs when the landowner is
not completely deprived of all economic value, see 

(citing United States v. Causby,

(1946)).

        At issue here is whether the government’s actions constitute a physical, permanent, non-
categorical taking for a flowage easement.17 The alleged taking is physical, in the sense that
actual flood waters physically entered the property; permanent, in the sense that the government
retains the rights to this flowage easement on a permanent basis with a continual right of re-
entry; and non-categorical, in the sense that the property owners are not deprived of all
economically viable use of their property as a result of the flowage easement. Further, this case
presents a scenario of inverse condemnation, i.e., where the landowner seeks to “recover[] just
compensation for a taking of his [or her] property when condemnation proceedings have not
been instituted.” United States v. Clarke, 

, 257 (1980).




       17
          Plaintiffs have alleged three separate takings: (1) a temporary, categorical, physical
taking for the temporary flooding; (2) a permanent, categorical, physical taking for the
destruction of plaintiffs’ personal property; and (3) a permanent, non-categorical, physical taking
for the flowage easements on each property. See Pls.’ Br. at 59. The court, however, finds these
distinctions unnecessary, as the first two alleged takings are simply the consequential result of
the third. Thus, the court here will apply its analysis to the consideration of whether the
government’s actions constituted a physical, permanent, non-categorical taking for a flowage
easement.
                                                  26
        Inverse condemnation cases in the flooding context, particularly those resulting from
government-induced flooding either of a permanent or temporary nature, are not new to this
court or others. See, e.g., Arkansas Game & 

-33 (citing Pumpelly v. Green
Bay Co., 80 U.S. (13 Wall.) 166 (1872), United States v. Cress, 

(1917), and United
States v. Dickinson, 

(1947)). In 2003, the Federal Circuit decided Ridge Line,
addressing the scenario of whether government-induced increased water runoff onto private
property constituted a taking of a flowage easement by inverse condemnation. Ridge Line, Inc.
v. United States, 

, 1355 (Fed. Cir. 2003). In Ridge Line, the Federal Circuit
applied a multi-pronged test to determine whether a taking, and specifically not a tort, had
occurred. 

prong evaluates whether “the government intends to invade a protected
property interest” or whether “the asserted invasion is the direct, natural, or probable result of an
authorized activity and not the incidental or consequential injury inflicted by the action.” 

Basin Orchard v. United States, 

, 709 (Ct. Cl. 1955)). In
other words, the government must intend to invade the property or the injury must be the
foreseeable result of the action. See 

n.18 (citing Cary v. United
States, 

, 1377 (Fed. Cir. 2009), and Ridge 

). Second, “the
nature and magnitude of the government action must be considered.” Ridge 

. This factor includes a requirement that “an invasion must appropriate a benefit to the
government at the expense of the property owner, or at least preempt the owner[’]s right to enjoy
his property for an extended period of time, rather than merely inflict an injury that reduces its
value.” 

background, the Supreme Court decided Arkansas Game & Fish in 2012,
expounding a list of six factors to consider when determining the existence vel non of a
compensable taking. 

. The factors, subsuming the considerations of the
Ridge Line test, include: (1) “time;” (2) “inten[t];” (3) “foreseeab[ility];” (4) “character of the
land;” (5) “reasonable investment-backed expectations;” and (6) “severity.” 

Game & Fish concerned a taking which was temporary in nature, the same
considerations remain relevant to the inquiry here, that is, whether the government’s actions with
regard to Addicks and Barker constitute a compensable taking, albeit a permanent one.

                                            ANALYSIS

        The court will begin its assessment by determining if each plaintiff has established a
cognizable property interest. Following this, the court will turn to the more fact-intensive
examination of the Arkansas Game & Fish considerations to evaluate whether plaintiffs have
met their burden of showing that they have suffered a compensable taking. Finding that the
government’s actions in this case constitute a taking, the court will then consider whether
defendant has any potential defenses to liability. The government argues that its actions do not
constitute a taking because the Corps was acting under the police power and under the doctrine
of necessity. The court concludes that these defenses are not applicable; therefore, the
government is liable for the taking of plaintiffs’ properties.




                                                 27
                                    A. The Takings Analysis

       1. Property interests.

        Plaintiffs must have a valid property interest at the time of the taking to be entitled to
compensation. See Wyatt v. United States, 

, 1096 (Fed. Cir. 2001) (citing
primarily Almota Farmers Elevator & Warehouse Co. v. United States, 

, 473-74
(1973)). Plaintiffs are owners of private properties not subject to flowage easements.18
Ownership of the properties by each plaintiff respectively and the lack of a previous flowage
easement are not in dispute. See generally Def.’s Br. at 93-97. The government, however,
disagrees that plaintiffs hold compensable property interests under principles of both state and
federal law. Namely, the government argues that: (1) the government has the right to mitigate
against floodwaters under Texas law; (2) that plaintiffs have no right to be free from invasions
because their ownership post-dates the dams’ construction; and (3) that the federal Flood Control
Act of 1928, ch. 569, § 3, 45 Stat. 534 (codified as amended in relevant part at 33 U.S.C. §
702c), limits plaintiffs’ rights to compensation. See id.; see also In re Upstream Addicks &

.

        The government misstates the interplay of these laws with the Takings Clause. While the
law cited in support of the government’s first contention, Tex. Water Code Ann. § 11.086(c),
exempts the government from liability for diversions of water caused by the “construction and
maintenance of levees and other improvements to control floods,” a conscious diversion of water
by the government onto private properties in a reservoir by a flood-control dam is not within this
exception. Cf. Harris Cty. Flood Control District v. Kerr, 

, 807 (Tex. 2016)
(“This is not a case where the government made a conscious decision to subject particular


       18
           This description, i.e., that plaintiffs are owners of private properties not subject to
flowage easements, is in a nutshell a finding respecting the character of the land at issue. In
other cases, the character of the land may be more complicated or may factor more heavily in the
takings determination. What is most relevant to the takings inquiry here is that defendant had no
legal right to cause flood waters to enter the properties. The character of the land in government
flooding cases is usually defined by whether, inherently, the property is “especially susceptible
to flooding.” See, e.g., 

. In this case, whether the private property
is used as farm land, as a residence, or commercially does not bear on liability. 

n.22.
Defendant contends that the character of the land at issue is land that has always been
“susceptible to flooding during extreme weather events” including “possible inundation
associated with the pools impounded by the [Addicks and Barker] Project.” Def.’s Br. at 100-01.
The government thus appears to be arguing that the character of plaintiffs’ lands is property
located within a reservoir in an area of the country susceptible to storms. That plaintiffs’
properties may be susceptible to flooding during extreme weather events is of some relevance,
but it is independent from the fact that plaintiffs’ properties are privately-owned land within a
reservoir that only flooded in this case because of the government’s construction of the Addicks
and Barker Dams (for a discussion on causation, see infra, at 35-39). Even if this geographical
area is generally susceptible to flooding during extreme weather events, the character of
plaintiffs’ land would not be especially susceptible to flooding without the construction of the
dams. Therefore, the character of the land at issue in this case is most simply described as
private property not subject to a flowage easement.
                                                      28
properties to inundation so that other properties would be spared, as happens when a government
builds a flood-control dam knowing that certain properties will be flooded by the resulting
reservoir. In such cases of course the government must compensate the owners who lose their
land to the reservoir.”); see also In re Upstream Addicks & 

. The
government’s second contention, that plaintiffs’ claims fail because they acquired their land after
the completion of the Addicks and Barker Dams, also does not bar relief, see 

at
669, and is more appropriately addressed as a consideration in regard to plaintiffs’ reasonable
investment-backed expectations. Lastly, defendants argue that “Section 702c of Flood Control
Act of 1928 . . . supports the conclusion that landowners in the vicinity of a federal project
constructed and operated to reduce flood risk lack a right to compensation for damages caused
by floodwaters not fully controlled by the Project.” Def.’s Br. at 97. This argument is
unpersuasive. The Flood Control Act of 1928 does not supersede or bar this court’s jurisdiction
over takings claims for flooding. See In re Upstream Addicks & 

;
accord California v. United States, 

, 1383 (Fed. Cir. 2001) (“Our review of the
Flood Control Act of 1928 leaves us with the firm conviction that Congress did not partially
impliedly repeal the Tucker Act.”) (addressing immunity in Tucker Act contract claims); see also
Scranton v. Wheeler, 

, 153 (1900) (“Congress may not override the provision that
just compensation must be made when private property is taken for public use.”). The court
finds defendant’s arguments unconvincing; therefore, plaintiffs have met their burden of
establishing a valid property interest.

       2. Takings factors.

            a. Nature and magnitude of the government action.

               i. Time & duration of the taking.

        The time and duration of the government invasion is an important consideration in many
takings cases. See, e.g., Arkansas Game & 

(citing Loretto v.
Teleprompter Manhattan CATV 

n.12; 

;
National Bd. of YMCA v. United States, 

, 93 (1969)). Particularly, the time and
duration of the government action may be a highly relevant inquiry when determining whether
the action constitutes a tort or a taking in the context of temporary takings. See 

(collecting cases applying the time factor in a temporary takings analysis). But
when the taking is one of a permanent nature, as it is here, the time and duration of the invasion
is essentially undisputed and manifestly supports the finding of a taking. The government,
through its construction, maintenance, and operation of the Addicks and Barker Dams in the
past, present, and future, has taken a permanent flowage easement on plaintiffs’ properties. 19
Defendant argues that its actions had only temporary consequences, because flood waters from
Harvey were only on the properties for a matter of days. See Def.’s Br. at 108-09. This


       19
          The parties in their briefing for the motion to dismiss disputed whether plaintiffs’
claims related only to government inaction, as opposed to affirmative government actions, and
thus would not state viable takings claims. See Def.’s Motion to Dismiss at 4; Pls.’ Opp’n to
Motion to Dismiss at 17, ECF No. 99. For the reasons set forth in the court’s opinion deferring
ruling on the motion to dismiss, see In re Upstream Addicks & 

,
the court concludes that plaintiffs’ claims are properly based on government action, not inaction.
                                                  29
argument, however, fails to account for the fact that the government’s actions have subjected
plaintiffs’ private properties to the possibility, rather probability, of government-induced
flooding ever since the construction of these dams, throughout subsequent changes to the dams
and reservoirs, and for at least the foreseeable future. The time and duration of the government’s
actions at issue here is not measured by “the length of time the water inundates the properties,”
as the government would have it, 

rather, it is measured by a permanent right to
inundate the property with impounded flood waters. Thus, this factor weighs in favor of
plaintiffs.
                ii. Severity.

        Another factor that warrants consideration in the determination of liability under
Arkansas Game & Fish is the “[s]everity of the 

. The severity
factor aids in differentiating a taking from a tort. See Ridge 

(noting that
“[t]he tort-taking inquiry . . . requires consideration of whether . . . the government’s actions
were sufficiently substantial to justify a takings remedy”). In effect, it requires the court to
assess whether the government’s interference with plaintiffs’ property rights “was substantial
and frequent enough to rise to the level of a taking.” 

357 (citation omitted). As the
Supreme Court stated in Portsmouth Harbor Land & Hotel Co. v. United States, where the Court
held that repeated firing of military guns over a beach resort could constitute a taking if frequent
enough, “[w]hile a single act may not be enough, a continuance of them in sufficient number and
for a sufficient time may prove [a taking]. Every successive trespass adds to the force of the
evidence.” 

, 329-30 (1922); see also 

, 265 (holding that
repeated overflights of governmental aircraft above a farm constituted a taking). In the flooding
context, “property may be taken by the invasion of water where subjected to intermittent, but
inevitably recurring, inundation due to authorized [g]overnment action.” Barnes v. United
States, 

, 870 (Ct. Cl. 1976) (holding the government liable for taking a flowage
easement) (citing numerous cases). Thus, intermittent inundation of land, as contrasted to
continuous overflow, can give rise to a viable permanent taking claim. See 

. Moreover, even a single flooding event may give rise to a taking where the defendant uses
a permanent structure to “purposely flood[] a property once and expressly reserves the right to do
so in the future.” Quebedeaux v. United States, 

, 323 (2013). In that event, the
“defendant’s actions may be viewed not as an ‘isolated invasion,’ but rather as reserving a
flowage easement over the affected property.” 

omitted); see also 

(holding that a permanent physical taking occurred, “even though no particular
individual [was] permitted to station himself permanently upon the premises,” when the
government reserved a “continuous right [of individuals] to pass to and fro”).

        Here, plaintiffs contend that “the [g]overnment’s actions resulted in an invasion by
flooding that preempted [their] right to use and enjoy the protectable real and personal property
interests they owned in the manner expected” and were therefore “sufficiently severe to
constitute a taking.” Pls.’ Br. at 52. To support this assertion, plaintiffs note that the
government-induced flooding caused “[t]he disruption of their lives, the devaluation of their
properties, the destruction of their real and personal property, and their displacement from their
homes and businesses for an extended period.” 

They emphasize that the flooding
“restricted access to and from their property, causing their eviction from their properties for a
period long after the water receded due to necessary repairs [and] significantly limited use of that
property.” 

other hand, the government asserts that the severity factor “favors a
finding of no liability because repairable damage resulting from temporary flooding during a
                                                  30
single flood event is not the type of severe impact that can support a claim for compensation
under the Fifth Amendment.” Def.’s Br. at 101. Stressing that each property was repaired or
capable of repair by the time of trial, the government asserts that such “repairable” damage or
“temporary harm” is “manifestly different from the type of injury that can support a Fifth
Amendment claim.” 

government maintains that some plaintiffs recovered
“significant flood insurance” and received federal benefits that mitigated actual out-of-pocket
expenses, and that the damage attributable to the government action was “relatively minor” for
many of the properties. 

02.

        The flooding at issue here went well beyond a tort and was sufficiently severe to rise to
the level of a compensable taking. The government’s suggestion that this flooding is not a
compensable taking because it was temporary and confined to a single flood event carries no
water. Even if a single event of this nature were insufficient to rise to a taking, the sheer
frequency of significant storms in the region both before and since construction of the dams—the
Hearne storm, the Taylor storm, the 1929 and 1935 storms, Tropical Storm Claudette in 1979,
the 1992 series of storms, Tropical Storm Allison in 2001, and the Tax Day Storm—suggests
that this was more than an isolated event, and that it is likely to recur. See also Tr. 1199:13 to
1200:7 (Kappel) (noting that Harvey’s maximum rainfall was not unprecedented in the region
when Harvey occurred). Indeed, this was not the first time that water had exceeded government-
owned land, and the Corps itself had fully anticipated a storm the likes of Harvey. The future
recurrence of a similarly large storm, producing comparable rainfall, remains likely to occur
again. See Tr. 1198:4-8 (Kappel) (stating that “[t]here is a probability that [a rain event similar
to Harvey] could happen again in the future [over the Addicks and Barker watersheds]”); Tr.
1494:3-11 (Long) (noting the “inevitably recurring” continuation of storms “that are of large
magnitudes that could have impacts similar to those of Harvey”). Nor is there any reason to
expect that the government would, or that it ought to, operate the dams to release more water
downstream any differently in a future storm than it did during Harvey. As noted previously, the
Corps operated the dams as prescribed by the Water Control Manual. Hence, in the nearly
inevitable event of a future storm of significant magnitude, it can be expected that the
government would similarly impound water on plaintiffs’ properties to prevent what would be
catastrophic flooding downstream. As a result, the likelihood of recurrent flooding is high,
weighing strongly in favor of the finding of a compensable taking.20

        The significant harm caused to plaintiffs’ properties, almost entirely preventing their
normal use and enjoyment, is also relevant to the severity analysis. Water measuring as much as
several feet in some cases inundated plaintiffs’ homes—for as long as a week in multiple
instances—destroying substantial personal property, causing structural damage, and rendering
properties uninhabitable or unusable until repairs could be completed months or years later. And
even in the case of the Popovici residence, where water came within inches but did not actually


       20
          The government also avers that plaintiffs suffered “no lasting infringement on their
property rights” differing from those of “thousands of other Texans in nearby areas whose homes
flooded.” Def.’s Br. at 102. But contrary to the government’s contention, the flooding on
plaintiffs’ property did differ from that experienced by others because it was directly caused not
by the storm itself but by the impoundment of water behind the dams, and that infringement is
lasting because the government reserves the right to repeat the impoundment in the future.

                                                31
enter the home, seepage around the foundation caused structural damage. See Tr. 1242:13 to
1243:6 (Popovici). Some homeowners expressed concern about the substances and materials
absorbed into the soil from the composition of the water itself, which became putrid, smelling of
“fecal material and dead animal material and chemicals.” Tr. 66:5-12 (Site Visit) (Popovici).
Given that the water at the test properties was classified as Category 3 “black water,” i.e., water
with “a greater potential to harbor pathogens, including sewage, chemicals, fertilizer, [and]
organic material,” these fears are not unfounded. Tr. 2892:14-24 (Hooper).

        Furthermore, while water was present it prevented basic ingress and egress at all the
properties, with some accessible only by watercraft. See, e.g., Tr. 1299:1-12 (Micu). Notably,
the amount of water that actually entered the structures is not reflective of, and actually much
less, than the severity of the water level outside the structures in the lawn and streets. See Tr.
1952:24 to 1953:21 (Bedient). The streets in these areas were designed to serve as a drainage
system that channeled water, and to that effect the elevation of street levels is lower than the
elevation of structures by at least 18 inches, meaning that the water would be much deeper in the
streets and lawns than in the structures themselves. See 

to the severity analysis is the substantial decline in property value caused
by the flood event and the likelihood of similar events in the future. The court takes judicial
notice of a recently enacted Texas law that requires a seller of residential real property to
disclose to potential purchasers whether the property is located wholly or partially in a reservoir
subject to controlled inundation by the Corps and whether the property has previously flooded.
See Tex. Prop. Code Ann. § 5.008 (West, Westlaw through the end of the 2019 Reg. Sess. of the
86th Leg.). All the properties at issue here fall within these disclosure requirements and the
adverse impact of the government-induced flooding on their market value is evident.
Additionally, Dr. Bell testified at trial that flooding events of this kind generally cause a decline
in property value of at least 20 to 30 percent, although he did not specifically opine on the
properties involved here. Tr. 1353:13 to 1354:2 (Bell). While the government seeks to discredit
his testimony as “decidedly qualitative, not quantitative,” Def.’s Br. at 104, the severity factor is
by nature a qualitative inquiry, lacking any definite quantitative thresholds. Nor is the dollar
amount of damages in issue at this juncture. Therefore, although Dr. Bell’s testimony did not
provide quantitative analyses for the specific properties, there is little room to question his broad
conclusions about the severe impact of this kind of flooding at these kinds of properties on
market value. Simply put, the absence of specific quantitative calculations does not serve to
discredit those conclusions. Likewise, Mr. Deal’s expert opinion concluded that plaintiffs’
properties “suffer[ed] permanent damage, damage that wouldn’t be healed by itself” and “would
require [a] significant amount of investment and risk of capital in order to get them all the way
back to [being] habitable.” Tr. 2210:6-10 (Deal). After inspecting the residential properties
involved here, Mr. Deal identified nearby comparable sales and compared the two. See
generally PX2205. He concluded “[t]hat the inundated properties suffered a significant
diminution in price levels.” Tr. 2210:25 to 2211:22 (Deal).

        The government maintains that the testimony of Dr. Bell and Mr. Deal simply “show[s]
that the flood-related impacts to the [t]rial [p]roperties were temporary and repairable,”
suggesting that such impacts do not rise to the level of a taking. Def.’s Br. at 105. But contrary
to the government’s assertion, the fact that property has been or could be, with sufficient outlays,
restored to its pre-flood condition is not a relevant consideration in the severity analysis. See
Arkansas Game & 

(finding a compensable taking even though the
                                                   32
damaged terrain could be repaired by “costly reclamation measures”); see also 

(holding that “no use to which [plaintiff] could subsequently put the property by his
reclamation efforts changed the fact that the land was taken when it was taken and an obligation
to pay for it then arose”). Under the government’s theory, seemingly any takings claim based on
government-caused damage to property could not be sustained if the damage were susceptible to
repair. Even catastrophic damage can often be mended by enough time and expense, but the
mere capacity for repair in no way mitigates the severity of the harm itself. Furthermore, the
government’s argument that the ability to repair damages caused by flooding weighs against
imposing liability on the Corps is also heedless of the recurrent nature of the flooding involved
here. Plaintiffs may have—in many, if not most, instances—been able to repair their real
property, if not their personal property, but the taking here involves more than the damage
already incurred; it encompasses a loss of the property owners’ right to exclude future floodwater
incursions onto their land and into their homes. Thus, that most of the bellwether plaintiffs were
able to repair their property is likewise irrelevant to their inability to prevent future government-
induced flooding on that property. It is not defensible to propose that an action which destroys
property, all the while reserving the right to do so again, is not a taking simply because the
property owner can, at great inconvenience and expense, repair the property after the fact.
Moreover, the government fails to consider that some plaintiffs could not even afford to conduct
the necessary repairs to render their homes habitable. Plaintiff Scott Holland, for example, could
not afford to fix his property, was forced to move away, and continued living in a trailer at the
time of trial. Tr. 1844:15 to 1845:11 (Holland).

        Likewise, that some affected property owners recovered insurance money is not apposite
to whether the government deprived plaintiffs of their right to use and enjoy their property.
Insurance proceeds that mitigate the amount of out-of-pocket expenses incurred to repair
damages in no way lessens the degree of harm caused by the initial infliction of that damage.
The same is true of other government benefits plaintiffs may have received, such as FEMA
grants. Independent awards of aid might have lessened the ex post facto cost of recovery for
some plaintiffs and be relevant to damages, but that has little bearing on whether the government
effected a taking initially.

        Given the extensive damage caused to plaintiffs’ real and personal property, their
inability to exercise the right to exclude floodwaters, the interference with their right to use and
enjoy their property, the high likelihood of recurring floods, and the significant diminution of
property values, the court finds that the severity calculus weighs in favor of finding a taking.

               iii. Benefit to the government.

        Another consideration in the takings analysis is whether the invasion “appropriate[s] a
benefit to the government at the expense of the property owner,” Ridge 

,
as opposed to inflicting a mere “consequential” injury, id.; 

. The line
between which destructions of property by government action are compensable takings and
which are simply consequential, and therefore subject to a tort analysis or not compensable, is
not always easy to distinguish. But when the direct result of the government’s actions is the
destruction of property for its own, and thus the public’s, benefit, the affected property owners
are entitled to just compensation for a taking. See 

-49. In the case of
Addicks and Barker, the government received a notable benefit at the expense of the upstream
private property owners. That the dams protected downstream Houston is not the point. It
                                                 33
suffices to say that, consistent with the purpose for the construction of the Addicks and Barker
flood-control projects, the government protected downstream properties from an estimated $7
billion in losses during Harvey, see Tr. 164:24 to 165:8 (Thomas), while concurrently causing
upstream properties to suffer from severe flooding.

         The government argues that “the direct . . . result of the government action is to reduce
the risk of catastrophic downstream flooding” and “[t]hat such [upstream] flooding occurred in
connection with Hurricane Harvey was merely a consequential result.” Def.’s Br. at 82. But the
precedents suggest otherwise. This is not a case where the damage to plaintiffs’ land was a
residual effect of government actions on other property. Cf. Southern Pac. Co. v. United States,

, 432 (1923) (finding that the injury to plaintiff’s railroad from construction of a
jetty nearby that may have altered ocean currents was not compensable as a taking because the
injury was incidental and indirect to the government actions). Notably, here the same actions
which benefitted the downstream properties are those which caused harm to plaintiffs. The
damage to plaintiffs’ properties was the direct result of the government’s construction,
modification, and operation of the Addicks and Barker Dams, reflecting the sheer fact that
plaintiffs’ properties are, by government design, within the dams’ flood-pool reservoirs. The
flooding suffered by plaintiffs and the associated “damages were not merely consequential. They
were the product of a direct invasion of [the plaintiffs’] domain.” 

; see
also 

(“[T]his is not a case of temporary flooding or of consequential
injury, but a permanent condition, resulting from the erection of the lock and dam, by which the
land is subject to frequent overflows of water from the river.”). The invasion of the Addicks and
Barker flood pools onto plaintiffs’ properties from the construction and modification of the dams
and their operation during Harvey, appropriated a benefit to the government at the direct expense
of inflicting significant injury to plaintiff property owners. Therefore, this factor weighs in favor
of plaintiffs.

           b. Intentional or foreseeable.

        Arkansas Game & Fish next requires the court to assess “the degree to which the
invasion is intended or is the foreseeable result of authorized government 

. A taking occurs either where the government intended to invade the property or where the
invasion is the “direct, natural, or probable result of an authorized activity and not the incidental
or consequential injury inflicted by the action.” Ridge 

(quoting Columbia
Basin 

). The presence of the disjunctive “or” in the Arkansas Game
& Fish factors, 

(“intended or is the foreseeable result”) (emphasis added),
makes evident that one of these circumstances must be present to support the finding of a taking.
See 

(“[P]laintiffs need not allege or prove that defendant specifically
intended to take property. There need be only a governmental act, the natural and probable
consequences of which effect such an enduring invasion of plaintiffs’ property as to satisfy all
other elements of a compensable taking.”) (internal citations omitted). Despite being separate
inquiries, the two factors are interrelated—one cannot find intent without foreseeability; but what
is an objectively foreseeable result may not have been the intended result. See John Horstmann
Co. v. United States, 

, 146 (1921) (“[I]t would border on the extreme to say that the
government intended a taking by that which no human knowledge could even predict.”);
Columbia Basin 

(“Such [results were] not the direct, natural or
probable consequence[s] of the Government’s act, and for this reason no intent to take can be

                                                 34
implied.”). Here, both intent and foreseeability were present. Because foreseeability bears on
intent, foreseeability is addressed first.

               i. Foreseeability and causation.

        Whether the asserted invasion is the “direct, natural or probable result of an authorized
activity” is a critical part of the takings analysis. Columbia Basin 

.
Put in other terms, the court should determine here “whether the [flooding] on the claimants[’]
property was the predictable result of the government action.” Ridge 

(citing Sanguinetti v. United States, 

, 149-50 (1924)). If the most that can be said is
that the government’s actions are only “a contributing factor towards” the flooding, in contrast to
the flooding being “the natural or probable consequence” of those actions, then “a tort action
may lie in the proper forum for such an incidental or consequential injury,” but not an action for
taking. Columbia Basin 

.

        The invasion asserted here by plaintiffs is that of impounded floodwaters entering onto
their properties resulting from the Corps’ construction, modification, maintenance, and operation
of the Addicks and Barker Dams. Plaintiffs contend that “the inundation of private properties
from the reservoir pools behind each dam was the predictable result of the [g]overnment’s
actions.” Pls.’ Br. at 30. Plaintiffs support this assertion by arguing that the Corps always
“knew there would be recurrent storms of such a magnitude [as Harvey],” 

and that the
foreseeability of plaintiffs’ properties flooding was obvious in light of the fact that the properties
are located within the maximum pool size for the reservoirs, see 

         Defendant, in response, puts forth several postulates why the inundation was not
foreseeable. Defendant first argues that the flooding of private properties in the reservoirs was
not foreseeable because “Hurricane Harvey was an extraordinarily rare and large storm.” Def.’s
Br. at 58 (heading) (capitals omitted). Further, defendant argues that, “[t]he relevant government
action . . . for purposes of this [foreseeability] analysis should be at the time the Corps
constructed the dams in the 1940s.” 

9. The government avers that “[t]he agency’s
knowledge at th[at] time [was] that [this] particular result is [] possible” and that “does not mean
it is a direct, natural or probable result.” 

(emphasis in original). Additionally, the
government argues that the Corps did not foresee “the resulting damage when it constructed the
Project in the 1940s” and that “[t]o find otherwise would hold the Corps responsible for
unforeseen urbanization.” 

        Defendant’s reliance on the contention that foreseeability in this scenario is most properly
measured from the viewpoint of the government in the 1940s, at the time the Addicks and Barker
Dams were constructed, is not appropriate because the foreseeability inquiry should not be so
constrained. Most importantly, defendant misstates the underpinnings of the foreseeability
analysis. Foreseeability—in contrast to intent, which more aptly accounts for subjective
positions—is not simply measured from the viewpoint of the government; foreseeability is an
objective inquiry. See, e.g., John Horstmann 

(considering whether the
results of the government actions could have been objectively foreseen); 

(explaining how the foreseeability inquiry depends on whether there was “any reason to
expect that such result would follow.”). That is, would an objective person reasonably foresee
that the actual results which occurred would have been the direct, natural or probable results of
the government’s actions? Whether the Corps subjectively foresaw the results may bear on
                                                 35
objective foreseeability, but it is not the only consideration. Therefore, it is irrelevant in this case
whether foreseeability is measured in the 1940s, 1970s, or even in the 2000s, because at all of
these points defendant should have objectively foreseen that the pools could and would exceed
government-owned land.

         Here, the evidence demonstrates the Corps was aware or should have been aware since
the initial construction of the dams and at every point onward, that the flood pools in the Addicks
and Barker Reservoirs would at some point (and thereafter) exceed the government-owned land,
inundating private properties. As early as the 1940s, the Corps understood that storms of
exceptionally large size were possible in the Houston metropolitan area. For example, the Corps
noted in the 1940 Definite Project Report that the Buffalo Bayou watershed is situated “in an
area subject to all of the circumstances making possible large storms” and that “only chance has
prevented the occurrence of a storm over the basin much larger than the 1935 storm.” See JX5 at
7. The Hearne storm of 1899, which served as a basis for the design of the maximum pool size
in each reservoir, occurred only 90 miles northwest of Houston. See 

the
Hearne storm occurred under meteorological conditions that the Corps noted “could be
approximated closely over the Buffalo Bayou watershed”). Notably, the Corps considered that
pool sizes beyond the extent of government-owned land were foreseeable in the 1940s during the
lifetime of the structures, when they conducted a cost-benefit analysis, see generally JX52,
ultimately determining that “the expected damages of inundating pastures and rice fields” would
be less than the cost of buying additional land, see Tr. 200:21-24 (Thomas). To an objectively
reasonable person, it was not a question of whether the pools would reach the level they did—it
was merely a question of when and how often. Just as in Cotton Land Co. v. United States, “The
events which occurred, although they took some time, were only the natural consequences of the
[government’s actions]. If engineers had studied the question in advance[,] they would . . . have
predicted what occurred.” 

, 233-34 (Ct. Cl. 1948). Accordingly, even measuring
foreseeability in 1940, as defendant advocates, leads to a conclusion that pools of a size at or
close to Harvey were objectively foreseeable.

        But the taking at issue here does not begin and end with the construction of Addicks and
Barker. The Corps’ modification, operation, and maintenance of the dams was and is ongoing,
continuing well into the years following the 1940s, and at each successive instance, the
likelihood of occurrence of flood pools exceeding government-owned land grew. By the 1960s
and 1970s, the Corps had a definite understanding that larger pool sizes were highly probable. A
study by the Corps in the 1960s explained that the now-permanent gates on the reservoir
conduits would lead to larger and more permanent pools. See JX15 at 44. In a 1973
memorandum, the Corps’ Chief of the Engineering Division in the Galveston, Texas district
noted that the Corps should “develop a history and rationale for our operating concept of
imposing flooding on private lands without benefit of flowage easement or other legal right.”
See PX37 at 1. A 1974 Corps inspection report echoed similar thoughts. See PX38 at 5
(USACE233674) (1974 Buffalo Bayou Inspection Report) (“Development of the area will
eventually place the [g]overnment in the position of having to flood the area within the reservoir
with the accompanying damages in order to protect downstream improvements in the event of a
severe future storm.”).

        Later events only magnified the risk of flooding beyond government-owned land,
rendering it virtually inevitable. Around March of 1992, a series of storms resulted in then-
record flood pools in both the Addicks and Barker Reservoirs. See Tr. 363:20 to 364:2
                                                 36
(Thomas). This result, known as the “ratcheting effect,” demonstrated that one Harvey-sized
storm was not necessary to create large flood pools—a series of consecutive moderate storms
could have the same effect. See Tr. 363:20 to 364:11 (Thomas). The Corps in the 1990s and
2000s, aware of the increased risk, surveyed properties in the reservoirs located beyond
government land to have a firmer idea as to the extent of the possible damage if flooded. See Tr.
100:5-16 (Thomas) (One such field study was conducted and completed in July 1994; another
study was completed in 2003.). Not only is it evident that the Corps believed flooding beyond
the extent of government-owned land was probable, it is unreasonable to contend otherwise.

        It is true that Tropical Storm Harvey was a record-setting storm. But the evidence
markedly shows that pools of this size and the attendant flooding of private property were, at a
minimum, objectively foreseeable. Thus, Harvey’s magnitude does not exculpate the
government of liability for its actions. Even so, the government suggests that “the claimed losses
were not the direct, natural or probable result” because the Corps could not have foreseen “such
significant development upstream of the reservoirs.” Def.’s Br. at 81 (emphasis added).
Essentially, the government suggests that because the properties that flooded were more
developed, i.e., homes and businesses occupied the land as contrasted to the more rural fields of
the 1940s, it should not be held responsible for the resulting damage. The government, however,
misapplies the foreseeability inquiry. That the monetary amount of damages may be more
significant than initially thought does not detract from the fact that it was foreseeable that the
land would be invaded by floodwater. In short, just because the nature of the invaded land has
changed from farm land to residential does not bear on the question of whether an invasion of
such land should have been foreseen.

        The parties also present opposing views on the causation analysis for the flooding at
issue. Establishing causation is a vital component of the foreseeability inquiry. “In order to
establish causation, a plaintiff must show that in the ordinary course of events, absent
government action, plaintiffs would not have suffered the injury.” St. Bernard Par. Gov’t v.
United States, 

, 1362 (Fed. Cir. 2018). Additionally, “the causation analysis must
consider the impact of the entirety of government actions that address the relevant risk.” 

364. Therefore, the relevant question here is whether the flooding on plaintiffs’ properties
would have occurred but for the government’s actions regarding Addicks and Barker.

        Causation for all thirteen properties was originally contested, see Pls.’ Br. at 39-49;
Def.’s Br. at 63-75, and expert testimony on the subject was presented from both Dr. Bedient for
plaintiffs and Dr. Nairn for defendant. As for ten of the thirteen properties, defendant has
essentially conceded that without the dams these properties would not have flooded. See Tr.
3258:8-12 (closing argument). Defendant’s expert opined that only the “finished first floors on
three of the thirteen upstream Test Properties would have experienced some flooding even in the
absence of the federal project.” DX608 at 166 (emphasis added). With respect to ten properties,
plaintiffs’ burden of causation thus has been met: (1) Banker; (2) Holland; (3) Lakes on
Eldridge; (4) Popovici; (5) Sidhu; (6) Soares; (7) Stewart; (8) Turney; (9) West Houston Airport
Corporation; and (10) Wind.




                                               37
        The remaining three properties require a more thorough analysis: (1) Burnham; (2) Giron;
and (3) Micu.21 The parties presented competing testimony about the causes of the flooding on
these three properties. Plaintiffs argue that “Dr. Bedient’s work establishe[d] that each of the
Test Properties’ [] flooding was in fact caused by the Addicks or Barker [D]am impoundment.”
Pls.’ Br. at 43. In his report, Dr. Bedient concluded that “all of the test properties were flooded
due to the impounding rainfall runoff waters by the [Corps] behind the Addicks and Barker
Dams,” PX526 at 46, and such flooding was not a result of the local drainage systems or due to
riverine flooding, see 

49, 54. Contrastingly, defendant argues that flooding was
unavoidable upstream due to the magnitude of Harvey. Def.’s Br. at 72. That is, defendant
asserts that the flooding on these three properties cannot be attributed to the pools created by the
Addicks and Barker Dams. See 

(arguing that the flooding on the three properties
was attributable to alternative sources such as diversion channels and riverine flooding).

         Dr. Bedient reached his conclusions by studying and analyzing real-time data collected
during the storm, whereas Dr. Nairn reached his conclusions through modeling and projections.
While modeling can be a useful tool for planning and analyzing hypothetical outcomes and at
times may be able to provide more sophisticated insights than even real-time data, in the case at
hand, Dr. Bedient’s analysis was more persuasive. Particularly, Dr. Nairn’s testimony suffered
from a major flaw—a failure to fully capture what actually occurred. For instance, Dr. Nairn’s
model concluded that flooding within the homes on the Giron and Micu properties due to
riverine overbanking had already occurred as of August 27. See DX608 at 125-26. But live
witness accounts and photographic evidence show that water did not enter either home until at
least an entire day later. See Tr. 1999:14 to 2000:13 (Bedient). Additionally, Dr. Nairn’s model
failed to account for stormwater drainage systems and improperly accounted for channel
diversions and drainage projects. See, e.g., Tr. 2002:14-25 (Bedient). These oversights render
the model scenario different from the real-life scenario, and likely caused an overstatement of
Dr. Nairn’s projections of riverine flooding. See Tr. 2004:19 to 2005:2 (Bedient); see also Tr.
1858:6-12 (Lesikar-Martin) (explaining that, in contrast to Dr. Nairn’s assertions, Bear Creek
was not overflowing beyond its banks during Harvey). A predictive modeling system which
relies on incorrect inputs and outputs used to align the model’s coefficients and factors, cannot
provide reliable projections. Accordingly, Dr. Nairn’s model, which relies on input data that do
not match what in fact occurred, cannot be fully reliable. Lastly, Dr. Nairn’s conclusions seem,
in part, to agree with that of plaintiffs’ expert, Dr. Bedient, even as to the three contested
properties. Dr. Nairn concludes that “[P]eak flood elevations at all of the upstream Test
Properties are attributed to backwater due to high pool elevations in Addicks or Barker
Reservoirs.” DX608 at iii (emphasis added). In other words, Dr. Nairn appears to agree that the
water would not have been as high in each of the three contested homes but for the Addicks and
Barker projects.

        Defendant has alleged a number of errors in Dr. Bedient’s calculations. For example,
defendant contends that Dr. Bedient failed to account for cumulative effects, and simply looked
at discrete six- and twelve-hour time periods when collecting certain data. Def.’s Br. at 73. But
these allegations do not suffice to discredit Dr. Bedient’s conclusions. Whether Dr. Bedient’s
model fully accounts for intervening hours does not detract from the fact that his conclusions are


       21
         Given that the causation issues were the same for all three properties, the properties can
be discussed collectively with regard to causation.
                                                38
more reliable because they align with what was actually witnessed.22 As such, plaintiffs have
met their burden of showing that but for the Addicks and Barker project, flooding would not
have occurred to the level it did on the three contested properties.

        Defendant also puts forth in rebuttal one additional argument on causation. Defendant
argues that because the government’s actions that address the relevant risk must be considered in
their entirety, plaintiffs’ failure to account for the impact of the outgrants is fatal to their
argument. See Def.’s Br. at 97-98 (citing St. Bernard 

). Defendant argues
that the outgrants that the “United States allowed to be built on the Project property were built to
reduce flood risks to upstream properties,” and because this is “‘government action [that]
mitigates the type of adverse impact that is alleged to be a taking, it must be considered in the
causation analysis.’” 

Bernard 

). It is defendant, however,
who fails to fully account for all the impacts of the outgrants. Defendant asserts that the
outgrants mitigated the flood risk upstream. To an extent, that allegation is correct. They
effectively allowed water to be removed from the upstream neighborhoods more rapidly. But the
outgrants also had the effect of causing “more frequent” and “larger” impoundments in the
reservoirs and “increase[d] flood damages resulting from reservoir impoundments.” JX52 at 16
(USACE01545). Thus, it cannot be said that the government’s granting of easements for
drainage systems consequently built by developers and local entities, as a whole, provided a
greater benefit than harm. As to the “but for” analysis, it would be wrong to say that but for the
granting of the outgrants, plaintiffs would have been worse off. Notably, the evidence actually
suggests the opposite. Moreover, in the counterfactual scenario where the federal government
refused to grant these easements, the evidence suggests that upstream developers would have
been required to seek other feasible remedies for drainage. See Tr. 817:1-11 (Vogler).

        Thus, considering the totality of the evidence, plaintiffs have met their burden of showing
causation for all thirteen properties. Plaintiffs have sufficiently demonstrated that the inundation
of floodwaters onto their private property was the “direct, natural, or probable result” of the
government’s activity. Ridge 

.

               ii. Intent.

        Intent does not concern whether the government meant to abridge a private property right
but whether it intended to occupy the pertinent property without lawful authority or excuse. See
LaBruzzo v. United States, 

, 474 (2019). Thus, the intent element is present if
the government intended its physical occupation even if it did not intend to effect a taking. See

by the Supreme Court of Texas, “build[ing] a flood-control dam knowing that
certain properties will be flooded by the resulting reservoir” is a “conscious decision to subject
particular properties to inundation so that other properties [will] be spared.” Harris Cty. Flood
Control 

(emphasis added). The requisite intent to invade is present in


       22
          Similarly, defendant’s assertions that Dr. Bedient’s testimony relied upon flawed gauge
data, see Def.’s Br. at 67, are unavailing. Defendant argues that “Dr. Bedient erroneously based
his critique on uncorrected data for [the upper Buffalo Bayou gauge].” 

gauges are
regularly inspected, see Tr. 2173:10-15 (Test. of Jeffrey East), and are considered reliable, 

the gauge was misreading, Dr. Bedient’s report corrects any misreading by comparing
and subsequently aligning the data with what was actually witnessed. See Tr. 3195:9-15.
                                                 39
such cases, and “of course the government must compensate the owners who lose their land to
the reservoir.” 

Tarrant Reg’l Water Dist. v. Gragg, 

, 555 (Tex.
2004) (citing City of Dallas v. Jennings, 

, 314 (Tex. 2004) (“[T]he requisite
intent is present when a governmental entity knows that a specific act is causing identifiable
harm or knows that the harm is substantially certain to result.”)). The government may not,
however, intend an outcome which it did not subjectively foresee as a “direct, natural, or
probable consequence” of its action. See John Horstmann 

.

         Here, the Corps knew from the outset that the land it purchased was inadequate to hold
the amount of water that would be contained in the reservoirs should the embankment-design
storm occur. See JX5 at 26. It knew then that if such a storm transpired, the water produced
would exceed government-owned land and flood private property. 

appears doubtful
that the Corps subjectively foresaw the occurrence of a storm event large enough to create pools
that exceeded government-owned land—although, as already noted, such a storm was objectively
foreseeable at that time, 

. It certainly knew that such a storm was possible
over the Addicks and Barker watersheds, see, e.g., Pls.’ Mot. to Reopen the Trial R. Ex. A at
USACE2019_0000013-14 (recording the Corps’ observation that the occurrence of a storm like
the Hearne storm was not “unreasonable”), but the Corps seems to have reckoned then that it was
an improbable event, see JX5 at 9-10 (concluding that the occurrence of a storm as severe as the
Hearne storm was “very remote”), or at least that it would not occur frequently. This conclusion
is also inferable from the cost-benefit analysis the Corps conducted around this time. To perform
such an analysis, the Corps needed to determine both how much it would cost the government to
flood beyond government-owned land and how frequently that was likely to happen. Comparing
the cost attributable to flooding rural land to what it would cost to purchase rights to the then-
undeveloped land, the Corps determined that the cost of flooding was less. That the calculus
reached the conclusion it did indicates that the Corps regarded such overflow as possible but that
it was willing to take the ensuing risk. That calculus did not withstand the test of time.
Nonetheless, the intent inquiry does not end there.

        Intent is present here because, like foreseeability, intent is not measured at one singular
point in time. Again, this is because the government action at issue is not simply the
construction of the dams, but the totality of their construction, modification, maintenance, and
operation over the project lifespan. The Corps subjectively knew by the 1940s, and particularly
by the 1960s, that storms larger than the design storm were likely to occur over Addicks and
Barker. 

, 31, 36. From that time forward, it had subjective knowledge that pools
exceeding government-owned land were probable at some point. Indeed, by 1973 the Corps
expected the possibility of flooding off of government-owned land to become a public concern.
See PX37 at 1. Thus, intent can be inferred here because the government knew flood waters
would likely occupy plaintiffs’ private properties at some point.

         Equipped with the knowledge that storms of the design storm magnitude were probable,
the Corps did not stray from its primary objective to prevent downstream flooding (indeed, it
probably could not), even when it knew that could well mean impounding water on private
property. For example, the 2012 Water Control Manual, which the Corps followed during
Harvey, instructs the Corps to operate the dams in a manner consistent with their original
purpose: to protect downstream property by impounding water in upstream reservoirs. It states
that “the general plan for reservoir regulation will be to operate the reservoirs in a manner that
will utilize to the maximum extent possible, the available storage to prevent the occurrence of
                                                 40
damaging stages on Buffalo Bayou.” JX110 at 7-4 (USACE016338). Notably, the “available
storage” that was to be “utilize[d] to the maximum extent possible” encompasses all land in the
intended reservoir behind the embankments, including land the government has never owned.
See Tr. 67:12 to 68:3 (Thomas). To accomplish its purpose of downstream protection, the Corps
planned all along to impound water to the maximum extent of the available storage—a
determination that never altered even when the Corps came to understand that rainfall events
reaching the design storm magnitude were probable rather than merely possible. In short, the
government had the requisite intent to invade plaintiffs’ properties because the Corps had been
well aware that storms capable of overflowing government-owned land were likely to occur, and
despite that knowledge it still intended to occupy the property concerned without lawful
authority or excuse. See 

.

           c. Reasonable investment-backed expectations.

        A property owner’s “reasonable investment-backed expectations regarding the land’s
use” is also a factor relevant to the takings inquiry under Arkansas Game & 

(citing Palazzolo v. Rhode Island, 

, 618 (2001)). As a threshold matter, plaintiffs
assert that this factor should not even be considered here because the concept applies only to
regulatory, not physical, takings. See Pls.’ Br. at 116-17 n.541. They correctly observe that
“time and again, the Supreme Court has underscored the distinctness of [the physical and
regulatory] lines of takings cases,” 

n.541, and there is no question that the reasonable
investment-backed expectations factor is ordinarily applied in the context of regulatory, and not
physical, takings. See, e.g., Penn Cent., 

. Noting this difference, the Federal
Circuit stated in Preseault v. United States that “[t]he Government’s attempt to read the concept
of ‘reasonable expectations’ as used in regulatory takings law into the analysis of a physical
occupation case would undermine, if not eviscerate, long-recognized understandings regarding
protection of property rights; it is rejected categorically.” 

, 1540 (Fed. Cir. 1996).
See also Palm Beach Isles Assocs. v. United States, 

, 1364 (Fed. Cir. 2000)
(explaining that “‘reasonable investment-backed expectations’ are not a proper part of the
analysis” in physical takings cases); 

(citing and quoting Love
Terminal Partners, L.P. v. United States, 

, 1345 (Fed. Cir. 2018) (“The reasonable,
investment-backed expectation analysis is designed to account for property owners’ expectation
that the regulatory regime in existence at the time of their acquisition will remain in place, and
that new, more restrictive legislation or regulations will not be adopted.”)) (emphasis added).

        The precept plaintiffs reach from this line of precedents is that the inclusion of the
investment-backed expectations factor in the Arkansas Game & Fish listing was not intended to
translate that factor outside the regulatory takings context. They seek to diminish its
enumeration as a relevant factor in Arkansas Game & Fish by noting the preamble to the list of
factors, which states that the ensuing factors are relevant “[w]hen regulation or temporary
physical invasion by government interferes with private 

.

        Even so, flooding cases can pose an exception to the quotidian rule that physical takings
do not involve consideration of “reasonable investment-backed expectations.” Plaintiffs fail to
take account of the context in which Arkansas Game & Fish arose. Arkansas Game & Fish
plainly involved a physical, not regulatory taking, but the Court nonetheless included the factor
as relevant for guiding the decision on remand. Although citing Palazzolo, a regulatory takings
case, for inclusion of the factor, the Court applied it to the physical taking before it because it
                                                 41
had accepted the finding that the flooded area at issue had flooded previously. The prior flooding
had occurred fairly often following Spring rains, but that flooding was transient and did not
affect the growing season of the management area’s forest. 

. Extensive
flooding that stretched over the growing season was quite a different matter. 

Court
acknowledged the plaintiff’s expectations that flooding at certain times and of limited duration
was possible, but that the flooding involved in the taking claim was of a different kind than that
which they could have anticipated or had previously encountered.

        The context of the case at hand is strikingly similar. In this case, the properties are
located in a geographical area that is generally susceptible to large storms and potential flooding,
and the landowners were aware of that fact. But the flooding that caused the alleged taking
before the court was different in kind from that which had occurred naturally and from what
plaintiffs had reason to anticipate; it was more severe than any prior flooding and it was not the
result of natural conditions but rather of deliberate government action. Reasonable investment-
backed expectations are therefore as equally applicable here as they were in Arkansas Game &
Fish.23 Despite the evident tension of transposing this factor from the regulatory to the physical
takings context, Arkansas Game & Fish clarifies that reasonable expectations are a relevant
consideration in connection with physical takings cases of this particular nature.

        Informing the application of the factor are two considerations. First, the landowner’s
expectations must be “reasonable,” meaning that while this is a fact-intensive inquiry, “it is
nonetheless an objective one.” Chancellor Manor v. United States, 

, 904 (Fed. Cir.
2003). Second, the matter at issue is a question of degree, that is “the extent to which the
[government action] interferes” with those expectations. 

.
Significantly, it is not the case that a takings claim must fail simply because a property owner
“acquired [] land while on notice that a taking was occurring or had the potential to occur.” In re
Upstream Addicks & 

(citing 

); see also
Cooper v. United States, 

, 764 (Fed. Cir. 1987) (finding a taking where the plaintiff
acquired property while on notice that the government-induced flooding was already occurring).
The law offers the government no loophole whereby it may escape takings liability by putting
landowners on notice of the risk that it could or would take their property. See 

(rejecting the “sweeping rule” that “a purchaser or a successive title holder . . . is
deemed to have notice of an earlier-enacted restriction and is barred from claiming that it effects
a taking.”). In short, the government gains no immunity for an uncompensated taking by giving
advance notice that it will take property. When the taking actually occurs, it still must provide
compensation.

       Even if notice had a bearing, plaintiffs would still prevail here because they neither knew,
nor reasonably should have known, of the risk posed by the dams. The government nonetheless
maintains that plaintiffs fail on this factor because “they lacked an objectively reasonable
expectation that their properties would not flood in a Hurricane Harvey-like event.” Def.’s Br. at



       23
          Perhaps the Supreme Court’s inclusion of the words “investment-backed” invites too
strong a reference to regulatory takings law. Simply referring to “reasonable expectations”
would capture the context in which the Court used the factor in Arkansas Game & Fish.

                                                42
106.24 To support this contention, it emphasizes that plaintiffs had notice of the possibility of
flooding, pointing out that “Harris and Fort Bend Counties have a long history of flooding during
large storms.” 

cites publicly available information demonstrating the possibility of
upstream flooding during large storms, notifications proliferated by local governments, and the
frequency with which residents in the region purchased flood insurance. See 

that not one of them had any knowledge that their property was situated in a reservoir,
see Pls.’ Br. at 117, and assert that there is no reason to think plaintiffs reasonably should have
known about that particular risk, different from natural flooding. See Pls.’ Reply at 24. See also
Tr. 1729:10-17 (Banker); Tr. 1758:15 to 1760:3 (Burnham); Tr. 1651:8-23 (Giron); Tr. 1834:14-
16 (Holland); Tr. 1413:15 to 1414:5 (Lakes on Eldridge); Tr. 1293:24 to 1294:15 (Micu); Tr.
1225:2-17 (Popovici); Tr. 1738:9-17 (Sidhu); Tr. 1076:22 to 1078:3 (Soares); Tr. 1607:19-22
(Stewart); Tr. 2151:16-20 (Turney); Tr. 1626:1 to 1627:10 (Wind); Tr. 2120:20 to 21221:5
(Lesikar).

        It is undisputed that plaintiffs did not know their properties were located within the
reservoirs and subject to attendant government-induced flooding. The point of contention here is
whether plaintiffs objectively ought to have known about that risk based on notice.25 First, the
government points to “[p]ublic documents” that discuss the “possibility of upstream flooding
during large storms.” Def.’s Br. at 106. But the mere fact that information is available does not
make it reasonable to assume that plaintiffs should have known about it or, if they did, that they
would understand that it related to government-induced flooding. Even if plaintiffs are assumed
to be aware of information in places such as Key Maps, FEMA Maps, or United States
Geological Survey quadrangle maps—an assumption that is hardly a given—the import of data
on these maps is far from obvious. For example, to infer the possibility of flooding from the Key
Maps would require a baseline knowledge about property elevations, something the average
homeowner does not generally know. It is highly tenuous to suggest that the average citizen
should know how to read and understand the information in these maps or recognize that the map
annotations refer to government-induced flooding rather than naturally occurring flooding.

        Next, the government cites the subdivision plats, which indicate that land was subject to
controlled inundation, as evidence that plaintiffs had notice of the risk when they purchased their
property. See, e.g., Def.’s Br. at 113-14. These subdivision plats are replete with miniscule


       24
         The court perceives the irony of the government’s simultaneous contentions that the
Corps could not have anticipated a storm of Harvey’s magnitude but that plaintiffs ought to have
foreseen the risk of their properties flooding in such an event.
       25
           At trial, the government sought to introduce the testimony of Dr. Gerald Galloway, a
retired Brigadier General in the Corps, as an expert witness. The government proposed that Dr.
Galloway testify as to “indicators that are available to laypeople that they could consider when
making a decision such as purchasing property.” See Tr. 2544:10-12 (Test. of Gerald Galloway).
The court declined to certify Dr. Galloway as an expert witness because the government did not
satisfy its burden to show by a preponderance of the evidence that the testimony was based on
scientifically valid principles rather than a subjective belief or unsupported speculation. See Tr.
2580:12 to 2581:17 (Galloway). Likewise, the proposed expert’s opinions about the subjective
views of the landowners were far less credible than the testimony offered by plaintiffs
themselves of their own personal knowledge and belief.
                                                   43
details. See, e.g., DX557. Even if one were to examine the plats, which it appears no purchaser
actually did, see, e.g., Tr. 1295:6-9 (Micu); Tr. 1660:14-16 (Giron), it would take an
uncommonly attentive eye to notice and comprehend the import of such a “disclosure.”
Moreover, the government’s own witness, the Fort Bend County Drainage District’s Chief
Engineer, testified that the plat language was not successful in informing the public of the risks
involved. See Tr. 682:10-16 (Vogler). Additionally, the fact that the Corps discussed the
possibility of upstream flooding with developers is not evidence that anyone who subsequently
purchased that property also should have been apprised of the information. The government
further cites the high rate of flood-insurance purchases compared to the national average,
concluding that this demonstrates that “the possibility of upstream flooding has long been
knowable in this region.” Def.’s Br. at 106. This argument also fails because no one disputes
that the Houston region is, and long has been, especially flood prone. Because that general flood
risk was well known, and because some residents purchased flood insurance to account for it, is
simply not evidence that plaintiffs should have been aware of the specific risk associated with
the very different type of flooding at issue here, namely, government-induced flooding.

        Perhaps the government’s strongest argument on the issue of notice is the fact that both
the Corps and local governments conducted public meetings, in which they disclosed
information about the possibility of flooding, during the decades leading up to Harvey. Def.’s
Br. at 106. But here, too, the government fails to show that plaintiffs should reasonably have
known of the risk. The mere fact that meetings occurred does not mean they were effective at
communicating the risk such that the public should have known about government-induced
flooding; there is no evidence that these meetings were heavily attended or particularly well
publicized in the community. And in rapidly developing suburbs of a large city like Houston, it
is reasonable to expect a regular flow of people moving in and out of the area, further reducing
the likelihood that new residents adapting to the area would know of the risk without an
especially aggressive public campaign. That not one of the plaintiffs in this case was aware of
the situation regarding government-induced flooding is also telling with respect to the
effectiveness of these meetings, suggesting that it is quite reasonable to conclude that the average
person in the community was likely unaware of the risk.

         Having determined that plaintiffs’ investment-backed expectations were reasonable, the
court next addresses the extent to which the government’s action interfered with those
expectations. Plaintiffs purchased their property for the same varied reasons people generally
buy real estate, e.g., for a home to live in safely or as an investment. See, e.g., Tr. 1704:12-19
(Banker) (noting that the property was purchased as home for retirement and was considered an
investment that would appreciate). As already noted, the degree of interference with these
expectations was acute—rendering properties uninhabitable for a significant time, requiring
substantial outlays to perform repairs, and resulting in a significant diminution in the resale value
of inundated properties. Subsequent developments prompted by the flooding, such as the
recently enacted Texas statute mandating disclosure when property is situated in a reservoir, can
further be expected to diminish property market value. Therefore, the court concludes that the
government-induced flooding severely interfered with plaintiffs’ reasonable investment-backed
expectations.

       Overall, each of the factors identified in Arkansas Game & Fish supports the finding of a
taking of a flowage easement on all thirteen of the bellwether test properties.

                                                 44
                                     B. Defenses to Liability

        The court must determine whether any of the government’s defenses would preclude the
finding of liability. The government asserts two defenses. First, the government argues that its
actions constituted an exercise of police powers, such that no viable taking claim exists. See
Def.’s Br. at 87-91. Second, the government argues that “the doctrine of necessity [] ‘absolv[es]
the State of liability for the destruction of real and personal property.’” See 

(citing
TrinCo Inv. Co. v. United States, 

, 1377 (Fed. Cir. 2013)) (internal citations
omitted). These defenses, however, are inapplicable to the case at hand.

        The government first argues that “[p]articularly in an emergency, where the government
action is part of an effort to reduce or mitigate unavoidable harms to the public, no viable taking
claim exists.” Def.’s Br. at 88 (citing Miller v. Schoene, 

, 279-80 (1928)) (emphasis
added). But that argument cuts against the defense, because the flooding at issue here was not an
unavoidable harm. Defendant asserts that in the situation at hand, the Corps had little to no
choice on how to act when Harvey hit, and that in an effort to protect lives, the Corps operated
the project in accordance with the 2012 Water Control Manual. See Def.’s Br. at 89-90. That is,
the Corps could open the gates and risk more severe downstream flooding or keep the gates
closed, as it did, flooding upstream properties. When Harvey struck, it was true that certainly
“the actions available to the government for dealing with the relevant emergency were
constrained by the design of the dams and impoundments, the Corps’ 2012 Water Control
Manual, and the Corps’ normal operating procedures.” See In re Upstream Addicks & 

. But these constraints only existed because the Corps’ design of the dams
contemplated flooding beyond government-owned land onto private properties. “Thus, it was
not that the government had to respond to Tropical Storm Harvey as an emergency that
necessitated the flooding of private land,” but rather that the government had made a calculated
decision to allow for flooding these lands years before Harvey, when it designed, modified, and
maintained the dams in such a way that would flood private properties during severe storms. 

now claim that this harm was unavoidable when it planned for years to
impound floodwaters onto plaintiffs’ properties.

        Similar reasoning applies to the government’s necessity defense. That defense rests on
the notion that “in times of imminent peril—such as when fire threatened a whole community—
the sovereign could, with immunity, destroy the property of a few [such] that the property of the
many and the lives of many more could be saved.” Def.’s Br. at 91 (quoting 

). Three requirements must be met for the necessity doctrine to apply: (1) “actual
emergency;” (2) “imminent danger;” and (3) “actual necessity of the [g]overnment action.”

. That this case involved a severe tropical storm, and a record-breaking
one at that, is not enough to infer an actual emergency. See 

(rejecting this court’s
“decision to extend the doctrine of necessity to automatically absolve the [g]overnment’s action
in any case involving fire control”). Where, as here, the government is responsible for creating
the emergency, granting the government immunity from liability under the necessity doctrine
would “stretch[] the doctrine too far.” 

term “emergency,” according to both
common usage and definition, refers to “a state of things unexpectedly arising.” Emergency,
Oxford English Dictionary,
https://www.oed.com/view/Entry/61130?redirectedFrom=emergency#eid (last visited Dec. 17,
2019) (emphasis added). The invasion alleged here was by no means unexpected—the Corps
knew that when a severe storm like Harvey came, flooding beyond the extent of government-
                                                 45
owned land upstream would result, in light of the design of the dams and the plans for their
operation. Thus, the necessity defense cannot apply here, because it cannot be said that
“necessity” existed in this case, when the flooding that occurred was the direct result of
calculated planning.

                                          CONCLUSION

        For the reasons stated, the court finds that the government’s actions relating to the
Addicks and Barker Dams and the attendant flooding of plaintiffs’ properties constituted a taking
of a flowage easement under the Fifth Amendment. Thus, the court finds defendant liable.26

        Because liability and damages were previously bifurcated, a plan for addressing damages
must now be put in place. The court proposes to adjudicate damages for five out of the thirteen
test properties. To that end, the parties shall each propose three properties for consideration as to
damages, thus providing to the court with a total of six candidates. The court will then select
five test properties from the six properties proposed. Each party is requested to file with the
court a notice detailing its three proposed test properties for damages and its respective
arguments for selection of those properties as bellwethers by January 21, 2020.

       It is so ORDERED.


                                                    s/ Charles F. Lettow
                                                    Charles F. Lettow
                                                    Senior Judge




       26
          The court had previously deferred resolution of the government’s earlier motion to
dismiss, see In re Upstream Addicks & 

(acting pursuant to RCFC
12(i), taking into account the fact-intensive inquiry involved). In light of the detailed post-trial
findings of fact and conclusions of law in this decision, that motion to dismiss is DENIED.

        Further, pursuant to the court’s previously stated reasons, 

n.9, Pls.’ Mot. to
Reopen the Trial R., ECF No. 245, is GRANTED, subject to the inclusion of the sworn statement
included in Def.’s Opp’n to Mot. to Reopen the Trial R., ECF No. 254. Also pending before the
court is Def.’s Mot. to Correct [the Trial] Transcript, ECF No. 241. This motion is GRANTED
as to those requests not opposed by plaintiffs, see Pls.’ Opp’n to Def.’s Mot. to Correct [the
Trial] Transcript, ECF No. 243, but the requests for correction opposed by plaintiffs are
DENIED.
                                                 46