GARLAND, Circuit Judge:

Shawali Khan, a detainee at the United States naval base at Guantanamo Bay, Cuba, appeals the denial of his petition for a writ of habeas corpus. The district court found that Khan was "part of" Hezb-i-Islami Gulbuddin (HIG), an associated force of al Qaeda and the Taliban engaged in hostilities against the United States and its coalition partners. Khan's primary contention on appeal is that the district court erred in concluding that intelligence reports offered by the government to prove his membership in an HIG cell were reliable. Finding no error in the district court's careful consideration of the evidence, we affirm its denial of Khan's petition.

I

Khan is an Afghan citizen who, at the time of his capture in mid-November 2002, lived in Kandahar, Afghanistan. By the government's account, Khan's affiliation with HIG began in the 1980s, when he served as a radio operator for the group under the command of his uncle, Zabit Jalil, during the anti-Soviet jihad. The government alleges that, after September 11, 2001, HIG formed an alliance with al Qaeda and the Taliban to attack U.S. and coalition forces operating in Afghanistan. According to contemporaneous reports by U.S. Army intelligence collectors, three Afghan informants told the collectors about the presence of a small HIG terrorist cell in Kandahar.1

One of the informants (hereinafter Informant A), later identified as a disaffected member of the HIG cell, told the intelligence collectors in late October 2002 that Khan was a communicator for the cell, facilitating radio contact among its members. He also reported that the cell was directed from Quetta, Pakistan, by Khan's uncle. Intelligence Information Report (IIR) 6 044 0266 03 at 1-3 (Oct. 29, 2002) (J.A. 1731-33). According to Informant A, this cell was responsible for explosions near the Kandahar airfield and was planning additional attacks on U.S. forces using radio-controlled binary explosive devices positioned on roads frequented by U.S. military vehicles. IIR 6 044 0249 03 at 1-4 (Oct. 29, 2002) (J.A. 1726-29).

Informant A gave the intelligence collectors specific information about the cell's method of operation. For example, he described how the cell would detonate two explosions timed to maximize casualties:

Id. at 3 (J.A. 1728); see IIR 6 044 0267 03 at 3 (Oct. 30, 2002) (J.A. 1738). Informant A also identified the precise radio frequencies the cell used in a particular sequence to detonate the two explosions. IIR 6 044 0249 03 at 3 (J.A. 1728); Classified J.A. 1728; see Gov't Br. 10. And in another meeting, he provided additional specific details about some of the cell's intended targets and the type of explosive device used — "a Chinese or Russian antitank mine approximately 12 inches in diameter" connected to an "electric blasting cap" that is in turn "attached to the radio-controlled electronic detonator." IIR 6 044 0267 03 at 2-3 (J.A. 1737-38).

Further details about Khan's role in the Kandahar HIG cell came from a second Afghan informant (hereinafter Informant B), who "did not personally witness the events he described," but had "indirect access" to the information. Decl. of Intelligence Collector 1 at ¶ 23 (Mar. 15, 2010) (J.A. 1543); see Decl. of Intelligence Collector 2 at ¶ 20 (Mar. 10, 2010) (J.A. 1572). Informant B reported that Khan was "a go-between and a facilitator" within the cell, that he "use[d] [his] oil shop to conduct meetings and as a contact point with other members within the cell," and that on November 9, 2002, he "delivered a radiocontrolled binary detonation device and two blasting caps to an operative working within his organization." IIR 6 044 0025 03 at 3 (Nov. 9, 2002) (J.A. 1723).

Finally, a third informant, an Afghan government official (hereinafter Informant C), revealed more about the Kandahar HIG cell's plans to attack U.S. and coalition vehicles at two specific locations. IIR 6 044 0300 03 (Nov. 3, 2002) (J.A. 1744-45). He told the intelligence collectors that "mines ha[d] already been emplaced" at these locations, but that "they ha[d] not been armed with a remote detonation device." Id. (J.A. 1745).

In mid-November 2002, U.S. military officials decided to neutralize the HIG cell. They planned an operation to capture Khan at his shop, using Informant A's tip that Khan would be there at a particular time. The operation was a success, and Khan's home and shop were searched after his arrest. As reflected in several reports and underlying exhibits, the searches uncovered a variety of physical evidence, including a notebook containing Arabic writing about various terrorist activities, a book of poems written in Arabic by a high-level al Qaeda leader, a notebook from the "al-Farouk camp" containing Arabic writing about the use and maintenance of weapons, a notebook with Arabic writing and pictures of Osama bin Laden, Pashtu music praising the Taliban, and a receipt for weapons. See Gov't Br. 13 n. 4 (summarizing material recovered from Khan's properties).

In addition, and more significant, heavily redacted classified intelligence reports state that the search of Khan's properties yielded further, particularly incriminating evidence that confirmed his role in the Kandahar HIG cell. Gov't Br. 13, 55. One of those reports states that [Redaction 1].

On November 22, 2002, Informant B told the intelligence collectors that, after Khan's capture by U.S. forces, Khan's uncle called a meeting with several HIG operatives in Pakistan to replace the entire Kandahar cell. IIR 6 044 0433 03 at 1-2 (Nov. 22, 2002) (J.A. 1753-54). According to the government, improvised explosive device (IED) attacks stopped in the area for about two months after Khan's capture. Decl. of Intelligence Collector 2 at ¶ 15 (Mar. 10, 2010) (J.A. 1570).

During an interrogation by U.S. personnel on December 17, 2002, Khan admitted to possessing [Redaction 2], but said that [Redaction 3].2 In subsequent interrogations, he at times admitted to possessing other pieces of physical evidence, and offered exculpatory explanations for some of them.3 In several interrogations, he also admitted that he had been part of the HIG army during the fight against the Soviet Union in the 1980s and that his uncle Jalil had commanded HIG forces against the Soviets; but at other times, he said he had never been an official member of HIG and/or denied any affiliation with HIG. See Pet'r Br. 11-12 (citing interrogation reports). In numerous interrogations, Khan told his questioners that he was just a shopkeeper, not a terrorist, and had been falsely accused by corrupt Afghans who told lies about him for money.

Khan has been detained at Guantanamo Bay since early 2003. The government contends that Khan's detention is lawful under the Authorization for Use of Military Force (AUMF) that Congress adopted after the September 11, 2001, al Qaeda attacks against the United States.4 We have held that the AUMF grants the President authority (inter alia) to detain individuals who are "part of forces associated with Al Qaeda or the Taliban." Al-Bihani v. Obama, 590 F.3d 866, 872 (D.C.Cir.2010); see Barhoumi v. Obama, 609 F.3d 416, 432 (D.C.Cir.2010).

In June 2008, the Supreme Court ruled in Boumediene v. Bush that "the constitutional privilege of habeas corpus" extends to aliens detained as enemy combatants at Guantanamo. 553 U.S. 723, 732, 128 S.Ct. 2229, 171 L.Ed.2d 41 (2008). Shortly thereafter, Khan filed a petition for a writ of habeas corpus. Early in the litigation, Khan moved for judgment on the record, contending that extensive discovery was unnecessary because the government had failed to produce sufficient reliable evidence to justify his detention.

In ruling on Khan's motion, the district court found that the Army intelligence collectors' informant reports lacked adequate indicia of reliability because, inter alia, "all sources are confidential (i.e., unidentified)." Khan v. Obama, 646 F.Supp.2d 6, 13 (D.D.C.2009). (The government had not yet revealed the informants' identities.) Accordingly, the court held that the informant reports could not, absent other reliable corroborating evidence, justify Khan's detention. At the same time, however, the court found that the government had already produced other reliable evidence suggesting that "petitioner was active in HIG during jihad against the former Soviet Union; HIG was a terrorist organization at the time of petitioner's capture;... petitioner was a HIG communicator at the time of his capture; and petitioner possessed some al Qaeda- or Taliban-related material at the time of his capture." Id. at 19. It also found that the government had "provided enough evidence to show that HIG qualifie[d] as an associated force ... engaged in hostilities against the United States or its coalition partners." Id. (internal quotation marks omitted). The court therefore allowed discovery to proceed, concluding that "the evidence that remains is sufficient ... to warrant denial of petitioners's motion." Id. at 20.

To address the district court's concerns about the informant reports, the government submitted new declarations by U.S. Army intelligence collectors. These collectors were the same ones who had interviewed the Afghan sources in 2002, authored the original reports, and participated in Khan's capture. The collectors identified two of the key informants by name and the third by position. In their declarations, the collectors stated that, based on their professional training and experience, and on their direct contact with the informants, they assessed the informants as reliable sources. They also set forth, in considerable detail, why they reached that conclusion. In so doing, the collectors explained both the manner in which they assessed the informants' motives and how elements of the informants' stories were independently verified. See Decl. of Intelligence Collector 1 (J.A. 1528); Decl. of Intelligence Collector 2 (J.A. 1556).

On May 13, 2010, the district court began a three-day evidentiary hearing. The government presented the various reports and exhibits described above, as well as others that will be discussed below. It also introduced three items it said were seized from Khan's property: the notebook containing Arabic writing about terrorist activities; the book of poems written in Arabic; and the al-Farouk notebook containing Arabic writing about weapons. Khan relied on the portions of his interrogation statements that were exculpatory, see, e.g., supra note 3; presented an expert witness to testify about HIG's history and activities; and testified himself via video link. He said he had remained at his family farm during the Soviet occupation and had not fought against the Soviets, but that he had assisted those who did fight. He repeated that he was just an innocent shopkeeper and denied participating in any attacks against Americans. He also said that he was not a member of any terrorist group, that HIG had not been in Kandahar since the Taliban came to power, that Afghans rather than Americans had seized items from his home, and that he did not recognize the items the government introduced into evidence. Hr'g Tr. 11-22, 35 (May 17, 2010) (J.A. 301-12; 324-25).

During the hearing, the court expressed concern about the heavy redactions of information in the classified intelligence reports describing evidence seized from Khan's properties, including particularly incriminating items that were not themselves introduced. The court said the redactions made it impossible to determine the source and timing of the reports or to assess their reliability. It described one of the documents as "perhaps the most redacted report in history," and urged the government to take it "back to your client agencies" for a declaration that might "confirm [its] legitimacy." Hr'g Tr. 47-50 (May 13, 2010) (J.A. 47-50). On the final day of the hearing, the court again urged the government to submit something "to remove th[e] question mark" over the intelligence reports. Hr'g Tr. 46 (May 17, 2010) (J.A. 336).

Thereafter, at the suggestion of Khan's counsel, the government made unredacted versions of those reports available to the court for its in camera review. Id. at 47 (J.A. 337). The government also submitted to both the court and counsel a classified declaration explaining, to the extent possible without disclosing particularly sensitive classified information, how and by whom the reports had been prepared. See Gov't Br. 24. Of particular importance, the declaration stated that [Redaction 6].

On September 3, 2010, the district court denied Khan's habeas petition. Khan v. Obama, 741 F.Supp.2d 1, 12-16 (D.D.C. 2010). The court found that the intelligence collectors' declarations supplied "reason to believe the information provided by [Informant A] is generally accurate." Id. at 13 (internal quotation marks omitted). It noted that the collectors judged the information "sufficiently reliable to plan the operation for Khan's capture based on it." Id. "Intelligence collectors in the field, facing dangerous life-or-death situations," the court observed, "would not... act on the basis of information they felt was unreliable." Id. The court found Informants B and C reliable because they provided information consistent with the information provided by Informant A. Id. at 14.

Based on its in camera review of the redacted portions of the intelligence reports, the unredacted portions of which "describ[ed] the items seized from Khan's properties," the court found that those reports "contain[ed] sufficient indicia of their reliability." Khan, 741 F.Supp.2d at 17. Although it could not discuss the sensitive redacted portions of the reports, the court stated that the declaration the government provided to the court and counsel "summarizes the hallmarks of reliability that the Court finds persuasive" and is itself "accurate and persuasive." Id.

In light of the government's additional evidence, the district court concluded that "the government has met its burden to establish by a preponderance of the evidence that Khan was a `part of' HIG." Id. The court also reaffirmed its earlier conclusion that the government had presented sufficient evidence to establish that "HIG was an `associated force' of al-Qaida and the Taliban at the time of Khan's capture in late 2002." Id. at 8.

On appeal, Khan does not challenge the detention standard applied by the district court, but rather the court's conclusion that the government produced sufficient reliable evidence to satisfy the standard. Khan contends there is insufficient reliable evidence establishing that: (1) he was "part of" HIG at the time of his capture in late 2002; and (2) HIG was an "associated force" of al Qaeda and the Taliban at that time.5 We consider Khan's "part of" arguments in Part II and his "associated force" arguments in Part III.

II

"We review the district court's findings of fact for clear error, its habeas determination de novo, and any challenged evidentiary rulings for abuse of discretion." Al Alwi v. Obama, 653 F.3d 11, 15 (D.C.Cir. 2011) (quoting Al-Bihani, 590 F.3d at 870) (internal citations omitted). Whether a detainee was "part of" an associated force is a mixed question of law and fact. Barhoumi, 609 F.3d at 423. That is, whether a detainee's alleged conduct is sufficient to make him "part of" a force and whether the alleged connections between that force and al Qaeda and/or the Taliban are sufficient to render it an "associated force" are legal questions that we review de novo. See id. Whether the government has proven that conduct and those connections, however, are factual questions that we review for clear error. See id. In addition, "[t]he question whether evidence is sufficiently reliable to credit is one we review for clear error." Al Alwi, 653 F.3d at 19; see Barhoumi, 609 F.3d at 424; Awad v. Obama, 608 F.3d 1, 8 (D.C.Cir.2010). But cf. Al Odah v. United States, 611 F.3d 8, 13-14 (D.C.Cir.2010) (concluding that a district court's decision that certain hearsay evidence was reliable "was no abuse of discretion").

The district court found, based on a preponderance of the evidence, that Khan is lawfully detained under the AUMF. See Al Odah, 611 F.3d at 13 ("It is now well-settled law that a preponderance of the evidence standard is constitutional in considering a habeas petition from an individual detained pursuant to authority granted by the AUMF."). Khan titles his argument regarding the "part of" portion of the AUMF detention standard as a claim that the government's evidence is "insufficient to prove by a preponderance" that he was part of an HIG cell. Pet'r Br. 45. With one exception, however, he does not seriously dispute that the evidence offered by the government — if judged reliable — satisfies the government's burden to show that he was "part of" HIG at the time of his capture. Instead, he primarily challenges the reliability of (A) the Army intelligence collectors' informant reports and subsequent declarations that describe the Kandahar HIG cell and Khan's role in it, and (B) the heavily redacted intelligence reports that describe items seized from Khan's properties. The exception is his contention that, regardless of the government's evidence, the court should instead have accepted the testimony of his expert, which he argues proves there was no HIG cell in Kandahar in 2002. We address these three arguments below.

A

Parhat v. Gates, 532 F.3d 834 (D.C.Cir.2008), "sets the guideposts for our inquiry" into the reliability of the collectors' reports and declarations. Barhoumi, 609 F.3d at 428; see Bensayah v. Obama, 610 F.3d 718, 725-26 (D.C.Cir.2010). In Parhat, the petitioner appealed the determination of a Combatant Status Review Tribunal (CSRT) that he qualified as an enemy combatant because of his affiliation with a Uighur independence group assertedly "associated" with al Qaeda and the Taliban. In particular, the petitioner challenged four intelligence reports on which the Tribunal primarily based its decision. In discounting those reports, we made clear that we were "not suggest[ing] that hearsay evidence is never reliable — only that it must be presented in a form, or with sufficient additional information, that permits the Tribunal and court to assess its reliability." 532 F.3d at 849 (emphasis omitted). Since Parhat, we have repeatedly held that "the fact that the district court generally relied on items of evidence that contained hearsay is of no consequence[;] [t]o show error in the court's reliance on hearsay evidence, the habeas petitioner must establish not that it is hearsay, but that it is unreliable hearsay." Awad, 608 F.3d at 7; see Al Alwi, 653 F.3d at 19; Al Odah, 611 F.3d at 14; Barhoumi, 609 F.3d at 422, 432; Al-Bihani, 590 F.3d at 879.

The government's evidence in Parhat was insufficient to enable the court to assess its reliability. The four intelligence reports at issue described activities "as having `reportedly' occurred, as being `said to' or `reported to' have happened, and as things that `may' be true or are `suspected of' having taken place." 532 F.3d at 846. "But in virtually every instance, the documents d[id] not say who `reported' or `said' or `suspected' those things. Nor [did] they provide any of the underlying reporting upon which the documents' bottom-line assertions [were] founded." Id. at 846-47. Those deficiencies made it impossible to assess the reliability of the reports standing on their own, and no additional evidence supported their reliability. Id. at 848. We therefore set aside the CSRT's enemy combatant determination. To facilitate meaningful review, we ruled, the government must do more than merely "submit[] documents that read as if they were indictments or civil complaints, and that simply assert as facts the elements required to prove that a detainee falls within the definition of enemy combatant." Id. at 850.

At an early stage of the proceedings in this case, the district court determined that the informant reports, standing alone, lacked adequate indicia of reliability. Khan, 646 F.Supp.2d at 16-17. It then properly afforded the government an opportunity to submit "sufficient additional information ... permit[ting the factfinder] to assess [their] reliability," Bensayah, 610 F.3d at 725-26 (quoting Parhat, 532 F.3d at 849). The government responded with the declarations of the Army intelligence collectors who had met with the sources and prepared the reports. The court found that "[t]hese declarations provide the information necessary to assess the sources' reliability," and "based on that information," the court concluded that the reports by the informants were reliable. Khan, 741 F.Supp.2d at 13-14.

Whether or not we would have regarded the informant reports as reliable on their own, it is clear that, together with the declarations, the combination looks nothing like the intelligence reports at issue in Parhat. Unlike in Parhat, we do not have a series of documents containing naked assertions about acts that "may have" occurred or that were "reported to" have taken place, where we do not know the identity of either the documents' authors or their sources. To the contrary, here we know that the authors of the documents are U.S. Army intelligence collectors who were on the ground in Kandahar at the time of the events in question, and we know that their sources are the three Afghan informants — two of whom are identified by name and one by position. Nor are the materials limited to the collectors' "bottom-line assertions"; rather, they contain the "underlying reporting" that was missing in Parhat. Also unlike the reports in Parhat, the collectors' reports and declarations contain great detail about what each source said about the detainee's activities.

In Barhoumi v. Obama, we were able to assess the reliability of a diary that was translated in an intelligence report "by evaluating the diary's internal coherence as well as its consistency with uncontested record evidence, including [the detainee's] own statements and the circumstances of his capture." 609 F.3d at 428. We stressed that the diary exhibited "first-hand knowledge" through "highly detailed descriptions of real-world persons, places, and events." Id. at 428-29, 432. We noted the significance of "independent verification," finding that the diary "refer[red] accurately and in great detail to verifiable real-world events." Id. And we further noted that "[the detainee's] own statements" buttressed the diary's reliability given the "numerous consistencies" between them. Id. at 429, 432.

These and other indicia of reliability are present here. The informants' reports themselves exhibit the informants' "first-hand knowledge" through "highly detailed descriptions of real-world persons, places, and events." Id. at 428-29, 432. The informants gave the intelligence collectors quite specific information about the cell's method of operation, including the type of explosive device the cell employed, the cell's use of two explosions to maximize casualties, the exact radio frequencies it used in a specific sequence to detonate the two explosions, and details about some of the cell's intended targets. In addition, the reports originally contained as enclosures digital photographs of objects provided by the informants to the collectors: an "[HIG] explosive device," IIR 6 044 0249 03 at 4 (J.A. 1729) (Informant A report), and an "[HIG] detonation device," IIR 6 044 0025 03 at 4 (J.A. 1724) (Informant B report).

Moreover, the informants' reports are reinforced by "verifiable, real-world" evidence, Barhoumi, 609 F.3d at 429. In particular, the items seized from Khan's home and shop corroborate the informants' statements by providing evidence of Khan's work for HIG. See supra Part I; infra Part II.B.6 Further support comes from "consistencies" with Khan's "own statements," Barhoumi, 609 F.3d at 429, 432, including his admission to possessing some of those items. See supra Part I. In addition, many of Khan's interrogation statements confirm that he was active in HIG during the jihad against the Soviet Union in the 1980s. Khan, 646 F.Supp.2d at 17; Pet'r Br. 11-12; cf. Salahi v. Obama, 625 F.3d 745, 751 (D.C.Cir.2010) (noting that the detainee's prior association was relevant to the question of whether he was "part of" al Qaeda at the time of his capture).

The collectors' declarations further buttress the informants' reliability. Khan argues that the fact that the collectors vouch for the informants' reliability cannot be sufficient. But the collectors' assessments are not the views of distant officials, based on nothing more than second-hand reports. Rather, the officers are experienced informant handlers who made their judgments based on face-to-face conversations in Kandahar. More important, their declarations explain, in detail, why they assess the informants as reliable.

With respect to Informant A, the collectors note that he spoke voluntarily, spontaneously, and in great detail. They state that he was a member of the HIG cell who came forward because he wanted to leave the cell as a result of an HIG attack that killed some of his tribesmen. Decl. of Intelligence Collector 1 at ¶¶ 37, 39 (J.A. 1548-49); Decl. of Intelligence Collector 2 at ¶¶ 15-16 (J.A. 1569-70). This undercuts Khan's speculation that Informant A lied in order to collect a bounty or avenge a grievance against him. The collectors were also able to independently verify some aspects of Informant A's story. See Barhoumi, 609 F.3d at 428-29 (considering "independent verification" an important indicator of reliability). He provided a detailed account of a previous attack on a U.S. patrol that was corroborated by members of the patrol, and an explosives expert confirmed that a device Informant A gave the collectors was part of a binary detonator that an average civilian would not have possessed. Decl. of Intelligence Collector 1 at ¶¶ 41-44 (J.A. 1549-50); Decl. of Intelligence Collector 2 at ¶ 15 (J.A. 1569-70). The collectors also note that U.S. forces planned their operation to capture Khan based on Informant A's tip that Khan would be at his shop on a certain day and time, and the tip proved accurate. Decl. of Intelligence Collector 1 at ¶ 46 (J.A. 1551); Decl. of Intelligence Collector 2 at ¶ 45 (J.A. 1578); see Parhat, 532 F.3d at 849 (suggesting that the government's use of documents "to take actions of consequence" is an indicator of their reliability). Finally, one of the collectors states that after Khan's capture, IED attacks stopped in the area for about two months. Decl. of Intelligence Collector 2 at ¶ 15 (J.A. 1570).

The collectors also judge Informant B to be reliable. Although he obtained his information from a sub-source in his neighborhood,7 he provided a detailed description of Khan's shop that was used to plan the capture, brought a detonation device to one interview, and was motivated to come forward by Taliban atrocities and a desire to improve business security in Kandahar. Decl. of Intelligence Collector 1 at ¶¶ 21-25, 32 (J.A. 1543-44, 1546-47); Decl. of Intelligence Collector 2 at ¶¶ 18-23 (J.A. 1571-73). Although the intelligence collectors have no independent recollection of Informant C, they judge him reliable as well because he provided highly detailed information that matched the information from the other two sources. Decl. of Intelligence Collector 1 at ¶¶ 50-56 (J.A. 1551-53); Decl. of Intelligence Collector 2 at ¶¶ 26-32 (J.A. 1573-75). While Informant C only repeated information he obtained from a sub-source, the collectors say they do not think the sub-source was either Informant A or Informant B, because if he had been they would have noted that fact in their original report. Decl. of Intelligence Collector 1 at ¶ 52 (J.A. 1552); Decl. of Intelligence Collector 2 at ¶ 29 (J.A. 1574).8

Although Khan attacks the collectors' credibility on a number of fronts, we review a district court's factfinding regarding the credibility of a witness only for clear error, Awad, 608 F.3d at 8, and we find no such error here. Khan complains that, because the collectors identify themselves with their alphanumeric codes and ranks rather than their actual names, their statements are not subject to the penalty of perjury and their oaths are meaningless. The district court properly rejected this argument, as it finds no support in either case law or the perjury statute, 18 U.S.C. § 1621. There is no serious dispute that the alphanumeric codes refer to specific individuals who could be identified for purposes of a perjury prosecution. Khan also argues that because, as the district court found, the declarations "inaccurately detail the timeline of when [one informant] introduced the collectors to [another informant]," Khan, 741 F.Supp.2d at 16, everything in the declarations should be disregarded as the product of wholesale fabrication. Although the court recognized that the timeline inaccuracies were disconcerting, it concluded that they did "not relate to the collectors' assessments of their sources' reliability," which were "supported by the collectors' independent verification of much of the reported information." Id. The district court's "decision to credit some of the statements of an individual but not others is reviewed for clear error," Awad, 608 F.3d at 8, and we find none here.

In short, the reports sourced to the three informants, supplemented by the Army intelligence collectors' declarations, are "a far cry from the `bare assertions' deemed unreliable in Parhat" because they "possess[] both endogenous and exogenous indicia of reliability." Barhoumi, 609 F.3d at 429. We find no error in the district court's determination that they were reliable.

B

Khan also challenges the reliability of heavily redacted intelligence reports that describe items recovered in searches of his properties. Those reports are highly incriminating, both because of the nature of the items themselves and because their presence on Khan's properties further corroborates the informants' description of Khan's role in the Kandahar HIG cell. Most incriminating of all is the report that [Redaction 7]; see also Redaction 1 (describing the seized items).

As we noted in Part I, at the habeas hearing the district court expressed concern that the extensive redactions made it impossible to determine the source and timing of the reports or to assess their reliability. The court described one of the documents as "perhaps the most redacted report in history," and pressed the government to take it "back to your client agencies" for a declaration that might "confirm [its] legitimacy." Hr'g Tr. 47-50 (May 13, 2010) (J.A. 47-50).

We agree with the district court that it is not possible to identify sources or assess reliability based on the redacted versions of the reports. But like the district court, we are not limited to the redacted versions. As we discussed in Part I, Khan's counsel indicated that he would be satisfied if the government gave the district court the unredacted reports to review in camera. The court seconded this suggestion, and the government obliged. The government also accepted the court's suggestion that it supplement the reports, which it did by providing both the court and Khan's counsel with a declaration (still classified, but with less sensitive information) that describes the reports' sources.

This circuit has previously suggested and endorsed the kind of search for reasonable alternatives that the parties and the court undertook here. In Parhat, for example, we said that where the source of classified information is "highly sensitive,... it can be shown to the court (and CSRT) alone." 532 F.3d at 849; see Bismullah v. Gates, 501 F.3d 178, 180 (D.C.Cir.2007) ("[T]he Government may withhold from counsel, but not from the court, certain highly sensitive information."), vacated on other grounds, Gates v. Bismullah, 554 U.S. 913, 128 S.Ct. 2960, 171 L.Ed.2d 881 (2008). We also observed that "there may well be other forms in which the government can submit information that will permit an appropriate assessment of the information's reliability while protecting the anonymity of a highly sensitive source," noting both the federal courts' practice in the Fourth Amendment context and the authorization in the Classified Information Procedures Act (CIPA) for using nonclassified substitutions in criminal cases. Parhat, 532 F.3d at 849 (citing 18 U.S.C.App. III, § 4). Similarly, in Al Odah v. United States, we again suggested by analogy to CIPA that the government may offer alternatives to providing classified information, as long as they "suffice to provide the detainee with `a meaningful opportunity to demonstrate that he is being held pursuant to the erroneous application or interpretation of relevant law.'" 559 F.3d 539, 547 (D.C.Cir. 2009) (quoting Boumediene, 553 U.S. at 779, 128 S.Ct. 2229).

The government contends that Khan waived any challenge to the reliability of the intelligence reports. We disagree. It is true that Khan's opening appellate brief did not focus on the reports' reliability, but rather on his then-pending post-judgment motion in the district court seeking unrestricted access to them. It is clear, however that Khan's ultimate goal was to "convince this Court that the [district] court erred in concluding that [the reports] have sufficient indicia of reliability." Pet'r Br. 63.

After Khan filed his opening brief, the district court denied his post-judgment motion, and Khan then filed a motion with this court for unredacted access to the intelligence reports. In the event we were to deny this motion, Khan requested that we review the documents in camera to determine whether his concerns had merit. See Pet'r Reply Br. 30. On April 8, 2011, we directed the government to submit the unredacted documents for our in camera review and deferred the motion for access. After examining the documents, we have — by separate order — denied the motion for unredacted access on the ground that the combination of the government's declaration and the in camera submission constitutes an "effectiv[e] substitute for unredacted access" that ensures Khan the "meaningful review of both the cause for detention and the Executive's power to detain" required by Boumediene. Al Odah, 559 F.3d at 547-48 (quoting 553 U.S. at 783, 128 S.Ct. 2229).

Like the district court, our review of the unredacted intelligence reports confirms that they "contain sufficient indicia" of reliability, and that the government's declaration (to which Khan has access) accurately represents the information contained in the reports. Khan, 741 F.Supp.2d at 17. That declaration states [Redaction 8]. It therefore adequately rebuts Khan's contention that [Redaction 9].

C

Khan further contends that, notwithstanding the evidence described above, the district court should instead have credited the testimony of Khan's expert witness that there was no HIG cell in Kandahar in 2002. The expert, Professor Brian Williams of the University of Massachusetts, testified that it is "strange, improbable, [and] unlikely" that HIG would have operated in Kandahar in 2002 because it was a Taliban stronghold and far from the home territory of HIG's founder, Gulbuddin Hekmatyar. Hr'g Tr. 110, 116-18 (May 13, 2010) (J.A. 110, 116-18). Khan argues that this testimony, together with Khan's own statements in interrogation reports, clearly establishes that there was no HIG presence in Kandahar that year.

But the expert's testimony did not remain so clear upon cross-examination. In response to the government's questions, Williams conceded there was a "possibility" that HIG operated in Kandahar in 2002, explaining that his original point was only that the presence of a cell at that time did not "fit the overall pattern that [he had] become used to." Id. at 121, 123 (J.A. 121, 123). Williams also conceded that he did not have access to the latest intelligence, had not traveled to Kandahar in 2002, and was not familiar with Zabit Jalil, Khan's uncle and the man whom the informants and a Defense Department intelligence report said commanded HIG's Kandahar cell from Pakistan. Id. at 123, 128 (J.A. 123, 128); see infra Part III. The district court did not clearly err in finding that the government's evidence, including all that we have discussed above, outweighs Professor Williams' testimony and Khan's statements.9

III

Finally, Khan challenges the district court's finding that HIG was an associated force of al Qaeda and the Taliban in November 2002. To meet its burden, the government introduced a classified expert declaration by [Redaction 10]. In his detailed declaration, the expert stated that [Redaction 11]. He also stated that HIG "played an important and deliberate role in supporting continued attacks against coalition and Afghan forces throughout 2002." Khan, 646 F.Supp.2d at 19 (quoting expert declaration).

In further support, the government proffered a Defense Department intelligence report indicating that, in August 2002, the Taliban and HIG opened a joint office in Pakistan to be led by Khan's uncle — HIG commander Zabit Jalil — and Taliban commander Hafez Majid. IIR 6 440 0230 02 (Sept. 3, 2002) (J.A. 1709-10). The purpose of the joint effort was to recruit new members and raise money for attacks on Afghan and U.S. security forces. Id. The government also introduced two public documents to similar effect. See U.S. DEP'T OF HOMELAND SEC., TERRORIST ORGANIZATION REFERENCE GUIDE (2004) ("HIG has long-established ties with Bin Ladin.... HIG has staged small attacks in its attempt to force U.S. troops to withdraw from Afghanistan, overthrow the Afghan Transitional Administration (ATA), and establish a fundamentalist state.") (J.A. 2167); CONGRESSIONAL RESEARCH SERVICE, AFGHANISTAN: POST-WAR GOVERNANCE, SECURITY, AND U.S. POLICY 23 (2004) (noting that HIG "is allied with Al Qaeda and Taliban remnants") (J.A. 2159).

To dispute the government's evidence, Khan offered the testimony of his expert, Professor Williams. But that testimony did not truly rebut the government's position. Professor Williams explained that, although al Qaeda and the Taliban were enemies of HIG in the late 1990s, the launch of U.S. operations in Afghanistan shortly after September 11 "change[d] the game." Hr'g. Tr. 106 (May 13, 2010) (J.A. 106). HIG's leader, Hekmatyar, returned to Afghanistan in early 2002 and "beg[an] launching low-level insurgent attacks ... in the northeast," at which time there was a burying "of the hatchet between the Taliban and Hekmatyar." Id. at 107. Williams acknowledged that there was a "collaboration" between the groups, and while he did not "know the specifics exactly of when" it began, he said it would have been an "urgent matter" to Hekmatyar "in spring of 2002." Id. at 107, 125-26. Although Professor Williams also testified that the collaboration "begins to get traction" in 2003 and 2004, that is not inconsistent with the conclusion that the collaboration existed in 2002. Id. at 107. In light of this evidence, the district court did not clearly err in finding that HIG was associated with al Qaeda and the Taliban in late 2002.10

IV

For the foregoing reasons, we reject Khan's challenge to the district court's findings that he was "part of" HIG and that HIG was an "associated force" of al Qaeda and the Taliban. We therefore affirm the court's determination that Khan is lawfully detained pursuant to the AUMF and its denial of Khan's petition for a writ of habeas corpus.

So ordered.