OPINION

CHIN, J. —

We granted review to determine whether an environmental impact report (EIR), issued as part of a master plan to develop a partial retirement community in Fresno, California, violates the California Environmental Quality Act (CEQA) for failing to include sufficient information on topics the act requires. (Pub. Resources Code, § 21000 et seq.)1 Our task is to review specific challenges to the final EIR2 that defendant County of Fresno (County) and its board of supervisors adopted, and the trial court approved. As we explain, we affirm in part and reverse in part the Court of Appeal's judgment.

FACTUAL AND PROCEDURAL BACKGROUND

The Friant Ranch project (Project) consists of real party in interest Friant Ranch, L.P.'s (real party in interest) planned development of the Central Valley's first master-planned "pedestrian friendly" community on a 942-acre site (formerly zoned agricultural) that sits adjacent to the unincorporated community of Friant in north central Fresno County, just south of the San Joaquin River. The County is the local governmental entity that acted as the lead agency for the CEQA review and for preparation of the Project's EIR.

The Project includes the Friant Ranch specific plan (Specific Plan), which contemplates the construction of approximately 2,500 single- and multifamily residential units that are age restricted to "active adults" age 55 and older, other residential units that are not age restricted, a commercial village center, a recreation center, trails, open space, a neighborhood electric vehicle network, and parks and parkways. The Project also includes 250,000 square feet of commercial space on 482 acres and the dedication of 460 acres to open space. An additional Friant community plan update expands the Specific Plan area and adds policies that are consistent with the Specific Plan and the County's general plan. The Project's construction is divided into five phases with an estimated 10-year build-out.

Through its board of supervisors, the County received written comments to the draft EIR, held a public hearing, and prepared responses to the comments. After making the findings required under section 21081, subdivision (a), for each significant effect noted in the draft, the County issued a statement of overriding considerations (Statement) that is required in CEQA approved projects to show that the Project's significant environmental effects have been identified, and avoided or mitigated, or that unmitigated effects will be outweighed by the Project's benefits. (Pub. Resources Code, §§ 21002, 21002.1, 21081; Guidelines, §§ 15091-15093.) The Statement noted: "The Project implements and furthers important plans and public policies adopted and endorsed by the County related to urban growth." The Statement also observed that the County "made a reasonable and good faith effort to eliminate or substantially mitigate the environmental impacts resulting from the Project by requiring implementation of the environmentally superior alternative — Project Alternative No. 3: Northeast Development Configuration and the Beck Property alternative wastewater treatment plant location — and various mitigation measures, goals and policies identified in the EIR, General Plan, the proposed Friant Community Plan Update, and the proposed Friant Ranch Specific Plan."

On February 1, 2011, the County's board of supervisors approved Project Alternative 3, certified the EIR, and approved a version of the Specific Plan that prohibited the discharge of treated effluent into the river from the wastewater treatment plant. The County also adopted a mitigation monitoring program (MMP), which noted in part that compliance with the mitigation measures would be "enforced through subsequent conditions of approval for future discretionary actions," including use permits and tentative subdivision maps for the Specific Plan area. By petition for writ of mandamus filed in the trial court, plaintiffs Sierra Club, Revive the San Joaquin, and League of Women Voters of Fresno (collectively, plaintiffs) challenged the County's certification of the EIR, alleging that it violated CEQA in several respects. (Code Civ. Proc., § 1094.5 [challenge to public agency's determination based on alleged CEQA noncompliance requires administrative mandamus proceeding].) The trial court rejected plaintiffs' challenges and approved the Project, noting in its judgment that in reviewing CEQA decisions, "it may not exercise its independent judgment on the evidence, but must determine only whether the act or decision is supported by substantial evidence." In reviewing the EIR, the court agreed with County's findings on traffic impact, biological resources, wastewater treatment, and air quality impact, among other considerations. It stated that the court "does not pass on the correctness of any EIR's environmental conclusions, but instead determines whether the EIR is sufficient as an informational document. All conflicts in the evidence and reasonable inferences must be resolved and drawn in favor of the agency's decisions and findings. The reviewing Court does not reweigh the evidence."

The court's judgment also observed that regarding air quality impacts, the County explained why the EIR's mitigation measures would reduce the Project's greenhouse gas emissions. The court agreed with the County that plaintiffs did not cite to the record in sufficient detail to show any error.

At the end of its judgment, the court noted that it retained jurisdiction to allow the County a reasonable amount of time to circulate a Park Impact analysis on the Project's effect on adjoining parks, including Lost Lake Park and Millerton Lake. This analysis is not at issue here. Otherwise, the court denied all of plaintiffs' claims and entered judgment in favor of real party in interest.

Plaintiffs appealed the judgment before the County could implement the mitigation measures. They claimed in relevant part that the Project's EIR failed to comply with CEQA because its discussion of air quality impacts was inadequate.

The Court of Appeal agreed with plaintiffs' contentions involving the EIR's consideration of the Project's air quality impacts on the following grounds: "(1) the EIR was inadequate because it failed to include an analysis that correlated the [P]roject's emission of air pollutants to its impact on human health; (2) the mitigation measures for the [P]roject's long-term air quality impacts violate CEQA because they are vague, unenforceable and lack specific performance criteria; and (3) the statement that the air quality mitigation provisions will substantially reduce air quality impacts is unexplained and unsupported. These defects must be cured by the preparation of a revised EIR." The Court of Appeal reversed the trial court's judgment on those grounds only.3

We granted real party in interest's petition for review on the issues concerning the Court of Appeal's reversal of the trial court judgment upholding the air quality impact findings and conclusions in the EIR's chapter 3 (discussing air quality impacts). The scope of our review concerns how courts should determine the adequacy of an EIR's discussion, including: What standard of review a court must apply when adjudicating a challenge to the adequacy of an EIR's discussion of adverse environmental impacts and mitigation measures, and whether CEQA requires an EIR to connect a project's air quality impacts to specific health consequences. We must also decide whether a lead agency impermissibly defers mitigation measures when it retains the discretion to substitute later adopted measures in place of those proposed in the EIR, and whether a lead agency may adopt mitigation measures that do not reduce a project's significant and unavoidable impacts to a less-than-significant level.

We conclude as follows: When reviewing whether a discussion is sufficient to satisfy CEQA, a court must be satisfied that the EIR (1) includes sufficient detail to enable those who did not participate in its preparation to understand and to consider meaningfully the issues the proposed project raises (Laurel Heights Improvement Assn. v. Regents of University of California (1988) 47 Cal.3d 376, 405 [253 Cal.Rptr. 426, 764 P.2d 278] (Laurel Heights I)), and (2) makes a reasonable effort to substantively connect a project's air quality impacts to likely health consequences. As explained below, the EIR in this case failed to do so. The EIR should be revised to relate the expected adverse air quality impacts to likely health consequences or explain in meaningful detail why it is not feasible at the time of drafting to provide such an analysis, so that the public may make informed decisions regarding the costs and benefits of the Project.

We further conclude that a lead agency may leave open the possibility of employing better mitigation efforts consistent with improvements in technology without being deemed to have impermissibly deferred mitigation measures. A lead agency may adopt mitigation measures that do not reduce the project's adverse impacts to less than significant levels, so long as the agency can demonstrate in good faith that the measures will at least be partially effective at mitigating the Project's impacts.

We therefore affirm the Court of Appeal's judgment finding the EIR's analyses of the Project's air quality impacts inadequate. However, we reverse the Court of Appeal's judgment that the EIR improperly deferred mitigation measures by proposing to substitute more effective measures if available in the future, and that the mitigation measures proposed were impermissibly vague and unlikely to reduce adverse health impacts to less than significant levels.

DISCUSSION

A. Adequacy of the EIR's discussion of health impacts of the Project's long-term effects on air quality

Plaintiffs claim that the EIR was insufficient as an informational document because it failed to adequately explain how the air pollutants the Project generated would impact public health. To address that claim, we must first decide what standard of review applies to a challenge to the adequacy of an EIR's discussion of a required topic.

1. Standard of review

(1) "The foremost principle under CEQA is that the Legislature intended the act `to be interpreted in such manner as to afford the fullest possible protection to the environment within the reasonable scope of the statutory language.'" (Laurel Heights I, supra, 47 Cal.3d at p. 390, quoting Friends of Mammoth v. Board of Supervisors (1972) 8 Cal.3d 247, 259 [104 Cal.Rptr. 761, 502 P.2d 1049].) "With narrow exceptions, CEQA requires an EIR whenever a public agency proposes to approve or to carry out a project that may have a significant effect on the environment. [Citations.]" (Laurel Heights I, supra, 47 Cal.3d at pp. 390-391; see Guidelines, § 15002, subd. (f).) The basic purpose of an EIR is to "provide public agencies and the public in general with detailed information about the effect [that] a proposed project is likely to have on the environment; to list ways in which the significant effects of such a project might be minimized; and to indicate alternatives to such a project." (Pub. Resources Code, § 21061; see Guidelines, § 15003, subds. (b)-(e).)4 "Because the EIR must be certified or rejected by public officials, it is a document of accountability. If CEQA is scrupulously followed, the public will know the basis on which its responsible officials either approve or reject environmentally significant action, and the public, being duly informed, can respond accordingly to action with which it disagrees." (Laurel Heights I, supra, at p. 392.) The EIR "protects not only the environment but also informed self-government." (Ibid.)

The standard of review in a CEQA case, as provided in sections 21168.5 and 21005, is abuse of discretion. Section 21168.5 states in part: "In any action or proceeding ... to attack, review, set aside, void or annul a determination, finding, or decision of a public agency on the grounds of noncompliance with this division, the inquiry shall extend only to whether there was a prejudicial abuse of discretion." (See § 21005, subd. (a) [noncompliance with information disclosure requirements may "constitute a prejudicial abuse of discretion"].) Our decisions have thus articulated a procedural issues/factual issues dichotomy. "[A]n agency may abuse its discretion under CEQA either by failing to proceed in the manner CEQA provides or by reaching factual conclusions unsupported by substantial evidence. (§ 21168.5.) Judicial review of these two types of error differs significantly: While we determine de novo whether the agency has employed the correct procedures, `scrupulously enforc[ing] all legislatively mandated CEQA requirements' (Citizens of Goleta Valley v. Board of Supervisors (1990) 52 Cal.3d 553, 564 [276 Cal.Rptr. 410, 801 P.2d 1161]), we accord greater deference to the agency's substantive factual conclusions. In reviewing for substantial evidence, the reviewing court `may not set aside an agency's approval of an EIR on the ground that an opposite conclusion would have been equally or more reasonable,' for, on factual questions, our task `is not to weigh conflicting evidence and determine who has the better argument.' (Laurel Heights I, supra, 47 Cal.3d at p. 393.)" (Vineyard Area Citizens for Responsible Growth, Inc. v. City of Rancho Cordova (2007) 40 Cal.4th 412, 435 [53 Cal.Rptr.3d 821, 150 P.3d 709] (Vineyard).)

This distinction between de novo review and substantial evidence review has worked well in judicial review of agency determinations. In most cases, the question whether an agency has followed proper procedures will have a clear answer. Did the agency provide sufficient notice and opportunity to comment on a draft EIR? (Pub. Resources Code, § 21092; Guidelines, § 15087.) Did the agency omit the required discussion of alternatives? (Guidelines, § 15126.6.) As to these legal requirements, the agency has no discretion, and courts will invalidate an EIR that fails to meet them. In that sense, judicial review is de novo.

But the question whether an agency has followed proper procedures is not always so clear. This is especially so when the issue is whether an EIR's discussion of environmental impacts is adequate, that is, whether the discussion sufficiently performs the function of facilitating "informed agency decisionmaking and informed public participation." (California Native Plant Society v. City of Santa Cruz (2009) 177 Cal.App.4th 957, 988 [99 Cal.Rptr.3d 572] [relying on Laurel Heights I, supra, 47 Cal.3d at pp. 404-405].) The review of such claims does not fit neatly within the procedural/factual paradigm.

This court's decision in Laurel Heights I illustrates how a court should assess a claim of inadequate discussion. The case involved a challenge to an EIR's discussion of alternatives to the proposed construction of the University of California, San Francisco's (UCSF) Laurel Heights campus. This court concluded that the discussion was inadequate: "UCSF's treatment of alternatives was cursory at best. The draft EIR identified three types of alternatives: no project anywhere, alternative sites on the UCSF Parnassus campus, and alternative sites off-campus. The three categories received a scant one and one-half pages of text in an EIR of more than 250 pages. The EIR stated the obvious conclusion that the `no project' alternative, i.e., no relocation to Laurel Heights, would not have the environmental effects identified in the EIR. It then stated in a mere two-sentence paragraph that `... no alternative sites on [the Parnassus] campus were evaluated as possible candidates for the location of the basic science units of the School of Pharmacy.' This is not a sufficient discussion of on-campus alternatives; it is merely an admission that such alternatives were not considered." (Laurel Heights I, supra, 47 Cal.3d at p. 403.)

Laurel Heights I continued: "Even if the Regents are correct in their conclusion that there are no feasible alternatives to the Laurel Heights site, the EIR is nonetheless defective under CEQA. As we stated in a context similar to CEQA, there must be a disclosure of the `analytic route the ... agency traveled from evidence to action.' (Topanga Assn. for a Scenic Community v. County of Los Angeles [(1974)] 11 Cal.3d 506, 515 [113 Cal.Rptr. 836, 522 P.2d 12] [construing requirements of Gov. Code, § 65906 for zoning variances]; [citation].) The EIR prepared by UCSF contains no analysis of any alternative locations. An EIR's discussion of alternatives must contain analysis sufficient to allow informed decision making. (San Bernardino Valley Audubon Society, Inc. v. County of San Bernardino (1984) 155 Cal.App.3d 738, 751 [202 Cal.Rptr. 423].)" (Laurel Heights I, supra, 47 Cal.3d at p. 404.)

In Laurel Heights I this court was clear that its inquiry was not a matter of reviewing the record for substantial evidence: "The Regents also contend the [project opponents] failed to point to any evidence in the record that demonstrates reasonable alternatives to moving the School of Pharmacy research units to Laurel Heights. This argument is somewhat disingenuous given the Regents' own failure to provide any meaningful information regarding alternatives. It is the project proponent's responsibility to provide an adequate discussion of alternatives. (Guidelines, § 15126, subd. (d).) That responsibility is not dependent in the first instance on a showing by the public that there are feasible alternatives. If the project proponent concludes there are no feasible alternatives, it must explain in meaningful detail in the EIR the basis for that conclusion." (Laurel Heights I, supra, 47 Cal.3d at p. 405.)

(2) Recently, in Cleveland National Forest Foundation v. San Diego Assn. of Governments (2017) 3 Cal.5th 497, 514-515 [220 Cal.Rptr.3d 294, 397 P.3d 989] (Cleveland National Forest), this court made a similar point that the adequacy of an EIR's discussion of environmental impacts is an issue distinct from the extent to which the agency is correct in its determination whether the impacts are significant. "[A]n EIR's designation of a particular adverse environmental effect as `significant' does not excuse the EIR's failure to reasonably describe the nature and magnitude of the adverse effect. (See Berkeley Keep Jets Over the Bay Com. v. Board of Port Cmrs. (2001) 91 Cal.App.4th 1344, 1371 [111 Cal.Rptr.2d 598] [`The EIR's approach of simply labeling the effect "significant" without accompanying analysis of the project's impact on the health of the Airport's employees and nearby residents is inadequate to meet the environmental assessment requirements of CEQA.']; Galante Vineyards v. Monterey Peninsula Water Management Dist. (1997) 60 Cal.App.4th 1109, 1123 [71 Cal.Rptr.2d 1].) An adequate description of adverse environmental effects is necessary to inform the critical discussion of mitigation measures and project alternatives at the core of the EIR. (See Guidelines, § 15151 [`An EIR should be prepared with a sufficient degree of analysis to provide decisionmakers with information which enables them to make a decision which intelligently takes account of environmental consequences.'].)" (Ibid.)

However, there are instances where the agency's discussion of significant project impacts may implicate a factual question that makes substantial evidence review appropriate. For example, a decision to use a particular methodology and reject another is amenable to substantial evidence review, as Sierra Club concedes. But whether a description of an environmental impact is insufficient because it lacks analysis or omits the magnitude of the impact is not a substantial evidence question. A conclusory discussion of an environmental impact that an EIR deems significant can be determined by a court to be inadequate as an informational document without reference to substantial evidence.

Our Courts of Appeal have consistently recognized that adequacy of discussion claims are not typically amenable to substantial evidence review. As the court explained in County of Amador v. El Dorado County Water Agency (1999) 76 Cal.App.4th 931, 945-946 [91 Cal.Rptr.2d 66], "Substantial evidence challenges are resolved much as substantial evidence claims in any other setting: a reviewing court will resolve reasonable doubts in favor of the administrative decision, and will not set aside an agency's determination on the ground that the opposite conclusion would have been equally or more reasonable. [Citations.] [¶] A claim that an agency failed to act in a manner required by law presents other considerations. Noncompliance with substantive requirements of CEQA or noncompliance with information disclosure provisions `which precludes relevant information from being presented to the public agency ... may constitute prejudicial abuse of discretion within the meaning of Sections 21168 and 21168.5, regardless of whether a different outcome would have resulted if the public agency had complied with those provisions.' (§ 21005, subd. (a).) ... [W]hen an agency fails to proceed [as CEQA requires], harmless error analysis is inapplicable. The failure to comply with the law subverts the purposes of CEQA if it omits material necessary to informed decisionmaking and informed public participation. Case law is clear that, in such cases, the error is prejudicial. (Sierra Club v. State Bd. of Forestry (1994) 7 Cal.4th 1215, 1236-1237 [32 Cal.Rptr.2d 19, 876 P.2d 505]; Fall River Wild Trout Foundation v. County of Shasta (1999) 70 Cal.App.4th 482, 491-493 [82 Cal.Rptr.2d 705]; Kings County Farm Bureau v. City of Hanford (1990) 221 Cal.App.3d 692, 712 [270 Cal.Rptr. 650]; East Peninsula Ed. Council, Inc. v. Palos Verdes Peninsula Unified School Dist. (1989) 210 Cal.App.3d 155, 174 [258 Cal.Rptr. 147] (East Peninsula); Rural Landowners Assn. v. City Council (1983) 143 Cal.App.3d 1013, 1021-1023 [192 Cal.Rptr. 325].)" (Italics added.) The court in that case concluded that the EIR was insufficient because among other things it failed to adequately describe environmental baseline conditions. (County of Amador, at pp. 952-956.)

We also affirm that in reviewing an EIR's discussion, we do not require technical perfection or scientific certainty: "`"[T]he courts have looked not for an exhaustive analysis but for adequacy, completeness and a good-faith effort at full disclosure."'" (California Native Plant Society v. City of Santa Cruz, supra, 177 Cal.App.4th at p. 979; accord, Laurel Heights I, supra, 47 Cal.3d at p. 406; see Guidelines, § 15151 ["An evaluation of the environmental effects of a proposed project need not be exhaustive, but the sufficiency of an EIR is to be reviewed in the light of what is reasonably feasible."].)

Three basic principles emerge from our decisions and those of the Court of Appeal: (1) An agency has considerable discretion to decide the manner of the discussion of potentially significant effects in an EIR. (2) However, a reviewing court must determine whether the discussion of a potentially significant effect is sufficient or insufficient, i.e., whether the EIR comports with its intended function of including "`"detail sufficient to enable those who did not participate in its preparation to understand and to consider meaningfully the issues raised by the proposed project."'" (Bakersfield Citizens for Local Control v. City of Bakersfield (2004) 124 Cal.App.4th 1184, 1197 [22 Cal.Rptr.3d 203] (Bakersfield).) (3) The determination whether a discussion is sufficient is not solely a matter of discerning whether there is substantial evidence to support the agency's factual conclusions.

The ultimate inquiry, as case law and the CEQA guidelines make clear, is whether the EIR includes enough detail "to enable those who did not participate in its preparation to understand and to consider meaningfully the issues raised by the proposed project." (Laurel Heights I, supra, 47 Cal.3d at p. 405; see Berkeley Keep Jets Over the Bay Com. v. Board of Port Cmrs., supra, 91 Cal.App.4th at p. 1356 ["Whether an EIR will be found in compliance with CEQA involves an evaluation of whether the discussion of environmental impacts reasonably sets forth sufficient information to foster informed public participation and to enable the decision makers to consider the environmental factors necessary to make a reasoned decision."]; Guidelines, § 15151 ["An EIR should be prepared with a sufficient degree of analysis to provide decisionmakers with information which enables them to make a decision which intelligently takes account of environmental consequences."].) The inquiry presents a mixed question of law and fact. As such, it is generally subject to independent review. However, underlying factual determinations — including, for example, an agency's decision as to which methodologies to employ for analyzing an environmental effect — may warrant deference. (Cf. Mountain Air Enterprises, LLC v. Sundowner Towers, LLC (2017) 3 Cal.5th 744, 751 [220 Cal.Rptr.3d 650, 398 P.3d 556]; Crocker National Bank v. City and County of San Francisco (1989) 49 Cal.3d 881, 888 [264 Cal.Rptr. 139, 782 P.2d 278].) Thus, to the extent a mixed question requires a determination whether statutory criteria were satisfied, de novo review is appropriate; but to the extent factual questions predominate, a more deferential standard is warranted. (Connerly v. State Personnel Bd. (2006) 37 Cal.4th 1169, 1175 [39 Cal.Rptr.3d 788, 129 P.3d 1].)

Real party in interest draws a distinction for standard of review purposes between claims that a required discussion has been omitted altogether and claims that a required discussion is insufficient, with the former subject to de novo review and the latter subject to substantial evidence review. But such a distinction is neither consistent with our precedent (see Laurel Heights I, supra, 47 Cal.3d at pp. 403-405) nor logically defensible. Whether or not the alleged inadequacy is the complete omission of a required discussion or a patently inadequate one-paragraph discussion devoid of analysis, the reviewing court must decide whether the EIR serves its purpose as an informational document.

2. The EIR's air quality discussion

The Court of Appeal's opinion presents a concise summary of the EIR's discussion regarding the Project's air quality impacts on public health.

"The EIR's discussion of Impact No.3.3.2, the long-term area and operational emissions, estimated that, at build-out, the proposed Friant Community Plan would emit approximately 117.38 tons per year of PM10 [particulate matter 10 microns in diameter or smaller], 109.52 tons per year of reactive organic gases (ROG), and 102.19 tons per year of nitrogen oxides (NOx). Estimates were made for ROG and NOx because they are precursors to ozone, which is formed when ROG and NOx undergo chemical reactions in the presence of sunlight.

"The Air District's thresholds of significance are 15, 10 and 10 tons per year for PM10, ROG and NOx, respectively. Because the project's estimated emission of PM10, ROG and NOx were from seven to 10 times larger than that of the thresholds of significance, the EIR concluded these air pollutants would have a significant adverse effect on air quality. Because Mitigation Measure 3.3.2 could not reduce these emissions below the Air District's thresholds of significance, the EIR concluded that the significant impacts were unavoidable.

"The draft EIR included a page of background information about ozone and nearly a page of background information about PM10. Each included a paragraph about the adverse health effects associated with the pollutant. The discussion of the adverse health effects, however, was not connected to the levels of the pollutant that would be emitted by the completed project. Instead, the discussion of adverse health effects was general in nature. For example, the description of the health effects of ozone noted that the effects were primarily to the respiratory system and stated: `Exposure to ambient levels of ozone ranging from 0.10 to 0.40 [parts per million] for 1 to 2 hours has been found to significantly alter lung functions by increasing respiratory rates and pulmonary resistance, decreasing tidal volumes, and impairing respiratory mechanics.'

"As to PM10, the EIR stated its adverse health effects depended upon `the specific composition of the particulate matter.' The EIR, however, provided no information about the composition of the particulate matter that was expected to be produced by the Project."

3. Adequacy of the EIR's discussion of public health impacts from air pollutants that the Project is expected to generate

Real party in interest contends that the EIR satisfied all CEQA requirements because it analyzed the Project's air quality impacts and disclosed the Project's likely general health impacts. Plaintiffs argue that the EIR failed to satisfy Guidelines section 15126.2, subdivision (a), which requires an EIR to "analyze any significant environmental effects the project might cause by bringing development and people into the area affected." In other words, they argue, the Project's health effects must be "clearly identified" and the discussion must include "relevant specifics" about the environmental changes attributable to the Project and their associated health outcomes.

The Court of Appeal held that the EIR's analysis of air quality impacts was inadequate because it did not connect the raw particulate numbers and their effect on air quality with specific adverse effects on human health in the built environment. (See Bakersfield, supra, 124 Cal.App.4th at p. 1193.) Bakersfield considered EIRs relating to the construction and operation of two shopping centers in the City of Bakersfield. (Ibid.) The shopping centers featured a Wal-Mart Supercenter as their primary tenant and anchor. (Id. at p. 1194.) Both EIRs concluded that the projects would have "significant and unavoidable adverse impacts on air quality." (Id. at p. 1219.) But neither EIR specifically identified the health impacts that would result from the adverse air quality effects. The appellate court criticized the EIRs because they lacked an "acknowledgement or analysis of the well-known connection between reduction in air quality and increases in specific respiratory conditions and illnesses. After reading the EIR's, the public would have no idea of the health consequences that result when more pollutants are added to a nonattainment basin." (Id. at p. 1220.) Bakersfield concluded that brief references to adverse health impacts on human respiratory health rendered the EIRs in that case inadequate as a matter of law because they failed to connect the adverse air impact with negative health effects. (Ibid.) The court held that "the health impacts resulting from the adverse air quality impacts must be identified and analyzed in the new EIR's." (Ibid., italics added.)

The Court of Appeal acknowledged that the EIR at issue here went "much further than" the Bakersfield EIRs, noting that the EIR not only listed the type and tons per year of the pollutants the Project is expected to produce, but also provided a general description of each pollutant and how it affects human health. The Court of Appeal found, however, that the EIR was inadequate under CEQA because its analysis failed to correlate the increase in emissions that the Project would generate to the adverse impacts on human health.

Real party in interest had argued below that "the reader can infer from the provided information that the Project will make air quality and human health worse." But the Court of Appeal concluded that "although the better/worse dichotomy is a useful starting point for analyzing adverse environmental impacts, including those to human health, more information is needed to understand that adverse impact."

The EIR does include some discussion of the health impacts of various pollutants and attempts to provide an explanation for its lack of specificity. It offers a general discussion of adverse health effects associated with certain Project-related pollutants. Notably, it also recognized that Fresno County suffers from the "most severe" ozone problems in the state and acknowledged the relationship between adverse ambient air quality and certain health risks to the respiratory system that could affect asthmatics, children, and healthy adults. These adverse effects, the draft EIR observed, could include "breathing and respiratory symptoms, aggravation of existing respiratory and cardiovascular diseases, alterations to the immune system, carcinogenesis, and premature death." The EIR explained, however, that a more detailed analysis of health impacts is not possible at this early planning phase. According to the EIR, "Health Risk Assessments are typically prepared for inclusion in development specific project EIRs when certain types of development commonly known to have the potential to result in a human health risk are being proposed (automobile fueling stations [for example]). Due to the broad nature of the planning approvals analyzed in this EIR, it is not possible to conduct a human health risk assessment based on specific proposed uses at specific locations within the boundaries of the Project Area because such specific information has not been determined."

(3) We agree with the Court of Appeal that the EIR's discussion of health impacts found in impact No. 3.3.2 is inadequate as an informational document, similar to what the court found in Bakersfield, supra, 124 Cal.App.4th at page 1220. The EIR's discussion of health impacts of the named pollutants provides only a general description of symptoms that are associated with exposure to the ozone, particulate matter (PM), carbon monoxide (CO), and nitrogen dioxide (NOx), and the discussion of health impacts regarding each type of pollutant is at most a few sentences of general information. The disclosures of the health effects related to PM, CO, and sulfur dioxide fail to indicate the concentrations at which such pollutants would trigger the identified symptoms. As in Bakersfield, "[a]fter reading the EIR's, the public would have no idea of the health consequences that result when more pollutants are added to a nonattainment basin." (Bakersfield, supra, 124 Cal.App.4th at p. 1220.) And as mentioned above, a sufficient discussion of significant impacts requires not merely a determination of whether an impact is significant, but some effort to explain the nature and magnitude of the impact. (See Cleveland National Forest, supra, 3 Cal.5th at pp. 514-515.) The EIR in this case fails to meet the standards articulated in Bakersfield and Cleveland National Forest.

Even in the one area in which the EIR goes into some detail about health effects — ozone — the analysis is inadequate. The EIR states: "Exposure to ambient levels of ozone ranging from 0.10 to 0.40 [parts per million of ozone] has been found to significantly alter lung functions by increasing respiratory rates and pulmonary resistance, decreasing tidal volumes, and impairing respiratory mechanics. Ambient levels of ozone above 0.12 [parts per million] are linked to symptomatic responses that include such symptoms as throat dryness, chest tightness, headache, and nausea."

At first glance, this information appears to potentially illuminate the health impacts of ozone produced by the Project. But the EIR presents no evidence of the anticipated parts per million (ppm) of ozone as a result of the Project. Rather, the EIR provides the estimated tons per year of reactive organic material (ROG) and NOx, the two components that react with sunlight to form ozone (i.e., ROG + NOx + sunlight → ozone). The raw numbers estimating the tons per year of ROG and NOx from the Project do not give any information to the reader about how much ozone is estimated to be produced as a result. Therefore, the disclosure of the health impacts associated with exposure to 0.10 to 0.40 ppm of ozone is not meaningful within the context of the Project because the reader has no idea how much ozone will be produced (i.e., whether the amount of ozone resulting from the ROG and NOx pollution will bring the ozone ppm within the 0.10 to 0.40 range).

Guidelines section 15126.2, subdivision (a) is instructive. It mandates that an EIR "identify and focus on the significant environmental effects of the proposed project ... examin[ing] ... changes in the existing physical conditions in the affected area," that it identify and describe "[d]irect and indirect significant effects of the project on the environment," and that the discussion should include, among other things, "relevant specifics of ... health and safety problems caused by the physical changes." (Guidelines, § 15126.2, subd. (a).) It also suggests that a connection be drawn between the two segments of information presented in the EIR — potential project emissions and human health impacts. Such a connection would meet CEQA's requirements.

(4) Relying on various amici curiae briefs submitted to the court, the County and real party in interest attempt to explain why the connection between emissions and human health that plaintiffs seek cannot be provided given the state of environmental science modeling in use at this time. The parties may be correct; we express no view on the question, except to note that scientific certainty is not the standard. But if it is not scientifically possible to do more than has already been done to connect air quality effects with potential human health impacts, the EIR itself must explain why, in a manner reasonably calculated to inform the public of the scope of what is and is not yet known about the Project's impacts. Contained in a brief, such explanation is directed at the wrong audience. The relevant informational document here is the EIR, and the EIR must communicate not to the reviewing court, but "the public and the government officials deciding on the project." (Vineyard, supra, 40 Cal.4th at p. 443.) For purposes of supplementing the EIR and bringing it in conformance with CEQA, the information contained in the briefs "is irrelevant ... because the public and decision makers did not have the briefs available at the time the project was reviewed and approved." (Vineyard, at p. 443 ["That a party's briefs to the court may explain or supplement matters that are obscure or incomplete in the EIR ... is irrelevant.... The question is ... not whether the project's significant environmental effects can be clearly explained, but whether they were."].)

We further reject real party in interest's argument that the EIR sufficiently accounted for its lack of specificity by explaining that a "Health Risk Assessment" is typically prepared later in the CEQA process, in connection with development-specific EIRs. A "[Health risk assessment]" is defined in the Health and Safety Code as a type of analysis undertaken in connection with the siting of hazardous substances, "a detailed comprehensive analysis... to evaluate and predict the dispersion of hazardous substances in the environment and the potential for exposure of human populations and to assess and quantify both the individual and populationwide health risks associated with those levels of exposure." (Health & Saf. Code, § 44306.)

(5) CEQA does not mandate such an indepth risk assessment. CEQA requires that the EIR have made a reasonable effort to discuss relevant specifics regarding the connection between two segments of information already contained in the EIR, the general health effects associated with a particular pollutant and the estimated amount of that pollutant the project will likely produce. This discussion will allow the public to make an informed decision, as CEQA requires. Because the EIR as written makes it impossible for the public to translate the bare numbers provided into adverse health impacts or to understand why such translation is not possible at this time (and what limited translation is, in fact, possible), we agree with the Court of Appeal that the EIR's discussion of air quality impacts in this case was inadequate.

The Court of Appeal identified several ways in which the EIR could have framed the analysis so as to adequately inform the public and decision makers of possible adverse health effects. The County could have, for example, identified the Project's impact on the days of nonattainment per year. But the Court of Appeal was clear that, ultimately — though the EIR must provide an analysis that is adequate to inform (Guidelines, § 15151) — the "County has discretion in choosing what type of analysis to provide...." We agree. The task for real party in interest and the County is clear: The EIR must provide an adequate analysis to inform the public how its bare numbers translate to create potential adverse impacts or it must adequately explain what the agency does know and why, given existing scientific constraints, it cannot translate potential health impacts further.

(6) To be sure, "`courts have looked not for perfection but for adequacy, completeness, and a good faith effort at full disclosure.'" (In re Bay-Delta etc. (2008) 43 Cal.4th 1143, 1175 [77 Cal.Rptr.3d 578, 184 P.3d 709].) But basic CEQA principles dictate there must be a reasonable effort to put into a meaningful context the conclusion that the air quality impacts will be significant. Although the EIR generally outlines some of the unhealthy symptoms associated with exposure to various pollutants, it does not give any sense of the nature and magnitude of the "health and safety problems caused by the physical changes" resulting from the Project as required by the CEQA guidelines. (Guidelines, § 15126.2, subd. (a).) Perhaps it was not possible to do more. But even in that case, we would have found the EIR insufficient because it failed to explain why it was not feasible to provide an analysis that connected the air quality effects to human health consequences.

B. Mitigation measures

1. "Substantially reduce air quality impacts"

At the outset of the discussion of proposed mitigation measure No. 3.3.2 (discussed more fully in pt. D. below), the EIR stated that, "Implementation of the following mitigation measures shall substantially reduce air quality impacts related to human activity within the entire Project area but not to a level that is less than significant...."

The Court of Appeal concluded that the EIR's use of the term "substantial" to describe the impact the proposed mitigation measures would have on reducing the Project's significant health effects, without further explanation or factual support, amounted to a "bare conclusion" that did not satisfy CEQA's disclosure requirements.

(7) We agree with the Court of Appeal on this point. (See Laurel Heights I, supra, 47 Cal.3d at p. 404 ["`To facilitate CEQA's informational role, the EIR must contain facts and analysis, not just the agency's bare conclusions or opinions.'"].) Here, the EIR included no facts or analysis to support the inference that the mitigation measures will have a quantifiable "substantial" impact on reducing the adverse effects. The EIR must accurately reflect the net health effect of proposed air quality mitigation measures. (Cleveland National Forest, supra, 3 Cal.5th at p. 514 ["an EIR's designation of a particular adverse environmental effect as `significant' does not excuse the EIR's failure to reasonably describe the nature and magnitude of the adverse effect"].)

2. Deferral of mitigation measures

We next decide whether, as the Court of Appeal concluded, the County, as the lead agency, impermissibly deferred mitigation measures when it approved real party in interest's EIR, which included mitigation measures to "at least partially reduce" the Project's air quality impacts, as well as a substitution clause for future mitigation methods. Plaintiffs contend that the Project's EIR is insufficient, because "the mitigation analysis is devoid of criteria for measuring the effectiveness of mitigation measures." (See Guidelines, § 15126.4, subd. (a)(1)(B); Sundstrom v. County of Mendocino (1988) 202 Cal.App.3d 296, 306-307 [248 Cal.Rptr. 352] [improper to defer formulation of mitigation measures until after project approved].) Plaintiffs agree with the Court of Appeal's conclusion that the formulation of future substitutions in this case was improperly deferred.

(8) The general rule is that an EIR is required to provide the information needed to alert the public and the decision makers of the significant problems a project would create and to discuss currently feasible mitigation measures. Mitigation measures need not include precise quantitative performance standards, but they must be at least partially effective, even if they cannot mitigate significant impacts to less than significant levels. (Laurel Heights I, supra, 47 Cal.3d at p. 404; Pub. Resources Code, §§ 21151, 21100; Guidelines, § 15370.)5

In the present matter, the Project's EIR noted that the air quality impacts will be significant and unavoidable. But the EIR's 12 mitigation measures in mitigation measure No. 3.3.2 were designed to reduce the Project's air quality impacts by providing shade trees, utilizing efficient PremAir or similar model heating, ventilation, and air conditioning (HVAC) systems, building bike lockers and racks, creating bicycle storage spaces in units, and developing transportation related mitigation that will include trail maps and commute alternatives.

Mitigation measure No. 3.3.2 includes a substitution clause that allows the lead agency to "substitute different air pollution control measures for individual projects, that are equally effective or superior to those proposed [in the EIR], as new technology and/or other feasible measures become available [during] build-out within the [Project]." In other words, the County retains the discretion to modify or substitute the adopted mitigation with equally or more effective measures in the future as better technology becomes available, unless the changes increase a project's significant impacts. (See Guidelines, § 15162, subd. (a)(3).)

(9) The County concluded that the Project's air quality impacts will be significant, and that the 12 mitigation measures set forth in the Specific Plan should be at least partially effective in reducing the significant impacts. The substitution clause will allow for additional and presumably better mitigation measures when they become available and it should be encouraged. (See Napa Citizens for Honest Government v. Napa County Bd. of Supervisors (2001) 91 Cal.App.4th 342, 357-358 [110 Cal.Rptr.2d 579] [recognizing county must have power to modify land use plans].) Allowing future substitutions for equal or more efficient technology to mitigate a project's acknowledged significant effects promotes CEQA's goal of environmental protection and is not an impermissible deferral of mitigation or an abuse of discretion. It is simply a recognition that substitutions of adopted mitigation measures may be implemented to further minimize the Project's environmental impacts.

3. Failure to reduce impacts to less than significant levels

Plaintiffs also ask us to decide whether a lead agency violates CEQA when its proposed mitigation measures will not reduce a significant environmental impact to less than significant levels. We conclude that as long as the public is able to identify any adverse health impacts clearly, and the EIR's discussion of those impacts includes relevant specifics about the environmental changes attributable to the project, the inclusion of mitigation measures that partially reduce significant impacts does not violate CEQA.

(10) We have stated that protection of the environment and of California's resources has long been considered of the utmost importance. However, in enacting CEQA to protect the environment, the Legislature did not seek to prevent all development. Section 21081, subdivision (b) allows a project to continue even if there are significant environmental effects that have not been mitigated, if "the public agency finds that specific overriding economic, legal, social, technological, or other benefits of the project outweigh the significant effects on the environment."

If, after the feasible mitigation measures have been implemented, significant effects still exist, a project may still be approved if it is found that the "unmitigated effects are outweighed by the project's benefits." (Laurel Heights I, supra, 47 Cal.3d at p. 391.) Even when a project's benefits outweigh its unmitigated effects, agencies are still required to implement all mitigation measures unless those measures are truly infeasible. (City of San Diego v. Board of Trustees of California State University (2015) 61 Cal.4th 945, 967 [190 Cal.Rptr.3d 319, 352 P.3d 883].) We recently held that "the lead agency must adopt feasible mitigation measures or project alternatives to reduce the effect to insignificance; to the extent significant impacts remain after mitigation, the agency may still approve the project with a statement of overriding considerations. [Citations.]" (Center for Biological Diversity v. Department of Fish & Wildlife (2015) 62 Cal.4th 204, 231 [195 Cal.Rptr.3d 247, 361 P.3d 342].) The inclusion of a mitigation measure that reduces an environmental impact is permitted even if the measure will not reduce the impact to a level below the threshold of significance.

4. Enforceability of mitigation measures

Plaintiffs argue that mitigation measures involving the installation of HVAC systems and tree planting, and any required mitigation efforts that "are fully enforceable through permit conditions, agreements, or other measures," are unenforceable. (§ 21081.6, subd. (b).) We note that the measures referred to in this section are proposed as "guidelines" that "shall be used by the County during review of future project-specific submittals for non-residential development ... with [the] intent that specified measures be required where feasible and appropriate."

The Court of Appeal found the EIR mitigation "provision about equipping HVAC units with a catalyst system does not identify who will determine if the system is `reasonably available and economically feasible'" and is unenforceable. In its analysis, the court omitted the next sentence, "[c]atalyst systems are considered feasible if the additional cost is less than 10% of the base HVAC cost." (Italics omitted.) This definition of what constitutes "economically feasible" catalyst systems eliminates the need to have individuals make such determinations. The Court of Appeal also found the phrase "`PremAir or similar catalyst system'" vague for not defining what performance criteria must be met to be a "`similar catalyst system.'" The term is not vague. PremAir is a brand name for an HVAC catalyst system. The individuals proposing new projects, or those tasked with evaluating the proposals for approval, would necessarily have knowledge of HVAC systems and catalyst systems, including PremAir. It is also impossible to require specific performance criteria, given that the type, size, model, and efficiency levels of the HVAC systems being installed in these future projects are unknown. Given the uncertainty of these future proposed projects, the language "`PremAir or similar catalyst system'" is sufficient under CEQA to provide an enforceable mitigation measure for any HVAC systems associated with those projects.

The Court of Appeal similarly found that the mitigation measure requiring trees be selected to provide shade did not specify the person(s) responsible for selecting which trees to plant. The measure instructs that "`[t]rees selected to shade paved areas should be varieties that will shade 25% of the paved area within 20 years.'" The instruction provides sufficient guidance for selecting appropriate shade trees. Any plan that the County approves must be complete, and must contain information about the trees selected for this mitigation measure. It seems clear that the person(s) selecting the trees would be the individuals or entity submitting the plans to the County for approval. The measure is not vague.

(11) In finding the mitigation measures cannot be enforced through permit conditions, agreements, or other measures, the Court of Appeal also misinterpreted section 21081.6, subdivision (b) and its significant effects provision, which provides that a public agency may set forth conditions of project approval required to avoid significant effects in "referenced documents which address [or incorporate the] required mitigation measures ... into the plan." (§ 21081.6, subd. (b).) The Project's MMP places the burden of enforcement on the County to "ensure that all construction plans and project operations conform to the conditions of the mitigated project." The Specific Plan additionally states that, "The County shall monitor compliance with the Specific Plan and mitigation measures," and it provides the stages of planning at which certain mitigation measures must be completed. These measures are not vague as to how they will be enforced; the County will enforce them during the approval process of future nonresidential development.6 Indeed, if the County were to approve a project that did not include a feasible mitigation measure, such approval would amount to an abuse of discretion, which could be corrected in a court mandamus proceeding. (See Rominger v. County of Colusa (2014) 229 Cal.App.4th 690, 727 [177 Cal.Rptr.3d 677] [holding dust control mitigation measures left to the county's discretion are enforceable through a judicial writ of mandamus]; see also, e.g., California Oak Foundation v. Regents of University of California (2010) 188 Cal.App.4th 227, 247 [115 Cal.Rptr.3d 631].)

CONCLUSION

In our view, the EIR's air quality impacts discussion and its mitigation measures meet CEQA requirements for specificity and enforceability with one exception: The EIR fails to provide an adequate discussion of health and safety problems that will be caused by the rise in various pollutants resulting from the Project's development. At this point, we cannot know whether the required additional analysis will disclose that the Project's effects on air quality are less than significant or unavoidable, or whether that analysis will require reassessment of proposed mitigation measures. Absent an analysis that reasonably informs the public how anticipated air quality effects will adversely affect human health, an EIR may still be sufficient if it adequately explains why it is not scientifically feasible at the time of drafting to provide such an analysis. Otherwise, the EIR is generally clear about the potential environmental harm under the Specific Plan, and it outlined mitigation measures to address those effects with factual support and scientific consensus.

Based on the foregoing analysis, we affirm in part and reverse in part the Court of Appeal's judgment and remand the matter for additional proceedings consistent with this opinion.

Cantil-Sakauye, C. J., Corrigan, J., Liu, J., Cuéllar, J., Kruger, J., and Robie, J.,* concurred.