*102The Orange County Water District (the District) appeals a postjudgment order awarding The Arnold Engineering Company (Arnold) approximately $615,000 in costs of proof under Code of Civil Procedure section 2033.420 based on the District's failure to admit certain fact-specific requests for admission (RFAs) during discovery.1 The District contends the trial court erred in making the award because (1) the District had reasonable grounds to believe it would prevail on the matters at issue under *103*356section 2033.420, subdivision (b)(3), and (2) even if it did not, Arnold did not adequately substantiate its costs with admissible evidence.
We conclude the court abused its discretion in awarding costs for certain RFAs because the District reasonably relied on percipient witness testimony, undisputed scientific testing, and the opinions of a qualified expert in denying the RFAs. The court did not err with respect to other RFAs, which covered matters the District did not pursue at trial. We further conclude that certain evidence, in the form of expert witness invoices, was inadequate to support an award under the circumstances here because it did not distinguish between recoverable and nonrecoverable costs. We will therefore reverse the order and remand for the court to make a new award consistent with this opinion.2
FACTUAL AND PROCEDURAL BACKGROUND
The District's Allegations and Arnold's Discovery Requests
The District is a public entity established by the California Legislature and empowered to manage, replenish, regulate, and protect groundwater supplies within its boundaries. (West's Ann. Wat. Code App. (2016 ed.) ch. 40, §§ 1, 2.) It brought this action to recover expenses associated with the North Basin Groundwater Protection Project (NBGPP), a proposed $200 million effort intended to address groundwater contamination in northern Orange County, California caused by volatile organic compounds (VOCs) and other chemicals. The District named as defendants a number of current and former owners and operators of industrial sites in the project area, including Arnold, that it believed were responsible in some way for VOC contamination in groundwater.
VOCs can be used in industrial solvents and cleaners. The primary VOCs at issue in this litigation include tetrachloroethylene (also known as perchloroethylene or PCE), trichloroethylene (TCE), 1,1-dichloroethylene (1,1-DCE), and 1,1,1-trichloroethane (1,1,1-TCA). The last chemical, 1,1,1-TCA, breaks down over time into 1,1-DCE and acetic acid. The detection of 1,1-DCE in soil or groundwater can therefore be evidence of past 1,1,1-TCA contamination. Another chemical, 1,4-dioxane, was commonly added to 1,1,1-TCA as a stabilizer.
*104Arnold owned and operated an industrial site at 1551 East Orangethorpe Avenue in Fullerton from 1960 through 1984. Arnold admitted using 1,1,1-TCA, but it contended that none of its operations resulted in VOC releases into the environment and none of its operations resulted in VOC contamination of groundwater. It denied using PCE or TCE at all.
After more than six years of litigation, and six months before trial, Arnold served RFAs on the District asking it to admit that (1) Arnold did not release PCE, TCE, 1,1,1-TCA, or 1,4-dioxane during its business operations at the site; (2) Arnold did not release PCE, TCE, 1,1,1-TCA, or 1,4-dioxane that caused contamination of groundwater in the North Basin area; and (3) Arnold did not contribute to VOC contamination in soil or groundwater in the area surrounding its site. The District denied these RFAs.
*357Evidence Revealed in Pretrial Discovery
Documents produced in discovery showed that Arnold applied for and obtained permits to operate a number of vapor degreasers and dip tanks (or "strippers") at its site. VOC solvents can be used in degreasers and strippers. A number of the documents produced in discovery identified 1,1,1-TCA as a VOC solvent used by Arnold, including one that described a 550-gallon above-ground storage tank for that chemical. Other documents did not specify any solvent.3
Arnold also operated one or more clarifiers. Although clarifiers do not themselves make use of VOC solvents, they can be a source of VOC discharge into the environment because they may process VOC-contaminated wastewater generated by other activities.
Donalee Farmer, a former Arnold maintenance manager, testified at deposition that he was aware of Arnold's use of only one chemical, 1,1,1-TCA, in a degreaser. He was not aware of any spills of 1,1,1-TCA at the site. He did not know which chemicals were used in Arnold's stripping operations.
Dan Hopen, another former Arnold employee, testified at deposition that Arnold used PCE as a stripper. He said "stripper solution" spilled onto the floor "all the time." Hopen also testified that Arnold used "trichloroethylene" (TCE) in its degreaser. When he was asked whether he ever saw a barrel with that name on it, Hopen answered, "Yeah, I think because-didn't actually-I don't know if it said trichloroethylene or Tri-111 or something like that. It might have had both on there. I'm not certain."
*105Farmer was aware of only one clarifier used by Arnold, which was inside the building. He testified that Arnold did not use a clarifier at the location identified as the "South Clarifier" in subsequent investigations. A 1986 inspection, however, revealed five different clarifiers at the site, including two outside the building, i.e., the "North Clarifier" and the "South Clarifier."
In the decades after Arnold left, various entities investigated potential VOC contamination at the site. Basic testing began in 1988, about four years after Arnold ceased operation. PCE contamination, along with small amounts of TCE and 1,1,1-TCA, was discovered. In 1995, more testing revealed PCE, TCE, 1,1,1-TCA, and 1,1-DCE contamination in the shallow soil. That year, TCE and 1,1-DCE were also detected in deeper samples, down to 105 feet below ground level, which was the deepest level tested. PCE was detected down to approximately 60 feet below ground level. 1,1,1-TCA was detected only once, at a level of 60 feet below ground level. Groundwater was encountered at 115 feet below the surface, but it was not tested. Further testing in 2007 revealed extensive PCE, TCE, 1,1,1-TCA, and 1,1-DCE contamination in shallow soil at additional locations on the site.
The District retained an expert witness, Richard Waddell, to analyze the extent and impact of VOC contamination at Arnold's former site. Waddell holds a doctoral degree in geology and is a registered geologist with the State of California. He described his experience in his resume as follows: "More than 30 years of post-graduate experience which includes RCRA/CERCLA [4 ] investigations; numerical *358modeling of flow and contaminant fate and transport processes in porous and fractured media; hydrogeologic characterization studies of deep and shallow groundwater systems; reclamation of uranium mill tailings; geochemical studies of inorganic contaminants; conceptual design and analysis of remedial actions; and litigation support." At trial, Arnold did not challenge Waddell's qualifications as an expert.
In his expert report, dated approximately two weeks prior to the District's RFA responses, Waddell summarized Arnold's activities at the site and the contamination found there. He believed, based on the evidence, that Arnold used PCE, TCE, and 1,1,1-TCA in its operations. He opined that Arnold's stripping activities involved a risk of spillage, and he found no evidence of VOC use by any other occupant of the site. For example, in his deposition, Farmer explained that he worked for the prior occupant of the site, Ensign Carburetor, and was not aware of any use of VOCs during their occupancy. A *106subsequent occupant of the site used solvents in its manufacturing processes for picture frames, but there was no evidence it used any of the chemicals at issue in this litigation.
Waddell explained, "The use of PCE, TCE, and 1,1,1-TCA by [Arnold] has contaminated soils beneath the building, beneath the clarifier, and a degreaser. Soils north of the building, including on the neighboring property [Johnson Controls] to the north, were also contaminated." With respect to the Johnson Controls site, Waddell's report noted that Johnson Controls' use of solvents was "limited" and the concentrations of VOC contamination in soil at the Johnson Controls site were lower than concentrations at the Arnold site. Waddell believed VOC releases at the Arnold site had impacted the Johnson Controls site.
To assess the effect of contamination at the Arnold site on groundwater, Waddell analyzed grab samples of groundwater upgradient and downgradient of the site. (There was no groundwater contamination data for the site itself.) A "grab sample" is "a sample that's collected at a particular depth in a one-time event." Waddell found that concentrations of TCE and 1,1-DCE increased as groundwater passed under site. Waddell opined that TCE and 1,1-DCE, as well as 1,4-dioxane, had impacted groundwater as a result of VOC releases by Arnold. However, Waddell wrote it was "not clear" whether PCE had similarly impacted groundwater.
Arnold's expert witness, Jonathan Rohrer, largely disagreed with Waddell's conclusions. Further details regarding his opinions will be discussed below. At his deposition, however, Rohrer admitted PCE "was likely used as part of Arnold's process. ... And there have been impacts to shallow soil as deep as 41 feet, at least in 1988, which would correspond to Arnold or an earlier tenant."
Evidence Admitted at Trial
Several former Arnold employees, including Hopen, testified at trial. Excerpts of Farmer's deposition were entered into evidence. In his testimony, Hopen confirmed Arnold used PCE in its stripper solution. He appeared uncertain about the identities of the other VOCs Arnold used. He first said he believed Arnold used "trichloroethylene" (TCE), but he later said he remembered seeing "1,1,1" on the barrels, which could only refer to 1,1,1-TCA.
*359Hopen then testified he was confident it was TCE.5
*107Hopen contradicted his prior deposition testimony about stripper solution spilling on the floor, but he did confirm that the stripping area was cleaned by spraying the floor with water, which then flowed into a clarifier. The clarifier handled "the solutions of etching material, the stripping material and the cleaning material" and discharged them into the sewer without removing any VOC solvents. VOC solvents can penetrate concrete and leak from sewer pipes to contaminate the surrounding environment. (In his deposition testimony, Farmer confirmed that the water sprayed in the stripper area drained to a clarifier and then to the sewer.) As to the degreasers, Hopen testified that leaking would occur when they were refilled, though he was impeached by his deposition testimony that he did not remember the pumps that were used to refill the degreasers. He asserted, however, his memory was better at trial.
Another former employee, Renee Otero, testified that the floor around the degreaser was never slippery, there were no chemicals dripping on the floor, and there was no mist in the air. She acknowledged, however, that the floor would get wet from the water spray used to clean the chemicals off the degreased metal. She wore boots to protect her feet from the chemicals and the water.
At trial, Waddell reiterated his opinion that Arnold used PCE, TCE, and 1,1,1-TCA in its operations, and its operations resulted in releases of those VOCs into the environment. He based his opinion on evidence of Arnold's operations and the contamination at the site. He said Arnold spilled liquids containing VOC solvents on the floor and washed them into a sump, where they were processed by a clarifier and dumped into the sewer. Nothing in this process removed the VOC solvents themselves from the wastewater. In Waddell's opinion, Arnold's practices were likely to cause spills or releases. Waddell identified two degreaser areas, the stripper tanks, and the Southern Clarifier as the chief sources of VOC contamination. As noted, soils testing revealed PCE, TCE, 1,1,1-TCA, and 1,1-DCE contamination at the site, including under the building slab.
Waddell testified he found no indication that any occupant of the site, other than Arnold, had used or released PCE, TCE, or 1,1,1-TCA. He concluded it was very unlikely that any other occupant was the source of VOC contamination at the site. TCE, in particular, would not have been used by any *108subsequent occupants. Its use was heavily regulated after 1976, when Arnold was still operating the site. After *360regulation, businesses switched to other chemicals such as PCE or 1,1,1-TCA.
Based on soil and groundwater data, Waddell concluded that TCE released by Arnold had reached groundwater and 1,1,1-TCA released by Arnold had impacted groundwater with its degradation product 1,1-DCE. Waddell supported his opinion in part with data from a groundwater monitoring well (MW-2) on the Johnson Controls site, which bordered the Arnold site to the north and west. The groundwater monitoring well (MW-2) was located directly north of the Arnold site, near the property line between the Arnold site and the Johnson Controls site. Waddell believed TCE and 1,1-DCE detected in groundwater there were caused by releases at the Arnold site, because in his opinion Johnson Controls had not used significant amounts of those solvents in its operations and its shallow soil was not as contaminated. Arnold, by contrast, had a chemical storage area adjacent to the Johnson Controls site and the monitoring well.
As noted, Waddell also relied on grab samples to support his opinion that the Arnold site contributed to VOC contamination in groundwater. Waddell was involved in the selection of the locations for the grab samples. Four locations were eventually chosen, one upgradient and three downgradient. The grab sample Waddell treated as his upgradient or baseline reading (GW-1) was located on adjacent property immediately to the east of the Arnold site. The downgradient grab samples were located on adjacent property to the northwest (GW-2), adjacent property to the west (GW-3), and adjacent property to the southwest (GW-4/4A).
Waddell's assessment of these grab samples was informed by his view of the prevailing groundwater flow in the vicinity of the Arnold site. In his testimony, he acknowledged that this flow could vary over time. Waddell identified areas of both northwesterly and southwesterly flows around the Arnold site. But, with respect to the grab samples, he believed the prevailing flow was to the northwest and west.
Comparisons of TCE concentrations between the upgradient location (GW-1) and two downgradient locations (GW-2 and GW-3) showed much higher concentrations at the downgradient locations. TCE concentrations in the third potentially downgradient location (GW-4/4A) were not elevated. Comparisons of 1,1-DCE concentrations between the upgradient location (GW-1) and two downgradient locations (GW-2 and GW-3) showed slightly higher concentrations at the downgradient locations. 1,1-DCE concentrations at the third potentially downgradient location (GW-4/4A) were actually lower.
Based on these data, Waddell opined that the increased TCE and 1,1-DCE concentrations were caused by VOC contamination at the Arnold site, as a *109result of Arnold's activities there. Waddell did not specifically address 1,4-dioxane, and he was precluded from offering the opinion that Arnold contributed to PCE groundwater contamination because he failed to disclose that opinion prior to trial. (As noted, Waddell wrote in his expert report that it was "not clear" whether PCE releases at the Arnold site had impacted groundwater.)
Waddell admitted that data from grab samples are less reliable than data from a groundwater monitoring well. He explained, "The changes in concentrations that you can observe through multiple sampling events provide information on, in essence, whether concentrations are going up or going down. You can compare those against water level measurements. It's just a more complete data set than a-single-event grab sample."
*361On cross-examination, Waddell admitted several errors in his calculations. In his expert report, he omitted a data point from a cross-gradient monitoring well that recorded a high TCE concentration, as well as several upgradient bore holes on a different contaminated property. In his testimony, and in a related demonstrative exhibit, he misstated the extent of TCE contamination in the shallow soil at the neighboring Johnson Controls site. As to the latter error, however, Waddell had used the correct contamination data in his expert report; he was only mistaken in his testimony.6
Waddell was impeached by evidence, contrary to his testimony, that Johnson Controls had operated several degreasers. He was also impeached by evidence that a solvent used by Johnson Controls, Safety-Kleen, may have incorporated VOCs including TCE. While Waddell was aware that Johnson Controls used Safety-Kleen, he was not aware of its specific chemical composition. The District, however, had memoranda in its possession that described the composition of Safety-Kleen, and it had provided discovery responses in other litigation that noted the possibility of VOC contamination from Safety-Kleen. The District did not provide this information to Waddell, and it also did not provide Waddell with deposition transcripts of two Johnson Controls employees describing its practices. In addition, other occupants of the Johnson Controls site may also have used VOCs.
Despite these specific shortcomings, however, Waddell had opined in his expert report that TCE and PCE had been released at the Johnson Controls site. He testified that TCE, in particular, had been found in the shallow soil *110there. He was therefore aware, separate and apart from the composition of Safety-Kleen, that the Johnson Controls site was a potential source of contamination. Waddell believed that, in addition to that contamination, VOC contamination from the Arnold site had migrated to the Johnson Controls site as well. When compared to the Johnson Controls site, the peak concentrations of TCE found in the shallow soil at the Arnold site were much higher.
Arnold's expert Rohrer testified that, in his opinion, the only "documented use" of a VOC by Arnold was 1,1,1-TCA. Given this limitation, and his review of the evidence, he believed Arnold had not released any VOCs into the shallow soil and Arnold's operations had not caused any groundwater contamination.
He acknowledged, however, that PCE, TCE, and some 1,1,1-TCA had been found in the shallow soil at the Arnold site. He said PCE and TCE contamination near the Southern Clarifier was an indication that there was a VOC release into the environment at or near the clarifier, but he did not believe Arnold was responsible. He also said the deeper detections of PCE and TCE may not have been caused by releases at the Arnold site at all; it was possible the detections were caused by contamination flowing from upgradient sources.
Rohrer criticized Waddell's reliance on grab sample data. He agreed with Waddell that the groundwater flows in the vicinity of the Arnold site ranged from west-northwest to west-southwest. Given this variation, he concluded the data were inadequate to determine whether the grab *362sample that showed the highest concentrations of TCE and 1,4-dioxane (GW-2) was truly downgradient from the Arnold site. He explained, "[G]iven the gradient discussion we had earlier, GW-1 at an instantaneous point in time may or may not have been upgradient of any of these in terms of the gradients we talked about being south of west to north of west and how gradients change with time. I do not know that GW-2 is indicative only of [the Arnold site] or any background contamination that may be coming from up in this vicinity." He was therefore unable to determine whether the Arnold site contributed to VOC readings at that grab sample. Rohrer identified several upgradient sites that were major sources of VOC contamination and that he believed may have affected the groundwater in the vicinity of the Arnold site as well.
Rohrer blamed other occupants of the Arnold site for shallow soil contamination, and he noted that several barrels of hazardous waste (including unknown paint-related solvents) were recovered from the site in 1993. But Rohrer admitted there was no "documented evidence" on which to base an opinion that subsequent occupants had used PCE, TCE, or 1,1,1-TCA. He admitted the only evidence of PCE use at the site was Hopen's testimony that *111Arnold used it. Rohrer also admitted there was no evidence that any subsequent occupant had released any VOCs into the shallow soil at the Arnold site.
The Court's Statement of Decision and Judgment
Following trial and briefing by the parties, the court issued a 74-page statement of decision covering the District's statutory claims against Arnold and the other defendants. It found in favor of the defendants, and against the District, on all of these claims. As relevant here, the trial court addressed the causes of contamination at defendants' sites in detail. The court's discussion opened with general comments on the state of the evidence. It provided an overview of its reasons for distrusting and failing to credit the District's evidence, including in particular its expert witness Waddell. The court criticized the District for relying on grab samples to determine the nature and extent of VOC contamination at defendants' sites. It credited testimony that grab samples provide only a snapshot of contamination at one moment in time, they are not reproducible, they are not necessarily indicative of past or future conditions, and they are not accepted by regulatory agencies as the sole method for determining groundwater contamination. The court questioned Waddell's credibility in light of the District's failure to provide relevant information to him and his misstatements in his direct testimony. The court noted that Waddell "assumed an advocate's demeanor" and appeared biased in favor of his opinions.
As to Arnold specifically, the court found that it was not responsible for VOC contamination in groundwater because the District had not shown that Arnold itself had released the VOCs at issue into the environment. As we explained in Alcoa , the court's statement of decision did not reflect the correct causation standard for the District's primary cause of action under the Carpenter-Presley-Tanner Hazardous Substance Account Act (HSAA; Health & Saf. Code, § 25300 et seq. ). (See Alcoa, supra , 12 Cal.App.5th at pp. 306-307, 316, 219 Cal.Rptr.3d 474.) However, this error is immaterial to the issues in this appeal. The RFAs at issue here are primarily factual and do not cover legal liability under the HSAA or any other cause of action. The factual issues covered by the RFAs were also discussed in the court's statement of decision, and we will summarize the court's findings here.
*363As to TCE, the court weighed the conflicting testimony regarding Arnold's use of that chemical and concluded the District had not shown Arnold used TCE at its site. It wrote, "Given the documentary evidence showing that Arnold used only 1,1,1-TCA (not TCE), Mr. Hopen's unreliable testimony regarding Arnold's VOC use, and the multiple operators at the site that may have released contaminants ... both before and after Arnold's occupancy, the *112District has not carried its burden of proving that Arnold used TCE during its operations at the Site. Therefore, Arnold is not responsible for TCE detections in soil or groundwater at the Johnson Controls site or any other part of the Project area." Similarly, for PCE, the court found the District had not proved that Arnold was responsible for contamination in groundwater.
Although Arnold acknowledged using 1,1,1-TCA, the court determined "there is insufficient evidence that Arnold caused a release of 1,1,1-TCA or 1,1-DCE into soil." The court noted that several subsequent occupants of the Arnold site used unknown solvents and paints with unknown constituents. The court explained, "Given the lack of foundation for Dr. Waddell's opinions, the evidence of subsequent occupants of the [Arnold] Site and Mr. Rohrer's substantiated expert opinion, there is insufficient evidence to find that the 1,1,1-TCA or 1,1-DCE found in shallow soil vapor on the [Arnold] Site originated from Arnold. Without a foundation for the opinion that Arnold released 1,1,1-TCA into soil (which would have eventually broken down to some extent to 1,1-DCE), there is no basis for Dr. Waddell's opinion that Arnold's operations contaminated groundwater or threaten today to contaminate groundwater." The court further found that each defendant, including Arnold, was "entitled to a judicial declaration that it has no liability to the District for damages, response costs, or other costs claimed by the District, or any future costs associated with the NBGPP."
The court later granted defendants' motion for judgment on the District's remaining common law claims. It entered judgment in favor of defendants, and against the District, on each of the District's causes of action. The judgment included a declaration that defendants "have no liability to the [District] for damages, response costs, or other costs claimed by the [District], or any future costs."
Arnold's Motion for Costs of Proof
After judgment was entered, Arnold moved for an award of costs of proof against the District based on its refusal to admit RFAs during discovery. It sought approximately $2.5 million in attorneys' fees and $400,000 in expert costs. The District opposed. Among other things, the District argued it should not be subject to a cost award because it had a reasonable ground to believe it would prevail on the issues covered by the RFAs.
After considering the initial briefing, the court requested further briefing regarding certain legal issues and additional documentation regarding Arnold's attorneys' fees and expert costs. In response to the court's comments, Arnold limited its motion to certain RFAs and reduced its cost request. The RFAs at issue asked the District to admit the following:
*113(1) Arnold did not release PCE, TCE, 1,1,1-TCA, or 1,4-dioxane during its business operations at the site;
(2) Arnold did not release PCE, TCE, 1,1,1-TCA, or 1,4-dioxane that caused contamination of groundwater in the North Basin area; and
(3) Arnold did not contribute to VOC contamination in soil or groundwater in the area surrounding its site.
*364In its reduced request, Arnold sought approximately $650,000 in attorneys' fees and $300,000 in expert costs. As proof of its expenses, Arnold submitted invoices from its attorneys and its expert. The attorney invoices included descriptions and a list of applicable RFAs for each time entry. The expert invoices included descriptions for each time entry, but they did not reference any RFAs. The District contended the expert invoices were inadmissible hearsay, and Arnold's showing was inadequate to tie the expert fees to the issues covered by the RFAs.7
After further briefing and argument, the court granted Arnold's narrowed motion for costs in part. As relevant here, it found that the District did not have a reasonable ground to believe it would prevail on the issues covered by the RFAs. ( § 2033.420, subd. (b)(3).) In its written order, the court explained, "Dr. Waddell based his conclusion that Arnold was responsible for TCE contamination on detection of that VOC at a location north of the Arnold site, even though he also testified the prevailing plume in that area consistently moved in a west/southwest direction. Former employee Hopen testified Arnold used TCE, but the only labels he could remember had '1,1,1' on them. That designation is associated not with TCE, but with TCA. While Arnold used 1,1,1-TCA in its operations, there was no substantial evidence that Arnold ever caused a release into the soil, much less groundwater. Dr. Waddell waited until trial to testify there was no evidence of PCE or 1,4-dioxane use by Arnold at its site. [¶] The court has carefully re-reviewed the arguments and admissible evidence proffered by the District on these issues in connection with this motion, but does not find them persuasive. The court is also not persuaded the September 7, 2011 denial of Arnold's motion for summary adjudication of issues gave the District any reasonable grounds for denying the Relevant RFAs one day later or for not taking the issues off the trial table in February 2012.... Even if one were to conclude the truth of the matters at issue in the Relevant RFAs was 'hotly' contested when the *114District denied the Relevant RFAs, that alone is not sufficient to give the responding party reasonable grounds to deny. [Citation.]" The court also found there was no other good cause for the District's failure to admit the RFAs.
The court awarded approximately $313,000 in attorneys' fees, or around half of what Arnold had requested. The court found that the descriptions for certain time entries were too general, or were too heavily redacted, to understand the work allegedly performed. It also found that the entries reflected some duplication of work, some work that was more clerical than legal, and some work that was not directly related to proving the facts covered by the relevant RFAs. It therefore awarded Arnold only the reduced amount.
However, the court awarded Arnold's requested expert fees in full, approximately $300,000. It concluded Arnold had offered a sufficient accounting and explanation of the expert fees. It found Arnold had sufficiently authenticated the expert invoices and identified the entries in the invoices for which it sought reimbursement. The court explained, "These entries are sufficiently detailed to permit the court to conclude that the work performed and *365expenses incurred were to prove the truth of the matters at issue."
The District appeals the court's order awarding costs of proof. The appeal was stayed pending resolution of the Alcoa appeal involving the merits of the judgment itself. In Alcoa , we affirmed the judgment as to Arnold. (See Alcoa, supra , 12 Cal.App.5th at p. 366, 219 Cal.Rptr.3d 474.) We will now consider the cost award.
DISCUSSION
I
Requests for Admission and Cost of Proof Awards
"A party to a civil action may propound a written request that another party 'admit the genuineness of specified documents, or the truth of specified matters of fact, opinion relating to fact, or application of law to fact.' " ( City of Glendale v. Marcus Cable Associates, LLC (2015) 235 Cal.App.4th 344, 351-352, 185 Cal.Rptr.3d 331 ( City of Glendale ).) Section 2033.420, subdivision (a) provides for an award of costs of proof where a party responding to such a request fails to admit the truth of a matter that is later proved: "If a party fails to admit the genuineness of any document or the truth of any matter when requested to do so under this chapter, and if the party requesting that admission thereafter proves the genuineness of that document or the truth of that matter, the party requesting the admission may *115move the court for an order requiring the party to whom the request was directed to pay the reasonable expenses incurred in making that proof, including reasonable attorney's fees." Such an award must be made unless, among other things, "[t]he party failing to make the admission had reasonable ground to believe that the party would prevail on the matter" or "[t]here was other good reason for the failure to admit." ( § 2033.420, subd. (b)(3)-(4).)8
"Requests for admission are not restricted to facts or documents, but apply to conclusions, opinions, and even legal questions. [Citations.] Thus, requests for admission serve to narrow discovery, eliminate undisputed issues, and shift the cost of proving certain matters. As such, the requests for admission mechanism is not a means by which a party obtains additional information, but rather a dispute resolution device that eliminates the time and expense of formal proof at trial." ( City of Glendale, supra , 235 Cal.App.4th at pp. 353-354, 185 Cal.Rptr.3d 331.)
"The primary purpose of requests for admissions is to set at rest triable issues so that they will not have to be tried; they are aimed at expediting trial. [Citation.] The basis for imposing sanctions under [the statute] is directly related *366to that purpose. Unlike other discovery sanctions, an award of expenses pursuant to [the statute] is not a penalty. Instead, it is designed to reimburse reasonable expenses incurred by a party in proving the truth of a requested admission where the admission sought was 'of substantial importance' [citations] such that trial would have been expedited or shortened if the request had been admitted." ( Brooks, supra , 179 Cal.App.3d at p. 509, 224 Cal.Rptr. 838.)
"Unlike sanctions imposed as a penalty for the nine types of discovery misconduct itemized in Code of Civil Procedure section 2023.010, an award of costs of proof for a denial of a request for admission involves the weighing of a number of factors, such as whether the matter denied was of 'substantial importance'; whether there was a 'reasonable basis' for the denial; whether the party making the denial knew or should have known at *116the time that the requested matter was of 'substantial importance' and was true; whether there were 'other good reasons for the denial'; and whether and to what extent the responding party made a good faith effort otherwise to resolve the matter. [Citations.] Also, unlike sanctions for discovery misconduct, costs of proof under section 2033.420 are awarded after trial; therefore, an award of such costs is not a device used by trial courts to control pretrial proceedings. Instead, as with attorney fees and costs awarded after judgment to a prevailing party, an award of costs of proof is a fee shifting and cost allocation mechanism that is available against parties. And, unlike sanctions for discovery misconduct, such costs cannot be awarded against attorneys." ( City of Glendale, supra , 235 Cal.App.4th at p. 354, 185 Cal.Rptr.3d 331.)
" 'In evaluating whether a "good reason" exists for denying a request to admit, "a court may properly consider whether at the time the denial was made the party making the denial held a reasonably entertained good faith belief that the party would prevail on the issue at trial." [Citation.]' " ( Grace v. Mansourian (2015) 240 Cal.App.4th 523, 529, 192 Cal.Rptr.3d 551 ( Grace ); accord, Laabs, supra , 163 Cal.App.4th at p. 1276, 78 Cal.Rptr.3d 372 ; Miller v. American Greetings Corp. (2008) 161 Cal.App.4th 1055, 1066, 74 Cal.Rptr.3d 776 ( Miller ).) A party's reasonable belief must be grounded in the evidence; it cannot be based merely on "hope or a roll of the dice." ( Grace , at p. 532, 192 Cal.Rptr.3d 551.) It is also not enough for a party making the denial to " 'hotly contest' " the issue; instead, "there must be some reasonable basis for contesting the issue in question before sanctions can be avoided." ( Brooks, supra , 179 Cal.App.3d at p. 511, 224 Cal.Rptr. 838.) Indeed, "the mere fact [that a party] presented evidence at trial is not an automatic justification for denial of the requests. Rather, the issue is whether, in light of that evidence, [the party] could reasonably believe they would prevail." ( Grace , at p. 531, 192 Cal.Rptr.3d 551.)
On the other hand, "[e]xpenses of proving disputed facts which an opposing party denies in response to a request for admission are not recoverable simply because the party promulgating the request prevails at trial." ( Brooks, supra , 179 Cal.App.3d at p. 513, 224 Cal.Rptr. 838.) The opposing party must have no reasonable basis to believe it would prevail. ( Grace, supra , 240 Cal.App.4th at p. 531, 192 Cal.Rptr.3d 551.)
Like other evidence, expert opinion evidence may provide a party with a reasonable ground to believe it will prevail on a matter covered by an RFA. (See, e.g., Grace, supra , 240 Cal.App.4th at p. 533, 192 Cal.Rptr.3d 551.) And, like other *367evidence, the credibility and persuasiveness of expert opinion evidence must be evaluated to determine whether it would be reasonable for a party to believe it would prevail based on it. Because expert opinion evidence necessarily covers matters that are to some extent beyond common experience (see Evid. Code, § 801, subd. (a) ), a party's assessment of the likelihood *117of prevailing on a matter heavily reliant on expert opinion evidence will be different from a matter that does not involve expert testimony. A party may not be able to fully assess an expert's choice of methodology and the substance of his or her analysis. For this reason, federal courts applying a similar standard have found that a party may reasonably deny an RFA based on credible expert opinion evidence, even where the expert opinion evidence is not ultimately believed by the trier of fact. (See, e.g., Benson Tower Condo. Owners Ass'n v. Victaulic Co. (D.Or. 2015) 105 F.Supp.3d 1184, 1197 ["Given the scientific and technical complexity of these issues, the Court finds that Defendant's denials were not unreasonable."]; Scheufler v. General Host Corp. (D.Kan. 1995) 915 F.Supp. 236, 239.)
But it is clear there is no "per se rule that reliance on an expert opinion provides a reasonable ground for a party to believe he would prevail at trial." ( Marchand v. Mercy Medical Ctr. (9th Cir. 1994) 22 F.3d 933, 937 ( Marchand ).) In limited circumstances, courts have found that a party's reliance on expert opinion evidence was unreasonable. For example, the federal appellate court in Marchand affirmed a trial court order awarding costs of proof where the factual basis for the expert's opinion was fatally undermined by the party's own admissions at trial. ( Id. at pp. 937-938.) In Wimberly v. Derby Cycle Corp. (1997) 56 Cal.App.4th 618, 637, 65 Cal.Rptr.2d 532 ( Wimberly ), this court reversed an order denying costs of proof where a party relied on one expert who had not been properly disclosed and another expert whose deposition testimony was inadmissible. Wimberly explained, "As the above facts establish, the only inference that can reasonably be drawn is that when [defendant] denied [plaintiff's] requests for admissions, it had no reasonable belief it could prevail on the causation and defect issues. [Defendant's] misunderstanding of the law regarding the use of expert witness depositions in lieu of live testimony, or its hope [plaintiff] would not object, does not provide reasonable grounds for denying the requested admissions, and accordingly an award of expenses was mandated." ( Id. at p. 638, 65 Cal.Rptr.2d 532.)
Whether a party has a reasonable ground to believe he or she will prevail, in the context of requests for admissions heavily reliant on expert opinion evidence, will depend on factors within the reasonable understanding of the party. Such factors may include whether the expert has sufficient qualifications and experience to opine on the matter at issue, whether the expert's opinions will likely be admissible at trial, whether the facts underlying the expert's opinions are supported by the evidence, whether the expert's methodology appears reasonable, and whether the expert's analysis is grounded in logic. A party cannot rely on a plainly unqualified expert, or a sham opinion, to avoid cost of proof sanctions.
*118Where a party's position is supported by a credible opinion from a qualified expert, the mere fact that an opposing party also has a credible opinion from a qualified expert will not in most cases preclude the party from reasonably believing it would prevail. Something about the state of the evidence must make the party's reliance on its own expert's opinion unreasonable. Whether a party has a reasonable *368ground to believe he or she will prevail necessarily requires consideration of all the evidence, both for and against the party's position, known or reasonably available to the party at the time the RFA responses are served. Beyond the expert opinions themselves, a party must consider other evidence, both lay and expert, that bears on the matter at issue. These general principles will govern our analysis.
II
Standard of Review
Courts have uniformly reviewed orders granting or denying cost of proof awards for abuse of discretion. (See, e.g., Grace, supra , 240 Cal.App.4th at p. 529, 192 Cal.Rptr.3d 551 ["The court's determination of whether costs of proof should be awarded is reviewed for abuse of discretion."]; City of Glendale, supra , 235 Cal.App.4th at p. 352, 185 Cal.Rptr.3d 331 ; Bloxham v. Saldinger (2014) 228 Cal.App.4th 729, 753, 175 Cal.Rptr.3d 650 ; Carlsen v. Koivumaki (2014) 227 Cal.App.4th 879, 903-904, 174 Cal.Rptr.3d 339 ; Laabs, supra , 163 Cal.App.4th at pp. 1275-1276, 78 Cal.Rptr.3d 372 ; Miller, supra , 161 Cal.App.4th at p. 1066, 74 Cal.Rptr.3d 776 ; Wimberly, supra , 56 Cal.App.4th at p. 637, fn. 10, 65 Cal.Rptr.2d 532 ; Brooks, supra , 179 Cal.App.3d at p. 508, 224 Cal.Rptr. 838 ; Haseltine v. Haseltine (1962) 203 Cal.App.2d 48, 60, 21 Cal.Rptr. 238.)
The District asks us to depart from this consensus. It contends we should review the order, or at least the "reasonable ground" determination, de novo. It primarily relies on authorities interpreting the probable cause element of a malicious prosecution tort. (See Sheldon Appel Co. v. Albert & Oliker (1989) 47 Cal.3d 863, 881, 254 Cal.Rptr. 336, 765 P.2d 498 ; Lanz v. Goldstone (2015) 243 Cal.App.4th 441, 462, 197 Cal.Rptr.3d 227 ; Knight v. City of Capitola (1992) 4 Cal.App.4th 918, 932, 6 Cal.Rptr.2d 874.) The probable cause element requires a showing that the prior action, on which the malicious prosecution tort is based, was brought without probable cause, i.e., no reasonable attorney would believe that the prior action was legally tenable. ( Sheldon Appel , at pp. 871, 886, 254 Cal.Rptr. 336, 765 P.2d 498.) Under that element, "[i]f there is ' "no dispute as to the facts upon which an attorney acted in filing the prior action, the question of whether there was probable cause to institute that action is purely legal." [Citation.]' " ( Lanz , at p. 462, 197 Cal.Rptr.3d 227.) Because it is a legal question, it is reviewed de novo on appeal. ( Knight , at p. 932, 6 Cal.Rptr.2d 874.) The District also cites other authorities in the criminal context where "reasonableness" standards govern and the trial court's determination on those issues is reviewed de *119novo. (See People v. Moore (2011) 51 Cal.4th 386, 395, 121 Cal.Rptr.3d 280, 247 P.3d 515 ; People v. Cromer (2001) 24 Cal.4th 889, 901, 103 Cal.Rptr.2d 23, 15 P.3d 243 ; People v. Leyba (1981) 29 Cal.3d 591, 597-598, 174 Cal.Rptr. 867, 629 P.2d 961.)
The District has not shown the established standard of review is incorrect. The question is not whether a reasonable litigant would have denied the RFAs. Nor is the question simply whether the litigant had some minimum quantum of evidence to support its denial (i.e., "probable cause"). The relevant question is whether the litigant had a reasonable, good faith belief he or she would prevail on the issue at trial. ( Grace, supra , 240 Cal.App.4th at p. 529, 192 Cal.Rptr.3d 551 ; Laabs, supra , 163 Cal.App.4th at p. 1276, 78 Cal.Rptr.3d 372.) Consideration of this question requires not only an assessment of the substantiality of the evidence for and against the issue known or available to the party, but also the credibility of that evidence, the likelihood that it would be admissible *369at trial and persuasive to the trier of fact, the relationship of the issue to other issues anticipated to be part of trial (including the issue's importance), the party's efforts to investigate the issue and obtain further evidence, and the overall state of discovery at the time of the denials and thereafter. Because the trial court supervises discovery and presides over trial, it is in a much better position to weigh these considerations and decide whether, in its discretion, the party who made the denials should be responsible for costs of proof on the issue. (See People v. Roldan (2005) 35 Cal.4th 646, 688, 27 Cal.Rptr.3d 360, 110 P.3d 289 ["The abuse of discretion standard ... reflects the trial court's superior ability to consider and weigh the myriad factors that are relevant to the decision at hand."]; see also In re Robert L. (1993) 21 Cal.App.4th 1057, 1065, 24 Cal.Rptr.2d 654.)
As noted, the primary purpose of RFAs is to narrow the issues in dispute and expedite trial. ( City of Glendale, supra , 235 Cal.App.4th at pp. 353-354, 185 Cal.Rptr.3d 331 ; Brooks, supra , 179 Cal.App.3d at p. 509, 224 Cal.Rptr. 838.) Whether a party had a reasonable, good faith belief he or she would prevail on an issue at trial, and therefore force a trial on that issue, implicates core discretionary functions of the trial court: management of the discovery and trial proceedings before it. (See In re Groundwater Cases (2007) 154 Cal.App.4th 659, 693, 64 Cal.Rptr.3d 827 ["Management of discovery lies within the sound discretion of the trial court ...."]; Moyal v. Lanphear (1989) 208 Cal.App.3d 491, 498, 256 Cal.Rptr. 296 ["Since trial management is a discretionary area, the proper standard of review for a challenge to trial management orders is abuse of discretion."].) An abuse of discretion standard is therefore proper. We must review the trial court's decision in light of the applicable legal principles, but it would be inappropriate to simply substitute our judgment for that of the trial court under these circumstances. (See Department of Parks & Recreation v. State Personnel Bd. (1991) 233 Cal.App.3d 813, 831, 284 Cal.Rptr. 839 ["The abuse of discretion standard ... measures whether, *120given the established evidence, the act of the lower tribunal falls within the permissible range of options set by the legal criteria."].)
The District makes much of the fact that the standard here involves a consideration of reasonableness. But, in similar contexts, we defer to the trial court's determination of reasonableness. (See Doe v. United States Swimming, Inc. (2011) 200 Cal.App.4th 1424, 1434, 133 Cal.Rptr.3d 465 [discovery sanctions for acting without substantial justification, i.e., without "clearly reasonable" justification]; Obregon v. Superior Court (1998) 67 Cal.App.4th 424, 430, 79 Cal.Rptr.2d 62 [reasonable and good faith attempt to meet and confer]; see also PLCM Group, Inc. v. Drexler (2000) 22 Cal.4th 1084, 1095, 95 Cal.Rptr.2d 198, 997 P.2d 511 [reasonable attorney fees].) The District also emphasizes that cost of proof awards may have a chilling effect on a litigant's ability to pursue his or her claims or defenses. But any such chilling effect is incidental. Unlike the tort of malicious prosecution, a litigant is not punished merely for pursuing a claim. The litigant must instead reimburse an opposing party's costs in proving an issue where the litigant did not have a reasonable, good faith belief it would prevail at trial on that issue, after being expressly requested to admit the issue. Moreover, the primary interest implicated by cost of proof awards is not constitutional, as in the criminal cases cited by the District, but managerial. We will review the court's order for abuse of discretion. (See Grace, supra , 240 Cal.App.4th at p. 529, 192 Cal.Rptr.3d 551, and cases cited, ante .)
*370III
Reasonable Grounds
We turn next to the merits of the order awarding Arnold its costs of proof. The RFAs at issue asked the District to admit that (1) Arnold did not release PCE, TCE, 1,1,1-TCA, or 1,4-dioxane during its business operations at the site; (2) Arnold did not release PCE, TCE, 1,1,1-TCA, or 1,4-dioxane that caused contamination of groundwater in the North Basin area; and (3) Arnold did not contribute to VOC contamination in soil or groundwater in the area surrounding its site. The District is not responsible for Arnold's costs if it shows it "had reasonable ground to believe [it] would prevail on the matter." ( § 2033.420, subd. (b)(3) ; see Grace, supra , 240 Cal.App.4th at pp. 529, 532, 192 Cal.Rptr.3d 551.)
In assessing whether the District had a reasonable ground to believe it would prevail, we are mindful of the nature of this litigation, which involves sophisticated scientific analyses of soil and groundwater contamination. As noted, where RFAs require sophisticated analyses of technical issues, courts are more willing to credit a party's reasonable belief that it would *121prevail based on expert opinion evidence. (See, e.g., Benson Tower Condo. Owners Ass'n v. Victaulic Co., supra , 105 F.Supp.3d at p. 1197.) Courts in the past have been willing to find a party's reliance on expert opinion evidence unreasonable only where the expert opinion evidence would be inadmissible at trial (see Wimberly, supra , 56 Cal.App.4th at pp. 637-638, 65 Cal.Rptr.2d 532 ) or where the expert opinion evidence was fatally undermined by facts known to the party (see Marchand, supra , 22 F.3d at pp. 937-938 ). Although those circumstances are not exclusive, and a court may act within its discretion in finding a party's reliance unreasonable in other contexts, these authorities illustrate the issues involved where courts consider expert opinion evidence supporting an RFA denial.
We note that Arnold's briefing often misperceives the substantive legal issue here. Arnold treats the District's arguments that it held a reasonable, good faith belief it would prevail as improper attempts to reargue the merits of its claims. Arnold variously contends that the doctrines of law of the case and collateral estoppel preclude the District from making these arguments. Arnold is incorrect. Whether Arnold was ultimately successful at trial is not the issue here. Rather, the issue is whether the District had a reasonable, good faith belief-prior to trial-that it would prevail. We must necessarily consider the basis for the District's alleged belief in determining its reasonableness. Arnold's subsequent success at trial does not alter our inquiry. We will address each chemical separately below.
PCE
The two PCE-related RFAs asked the District to admit Arnold "did not release PCE at any time during its business operations at [the Arnold site]" and Arnold "did not release PCE that caused contamination of groundwater in the North Basin area at issue in this action." The District does not contest that it had no reasonable grounds to believe it would prevail on the latter issue (PCE contamination of groundwater), given that Waddell did not and could not opine at trial that any such contamination had occurred. It does challenge the trial court's finding that it had no reasonable grounds to believe it would prevail on the former issue (PCE release). For reasons we will explain, we agree with the District that the trial court abused its discretion by finding that the District had no reasonable grounds to believe it would prevail on the issue of Arnold releasing *371PCE during its business operations at the site.
The threshold issue regarding this RFA is whether the District had a reasonable ground to believe it would prevail on the issue of whether Arnold used PCE at all. Persuasive evidence showed Arnold had used PCE. Hopen, Arnold's former employee, testified unequivocally at deposition and at trial that Arnold used PCE as a stripping agent. Rohrer, Arnold's expert, admitted *122at deposition that Arnold's operations "likely" involved the use of PCE. The soil at the Arnold site was contaminated with PCE, which showed that someone had used and released PCE there. There was no evidence any other occupant of the site used PCE. Waddell opined that it was very unlikely another occupant used PCE, based on the nature of their operations there.
Against this evidence, Arnold relied on the testimony of Farmer, another former employee, who testified that he was not aware of any PCE use. But Farmer admitted he did not know what chemical was used in Arnold's stripping operations. Arnold also relied on permits identifying 1,1,1-TCA as the solvent used in Arnold's operations. Other permits did not identify a solvent, however, so it was possible Arnold used PCE under those permits.
Given this evidence, it is clear the District had a reasonable ground to believe it would prevail on the threshold question of whether Arnold used PCE. It was reasonable for the District to believe that Hopen's testimony, combined with the PCE contamination at the site, would lead a reasonable trier of fact to believe Arnold had used PCE. Nothing about Arnold's evidence foreclosed the District's reasonable reliance on its own evidence regarding PCE. Arnold's reliance on contrary evidence did not make the District's reliance on its own evidence of PCE use unreasonable.
Similarly, given Arnold's plausible PCE use, the District plainly had a reasonable ground to believe it would prevail on the question whether Arnold released PCE during its operations at the site. The contamination at the site shows that PCE was released there by someone. Waddell and Rohrer agreed on that point. Given the evidence of (1) PCE use by Arnold, (2) the risk of PCE spills and leaks into the environment created by Arnold's stripping operations, and (3) the absence of any evidence that any other person used PCE at the site, the District plainly had a reasonable ground to believe that a trier of fact would find Arnold released PCE at its site.
In its order, the trial court appeared to base its award of costs on the fact that "Dr. Waddell waited until trial to testify there was no evidence of PCE ... use by Arnold at its site." But Waddell did not testify, at trial or otherwise, that there was no evidence of PCE use by Arnold. On the contrary, Waddell's position was that Arnold had used and released PCE at the site. PCE-related evidence was excluded at trial because Waddell could not testify that Arnold's use of PCE contaminated groundwater . But the question of PCE contamination of groundwater is different from the question of a PCE release. While they are intertwined for purposes of Arnold's liability for PCE groundwater contamination, they reflect separate RFAs, covering separate factual issues, and they must be analyzed separately.
*123While the trial court has wide discretion to award costs of proof, this discretion must be exercised reasonably and in conformance with applicable law. (See Horsford v. Board of Trustees of California State University (2005) 132 Cal.App.4th 359, 393, 33 Cal.Rptr.3d 644.) Given the evidence outlined above, the court *372abused its discretion by concluding the District did not have reasonable grounds to believe it would prevail when it denied Arnold's RFA seeking an admission that Arnold "did not release PCE at any time during its business operations at [the Arnold site]."9
TCE
The two TCE-related RFAs asked the District to admit Arnold "did not release TCE at any time during its business operations at [the Arnold site]" and Arnold "did not release TCE that caused contamination of groundwater in the North Basin area at issue in this action." The District contends the trial court erred by awarding costs of proof based on both RFAs because it had a reasonable ground to believe it would prevail on these issues. For reasons we will explain, we agree.
As with PCE, the threshold issue is whether the District had a reasonable ground to believe it would prevail on the issue of whether Arnold used TCE at all. While the evidence on this question was not as compelling as with PCE, it clearly provided a reasonable ground for the District to believe a trier of fact would find Arnold used TCE. Hopen testified that Arnold used TCE as a degreaser, using its full name "trichloroethylene" in both his deposition and at trial. At trial, following testimony on the subject in both direct and cross-examination, Hopen was asked whether his "best memory" was that the chemical used by Arnold was "trichloroethylene," and he confirmed it was "[t]richloroethylene."
While Hopen also testified that he saw "1,1,1" on barrels of VOC solvent-which could only refer to 1,1,1-TCA, not TCE-his confusing testimony does not make it inherently unreasonable for the District to believe that Arnold used TCE, especially given the other evidence in the record. As with PCE, extensive TCE contamination was found in the soil at the Arnold site, which showed that someone had used and released TCE there. And, again as with PCE, there was no evidence any other occupant of the site used TCE. Indeed, TCE's use was restricted after 1976, while Arnold still operated at the site. Only one other industrial company occupied the site before *124Arnold. Farmer, who was also employed by that company, testified that he was not aware of any VOC use during its occupancy. This evidence corroborated Hopen's recollection that Arnold used TCE.10 *373Arnold relied on Farmer's testimony and the permits described above, which identified only 1,1,1-TCA. The trial court, too, wrote in its written order awarding costs, "Former employee Hopen testified Arnold used TCE, but the only labels he could remember had '1,1,1' on them. That designation is associated not with TCE, but with TCA." Again, however, nothing about this evidence foreclosed the District's reasonable reliance on its contrary evidence. The District could reasonably believe that its own evidence would be believed by the trier of fact. The fact that the trier of fact here believed Arnold's evidence does not make the District's reliance on its own evidence unreasonable. These circumstances are far different from Grace, supra , 240 Cal.App.4th at pages 530 to 532, 192 Cal.Rptr.3d 551, where a party relied only on his own testimony to deny an RFA; where that testimony was contradicted by other percipient witnesses, a police report, and undisputed expert opinion evidence; and where the party's attorney appeared to acknowledge the weakness of his position in his opening statement.
If, as we have found, the District had reasonable grounds to believe that a trier of fact would find that Arnold used TCE, it follows from the evidence that the District would also have reasonable grounds to believe that a trier of fact would find that Arnold released TCE during its operations at the site. A TCE release undeniably occurred at the site. Rohrer and Waddell agreed on this point. And there was no evidence that any occupant of the site other than Arnold used TCE. Indeed, as noted, the evidence was to the contrary: Farmer testified that the prior occupant did not use any VOCs, and its use after Arnold's departure was heavily regulated. Moreover, the evidence showed that Arnold's degreasing activities carried a risk of spills and leaks into the environment. The only reasonable conclusion based on the evidence is that the District had reasonable grounds to believe it would prevail when it denied Arnold's RFA seeking an admission that Arnold "did not release TCE at any *125time during its business operations at [the Arnold site]." The trial court abused its discretion by finding otherwise.
The next RFA covered whether Arnold released TCE that caused contamination of groundwater. The undisputed evidence showed that TCE was detected in soil samples down to 105 feet below ground level, the lowest level tested, approximately 10 feet above groundwater. The groundwater itself was not tested.
In denying this RFA, the District relied on Waddell's expert opinion that TCE contamination at the Arnold site contributed to groundwater contamination. Waddell, in turn, primarily relied on grab samples of groundwater on adjoining properties. Testing revealed higher concentrations of TCE in groundwater to the northwest (GW-2) and west (GW-3), when compared to the east (GW-1). In Waddell's opinion, these results showed that the Arnold site contributed to groundwater TCE contamination because the concentrations increased as groundwater flowed in the north-northwest direction under the site. Waddell's opinion was supported by undisputed data showing that TCE was present in both the shallow soil and deeper soil layers in the vicinity of groundwater.
Arnold's evidence in opposition consisted primarily of Rohrer's opinion that the grab sample data was insufficient to prove groundwater contamination. Rohrer testified that the groundwater flow was too variable to conclude that the increases in TCE concentration were caused by groundwater flowing underneath the Arnold site. Arnold also undermined Waddell's credibility by showing that he omitted *374a data point from a cross-gradient monitoring well (not a grab sample) that recorded a high TCE concentration, as well as several upgradient bore holes on a different contaminated property. Waddell also misstated the extent of TCE contamination in the shallow soil at the neighboring Johnson Controls site and appeared to lack information on the composition of VOC solvents used by Johnson Controls and other occupants of that site.
In its postjudgment order awarding costs, the trial court appeared to believe that Waddell's opinion regarding TCE contamination relied on an obvious contradiction, i.e., that a TCE detection north of the Arnold site proved groundwater contamination, even though in the court's view Waddell testified that groundwater in that area "consistently moved in a west/southwest direction." But the court was incorrect that Waddell's opinion relied on such a contradiction. Waddell identified areas of both northwesterly and southwesterly flows around the Arnold site. With respect to the grab samples themselves, he testified the prevailing flow was to the northwest and west. Rohrer also did not definitively testify that the groundwater flow was southwesterly.
*126His criticism of Waddell relied primarily on the variable flow of groundwater, and the inability of Waddell (or anyone) to determine whether the allegedly downgradient grab samples were in fact downgradient of the Arnold site. The evidence regarding groundwater flow was therefore disputed; it was not irrational or illogical for Waddell to conclude that the grab samples to the northwest and west were downgradient of the Arnold site.
Based on our review of the record, it is clear Waddell's opinion could have been believed by a trier of fact. He had more than sufficient qualifications to opine on groundwater contamination issues, including a doctoral degree in geology and 30 years of relevant experience; his opinions were grounded in relevant and reliable evidence; and his methodology was reasonable and logical to an outside observer. His opinion was not a sham. It was reasonable for the District to believe Waddell's opinions would be credited by a trier of fact and therefore deny Arnold's RFA.
There is nothing in the record that would support the trial court's contrary decision. Waddell's use of grab samples was subject to criticism, but they logically and reasonably tended to prove that the Arnold site contributed to TCE groundwater contamination. And the grab samples were not the sole basis for Waddell's opinion on this matter. He also relied on soil testing, which showed extensive TCE contamination, including in deeper soils at the site. TCE contamination was detected down to the lowest level tested, only 10 feet above groundwater.
Arnold focuses on Waddell's misstatements and omissions regarding the neighboring Johnson Controls site. But these misstatements and omissions do not support the conclusion that the District's reliance on Waddell's opinion was unreasonable. First, Arnold points out that Waddell omitted from his trial testimony evidence of TCE contamination in the shallow soil at the Johnson Controls site. Waddell included the correct data in his expert report, however, so his erroneous trial testimony does not call into question the District's reliance on his earlier opinions.
Second, Arnold notes that Waddell was impeached with evidence that Johnson Controls used degreasers (contrary to Waddell's understanding) and that it used a VOC solvent that contained TCE. These latter two omissions, however, were not consequential under the circumstances. Waddell acknowledged that TCE had been released at the Johnson Controls site; the *375TCE concentrations found at the Arnold site were simply far higher. While Waddell may have been mistaken about the particulars of TCE use at the Johnson Controls site, it is not reasonable to believe that his opinions regarding TCE contamination at the Arnold site were fatally undermined by those mistakes. (Cf. Marchand, supra , 22 F.3d at pp. 937-938.) *127In sum, while the trial court was entitled to find that Waddell was not credible, that his use of grab samples was insufficient to prove TCE groundwater contamination, and that the District ultimately did not carry its burden of proving that Arnold contributed to TCE groundwater contamination, the record does not compel such findings. The District's contrary position was supported by TCE contamination data for both soil and groundwater, as well as the reasoned opinion of a highly qualified expert. This expert opinion, in turn, was neither inadmissible (cf. Wimberly, supra , 56 Cal.App.4th at pp. 637-638, 65 Cal.Rptr.2d 532 ) nor fatally undermined by facts known to the District (cf. Marchand, supra , 22 F.3d at pp. 937-938 ). The District's position was based on far more than mere "hope or a roll of the dice." (Cf. Grace, supra , 240 Cal.App.4th at p. 532, 192 Cal.Rptr.3d 551.) The District had a reasonable ground to believe it would prevail, and the trial court's decision otherwise was an abuse of discretion.
1,1,1-TCA
The two 1,1,1-TCA-related RFAs asked the District to admit Arnold "did not release 1,1,1-TCA at any time during its business operations at [the Arnold site]" and Arnold "did not release 1,1,1-TCA that caused contamination of groundwater in the North Basin area at issue in this action." The District contends the trial court erred by awarding costs of proof based on both RFAs because it had a reasonable ground to believe it would prevail on these issues. Again, for reasons we will explain, we agree.
Unlike PCE and TCE, Arnold admitted using 1,1,1-TCA. Farmer testified in his deposition that Arnold used 1,1,1-TCA in its degreasing operations. A number of permits produced in discovery identified 1,1,1-TCA for use in degreasers and other equipment, including one that described a 550-gallon above-ground storage tank for that chemical.
The shallow soil at the Arnold site was contaminated with 1,1,1-TCA and its degradation product 1,1-DCE. A release of 1,1,1-TCA plainly occurred there. While Arnold used 1,1,1-TCA for decades, there was no evidence that any other occupant of the site used 1,1,1-TCA. Waddell opined that it very unlikely another occupant used 1,1,1-TCA, based on the nature of their operations there. Moreover, as we have noted, the evidence showed that Arnold's degreasing activities carried a risk of spills and leaks into the environment. Given the extensive 1,1,1-TCA and 1,1-DCE contamination at the Arnold site, and the lack of any evidence that any occupant other than Arnold used 1,1,1-TCA, the only reasonable conclusion based on the evidence is that the District had reasonable grounds to believe it would prevail when it denied Arnold's RFA seeking an admission that Arnold "did not release 1,1,1-TCA at any time during its business operations at [the Arnold site]." The trial court abused its discretion by finding otherwise.
*128The next RFA covered whether Arnold released 1,1,1-TCA that caused contamination of groundwater. Because 1,1,1-TCA degrades into 1,1-DCE (and acetic acid), the District relied on 1,1-DCE detections to show that Arnold's release of 1,1,1-TCA caused groundwater contamination. As with TCE, the undisputed evidence *376showed that 1,1-DCE was detected in soil samples down to 105 feet below ground level, the lowest level tested, approximately 10 feet above groundwater.11
Also as with TCE, the District relied on Waddell's expert opinion to deny this RFA. Waddell primarily relied on 1,1-DCE detections in his groundwater grab samples, which showed slightly higher 1,1-DCE concentrations to the northwest (GW-2) and west (GW-3), when compared with the east (GW-1). Waddell opined that these data showed that the Arnold site contributed to groundwater 1,1-DCE contamination because the concentrations increased as groundwater flowed in the north-northwest direction under the site. The soil testing data supported Waddell's opinion, since it showed 1,1-DCE contamination down to the lowest level tested, 10 feet from groundwater. It was therefore plausible that 1,1-DCE had reached groundwater at the site.
Arnold again relied on Rohrer's criticism of Waddell's grab sample data. Rohrer testified that the data was inadequate to show that the Arnold site contributed to 1,1-DCE contamination in groundwater. Arnold also impeached Waddell with evidence that Johnson Controls had to some extent used solvents with 1,1,1-TCA at its neighboring site.
In its postjudgment order, the trial court based its award of costs on a perceived lack of evidence of any release of 1,1,1-TCA by Arnold: "While Arnold used 1,1,1-TCA in its operations, there was no substantial evidence that Arnold ever caused a release into the soil, much less groundwater." We disagree that there was no substantial evidence to support a finding that Arnold released 1,1,1-TCA at the site.12 Arnold used large quantities of 1,1,1-TCA during its long occupancy at the site, its operations carried the risk of releases into the environment, the site was contaminated by a 1,1,1-TCA
*129release by someone , and there was no evidence any other occupant of the site used 1,1,1-TCA. Based on this evidence, a trier of fact could reasonably infer that Arnold released 1,1,1-TCA during its operations at the site. (See Orange County Water Dist. v. Sabic Innovative Plastics US, LLC (2017) 14 Cal.App.5th 343, 381-382, 222 Cal.Rptr.3d 83 ; United Alloys, Inc. v. Baker (C.D.Cal. 2011) 797 F.Supp.2d 974, 1000.)
However, while the absence of substantial evidence to support a party's position may indicate that a party did not have a reasonable ground to believe it would prevail on that matter, the opposite is not necessarily true. "[T]he mere fact defendants presented evidence at trial is not an automatic justification for denial of the requests. Rather, the issue is whether, in light of that evidence, defendants could reasonably believe they would prevail."
*377( Grace, supra , 240 Cal.App.4th at p. 531, 192 Cal.Rptr.3d 551.)
The state of the evidence regarding 1,1,1-TCA and 1,1-DCE groundwater contamination is essentially the same as that regarding TCE groundwater contamination. Arnold primarily relied on Waddell's expert opinion, which was by all appearances credible. Waddell was a qualified expert, his opinions were grounded in the evidence, and his methodology was ostensibly sound. He did not provide a sham opinion. It was reasonable for the District to believe his opinions would be persuasive at trial, and for reasons we have already explained nothing in the record supports the trial court's contrary decision.
Arnold argues that early testing did not detect 1,1,1-TCA in any significant concentrations. But that testing was limited, and later testing revealed extensive 1,1,1-TCA and 1,1-DCE contamination at the site. The early testing does not preclude the District's reliance on later testing. Arnold also relies on a 1994 report, which it asserts found no VOC groundwater contamination, and a 1995 letter from the Orange County Heath Care Agency (OCHCA). But the 1994 report was limited to PCE; it did not consider 1,1,1-TCA or 1,1-DCE. The 1995 OCHCA letter addressed only surface health threats, finding that no further action was needed to address them. The OCHCA did not make any findings about groundwater contamination, except to note that contamination remained at the site and could potentially affect groundwater. Neither of these documents prevented the District from having a reasonable belief it would prevail.
Viewing the evidence as a whole, it is clear the District had a reasonable ground to believe it would prevail on the issue of groundwater contamination caused by 1,1,1-TCA. It had a reasonable and apparently credible opinion from a highly qualified expert, grounded in the evidence, that supported its position. While the trial court did not credit the expert's opinion at trial, his *130opinion was not revealed to be sham, inadmissible, or wholly without foundation. (Cf. Wimberly, supra , 56 Cal.App.4th at pp. 637-638, 65 Cal.Rptr.2d 532 ; Marchand, supra , 22 F.3d at pp. 937-938.) The District had far more than "hope or a roll of the dice" going into trial. (Cf. Grace, supra , 240 Cal.App.4th at p. 532, 192 Cal.Rptr.3d 551.) It had a reasonable ground to believe it would prevail, and the trial court's decision otherwise was an abuse of discretion.
1,4-Dioxane
The two 1,4-dioxane-related RFAs asked the District to admit Arnold "did not release 1,4-dioxane at any time during its business operations at [the Arnold site]" and Arnold "did not release 1,4-dioxane that caused contamination of groundwater in the North Basin area at issue in this action." The District contends the trial court erred by awarding costs of proof based on both RFAs because it had a reasonable ground to believe it would prevail on these issues. We disagree.
While Waddell testified that 1,4-dioxane was "commonly" added to 1,1,1-TCA as a stabilizer, there was no evidence of the 1,1,1-TCA formulation used by Arnold. Waddell testified that there were other stabilizers used for 1,1,1-TCA, and the District has cited no evidence quantifying the likelihood that Arnold's formulation included 1,4-dioxane rather than an alternative. And, because none of the investigations at the Arnold site tested for 1,4-dioxane, there was also no evidence it was present in the shallow soil there.
The only evidence regarding 1,4-dioxane at the site came from Waddell's grab samples. The data showed that concentrations of 1,4-dioxane were very slightly higher at *378one downgradient location (GW-2), significantly higher at another downgradient location (GW-3), and lower at the third potentially downgradient location (GW-4/4A). In his expert report, Waddell opined that 1,4-dioxane had impacted groundwater as a result of VOC releases by Arnold. But, importantly, Waddell did not offer any opinions regarding Arnold and 1,4-dioxane at trial.
Given the dearth of evidence regarding 1,4-dioxane, and the District's failure to pursue liability at trial based on 1,4-dioxane contamination, we conclude the trial court could reasonably find that the District had no reasonable ground to believe it would prevail on the 1,4-dioxane RFAs at issue here. " 'Where certain facts exist which the responding party does not intend to contest at trial, the proper time to admit and permit those facts to be established is during pretrial discovery.' " ( *131Wimberly, supra , 56 Cal.App.4th at p. 634, 65 Cal.Rptr.2d 532.) The District has not shown the court abused its discretion by awarding costs of proof based on these RFAs.13
General VOC Contamination
The two general VOC contamination RFAs asked the District to admit Arnold "did not contribute to VOC contamination in soil in the geographic area [identified as the area surrounding the former Arnold site]" and Arnold "did not contribute to VOC contamination in groundwater presently located in the geographic area [identified as the area surrounding the former Arnold site]." For reasons we will explain, our conclusions regarding these RFAs flow automatically from our discussion above. For both RFAs, we conclude the trial court abused its discretion by awarding costs of proof.
The first general VOC contamination RFA asked the District to admit Arnold "did not contribute to VOC contamination in soil in the geographic area [identified as the area surrounding the former Arnold site]." As noted, we have concluded that the District plainly had a reasonable ground to believe it would prevail on the issues of PCE, TCE, and 1,1,1-TCA contamination in soil at the Arnold site. Because PCE, TCE, and 1,1,1-TCA are VOCs, it necessarily follows that the District had a reasonable ground to believe it would prevail on the issue of VOC contamination in soil in the relevant area.
The same reasoning applies to the second general VOC contamination RFA, which asked the District to admit Arnold "did not contribute to VOC contamination in groundwater presently located in the geographic area [identified as the area surrounding the former Arnold site]." Because the District had a reasonable ground to believe it would prevail on the issues of TCE and 1,1,1-TCA contamination in groundwater at the Arnold site, the District also had a reasonable ground to believe it would prevail on the issue of VOC contamination in groundwater in the area surrounding the Arnold site.
IV**
*132DISPOSITION
The order is reversed. The matter is remanded to the trial court with directions *379to receive additional evidence, if necessary, and to enter a new order awarding reasonable costs of proof, in an amount to be determined in its discretion, based on the District's denials of Arnold's RFA Nos. 17 (PCE groundwater contamination), 22 (1,4-dioxane release), and 23 (1,4-dioxane groundwater contamination) only. In the interests of justice, the parties shall bear their own costs on appeal.
WE CONCUR:
BENKE, Acting P. J.
HUFFMAN, J.
Further statutory references are to the Code of Civil Procedure unless otherwise specified.
In an opinion filed last year, we addressed the District's appeal from the judgment itself. (See Orange County Water Dist. v. Alcoa Global Fasteners, Inc. (2017) 12 Cal.App.5th 252, 219 Cal.Rptr.3d 474 (Alcoa ).) That opinion provides a detailed discussion of the District's claims, the procedural history of the litigation, the evidence presented at trial, the court's decision, and the judgment. In this opinion, we will discuss only those aspects of the litigation relevant to the issues raised in this appeal.
One document identified PCE for use in a stripper. The trial court found that this document was inadmissible at trial.
RCRA and CERCLA are federal environmental protection statutes. RCRA is the Resource Conservation and Recovery Act (42 U.S.C. § 6901 et seq. ). CERCLA is the Comprehensive Environmental Response, Compensation, and Liability Act (42 U.S.C. § 9601 et seq. ).
On direct, Hopen testified as follows: "Q. Okay. Are you familiar with what was placed in the degreaser at Arnold? [¶] A. Trichloroethylene. [¶] Q. How did you know that was what was in the degreaser, could you explain? [¶] A. There were labels-they were black barrels, they had labels on them. And I believe it said trichloroethylene, 1,1,1, and we used to call it trike, for short." On cross-examination, Hopen testified: "Q. Have you ever heard of trichloroethane? [¶] A. I may have, but maybe it wasn't-I don't know if it was trichloroethylene or what you just said. I can't remember exactly. [¶] Q. So sitting here today you don't remember today whether it was trichloroethane or trichloroethylene that was on the barrels. [¶] A. Ethane, I think, trichloroethylene, yeah, with the 'l ,' I believe. [¶] Q. You can't remember one way or another? [¶] A. Not unless I saw-not unless I saw-I remember the '1,1,1' on it. [¶] Q. So you remember the 1,1,1 for sure? [¶] A. Yes, and I remember there was a VOC." On redirect, Hopen was asked, "Is it your best memory that the chemical used in the degreaser was trichloroethylene?" He responded, "Trichloroethylene."
Separately, Waddell was told by the District's counsel to exclude dates from his report, which might have assisted the defendants in establishing defenses based on the statute of limitations. The RFAs at issue here are not affected by statute of limitations issues. Arnold claims the District asked Waddell to omit certain contamination data from his report, but that claim is not supported by the record.
Arnold's reduction in expert costs reflected the elimination of expert work unrelated to the issues in the RFAs. For example, Arnold's expert Rohrer analyzed the effectiveness of the District's plan to address contamination in the North Basin and modeled the flow of contamination from the Arnold site assuming the truth of Waddell's conclusions regarding the contamination there.
Until 1987, the statute governing cost of proof awards did not explicitly mention an exception where the responding party had a "reasonable ground to believe that the party would prevail on the matter." Instead, the statute required the court to find "there were no good reasons for the denial" before an award could be made. (Former § 2034, subd. (c), repealed by Stats. 1986, ch. 1334, § 1; see Smith v. Circle P Ranch Co. (1978) 87 Cal.App.3d 267, 273-274, 150 Cal.Rptr. 828.) The "good reasons" exception was interpreted to incorporate, among other factors, whether "at the time the denial was made the party making the denial held a reasonably entertained good faith belief that the party would prevail on the issue at trial." (Brooks v. Am. Broad. Co. (1986) 179 Cal.App.3d 500, 511, 224 Cal.Rptr. 838 (Brooks ).) The enactment of the Civil Discovery Act of 1986 codified this factor as a separate exception (with some alterations), in addition to the general "good reason" exception. (Stats. 1986, ch. 1334, § 2; see § 2033.420, subd. (b)(3)-(4).) The Brooks formulation continues to inform our interpretation of the statute. (See, e.g., Laabs v. City of Victorville (2008) 163 Cal.App.4th 1242, 1276, 78 Cal.Rptr.3d 372 (Laabs ).)
In light of our conclusion, we need not consider the District's alternative argument that the trial court erred by awarding costs of proof because Arnold did not prove the matter covered by the RFA at trial, i.e., it did not release PCE during its business operations at the site. (See § 2033.420, subd. (a) ; Stull v. Sparrow (2001) 92 Cal.App.4th 860, 865-866, 112 Cal.Rptr.2d 239.)
Arnold contends that Hopen could not know whether Arnold used TCE because his employment allegedly started in 1978, two years after TCE's use was restricted. Arnold cites the trial court's statement of decision to support that fact, but the trial court's statement of decision is not evidence. Arnold must provide proper citations to the evidence supporting factual assertions in its briefing. (See Cal. Rules of Court, rule 8.204(a)(1)(C).) Arnold includes a citation to the trial transcript, but it does not appear to correspond to the transcript pagination in the record in this appeal. The pages cited do not relate to the dates of Hopen's employment. In any event, our review of the transcript shows that Hopen testified his "best estimate" was that he started in 1978, but he was not sure of the exact years. Given the equivocal nature of this testimony, and the lack of evidence of the type of restrictions placed on TCE at that time, it would not require a trier of fact to completely discount Hopen's testimony regarding Arnold's use of TCE, especially in light of the substantial TCE contamination found at the site.
Arnold points out that TCE can also biodegrade into 1,1-DCE, but that chemical reaction occurs only under certain limited circumstances. The trial court in its statement of decision following trial accepted that 1,1-DCE contamination would be evidence of a past 1,1,1-TCA release. The District was not unreasonable in crediting Waddell's opinion that 1,1-DCE detections were evidence of a past 1,1,1-TCA release.
"Substantial evidence" is of course a legal term of art, which is not equivalent to a lot of evidence or the preponderance of the evidence. (See Howard v. Owens Corning (1999) 72 Cal.App.4th 621, 631, 85 Cal.Rptr.2d 386.) Instead, it is the minimum showing necessary to sustain a judgment or finding in a party's favor, i.e., evidence and inferences that are reasonable, credible, and of solid value that provide proof of the essential elements that the law requires in a particular case. (Colombo v. BRP US Inc. (2014) 230 Cal.App.4th 1442, 1451-1452, 179 Cal.Rptr.3d 580.)
To the extent the District contends the denial of Arnold's motion for summary judgment justifies its denials of the 1,4 dioxane RFAs, we disagree. Even assuming the denial were relevant, it was not necessarily based on evidence of 1,4-dioxane contamination. It therefore has no bearing on the reasonableness of the District's reliance on its evidence regarding 1,4-dioxane.
See footnote *, ante .