Before: Jacobs, Leval, C.JJ.1 In cross-appeals from the judgment of the United States District Court for the Northern District of New York (Brenda K. Sannes, J.), Plaintiff MPM Silicones, LLC (“MPM”) appeals the grant of partial summary judgment dismissing its claims for recovery of “remedial action” costs under the Comprehensive Environmental Response, Compensation, and Liability Act (“CERCLA”) as barred by the statute of limitations in 42 U.S.C. §9613(g)(2). Defendant Union Carbide Corporation (“UCC”) cross-appeals from the district court’s holding after a bench trial that UCC is liable to MPM for 95 percent of the cost of future “removal action.” Held, the district court’s conclusion that MPM’s claims for recovery of remedial action costs were time-barred relied on an incorrect interpretation of our decision in New York State Electric and Gas Corp. v. FirstEnergy Corp., 766 F.3d 212 (2d Cir. 2014); the district court made no error in adjudicating the allocation of future removal action costs, or in allocating 95 percent against UCC. AFFIRMED in part; VACATED and REMANDED in part. DENNIS JACOBS and PIERRE LEVAL, D.JJ. These are cross-appeals by Plaintiff MPM Silicones, LLC (“MPM”) and Defendant Union Carbide Corporation (“UCC”) from different aspects of the judgment of the United States District Court for the Northern District of New York (Brenda J. Sannes, J.). Plaintiff MPM owns and operates the Sistersville site (“Sistersville” or the “Site”), a manufacturing facility in Friendly, West Virginia. The Site was previously owned and operated by Defendant UCC for many decades. In the course of UCC’s manufacturing operations at the Site in the 1960s and 1970s, UCC generated substantial amounts of toxic polychlorinated biphenyl (“PCB”) waste and buried that waste in various areas of the site. After MPM acquired the Site, it discovered some of this buried PCB waste and undertook a program of clean-up activity in response. MPM brought this suit against UCC under the Comprehensive Environmental Response, Compensation, and Liability Act (“CERCLA”) to recover from UCC the costs it had incurred (and would incur) in cleaning up UCC’s PCB contamination. As explained below, with respect to the timeliness of CERCLA suits to recover such response costs, the governing statute distinguishes between what it identifies as “removal” action, 42 U.S.C. §9601(23), which is generally urgent action taken to deal, at least temporarily, with an immediate health hazard, and “remedial action,” id. §9601(24), which is less urgent action whose objective is to eventually achieve a comprehensive, permanent remedy. See New York State Elec. and Gas Corp. v. FirstEnergy Corp., 766 F.3d 212, 230-31 (2d Cir. 2014). The timeliness of cost recovery suits is governed by very different standards, depending on whether the costs were incurred in “remedial action” or in “removal action.” 42 U.S.C. §9613(g)(2). In this opinion, we use the terms “removal” and “remediation” (or “remedial”) not in their colloquial senses, but to refer respectively to the statutory terms “removal action” and “remedial action.” With respect to MPM’s claims to recover remediation costs, the district court granted summary judgment to UCC on the grounds that, under 42 U.S.C. §9613(g)(2), the suit was time-barred. MPM Silicones, LLC v. Union Carbide Corp., No. 1:11-cv-1542, 2016 WL 3962630, at *14-19 (N.D.N.Y. July 7, 2016) (“MPM I”). On the other hand, as for removal costs, the court held by summary judgment that UCC was liable to MPM, id. at *31, but that MPM was obligated to cover a part of those costs, reserving the precise allocation for trial, MPM Silicones, LLC v. Union Carbide Corp., No. 1:11-cv-01542, Dkt. No. 165 (N.D.N.Y., Dec. 8, 2016) (“MPM II”). After a bench trial, the court allocated 95 percent of future removal costs to UCC with 5 percent to be borne by MPM. MPM Silicones, LLC v. Union Carbide Corp., No. 1:11-cv-1542, 2017 WL 6408611, at *15-21 (N.D.N.Y. Sept. 22, 2017) (“MPM III”). MPM and UCC each challenge aspects of those rulings on this appeal. We hold that the district court relied on invalid reasoning in concluding that MPM’s claim for recovery of the costs of its remediation efforts is time-barred by 42 U.S.C. §9613(g)(2). We therefore vacate that ruling and remand for further consideration of the timeliness of MPM’s suit to recover its remediation costs. As to costs of future removal, we hold that the district court correctly ruled that the issue of cost allocation was ripe for review and did not abuse its discretion in allocating responsibility for 95 percent to UCC. A. BACKGROUND i. UCC’s use of PCBs at the Site The Site consists of 1,300 acres of land in a rural area bordering the Ohio River. The site was undeveloped farmland until UCC acquired it and constructed manufacturing facilities. UCC operated those facilities to manufacture a variety of chemical products from approximately 1955 to April 1993. Today, the site is still used for chemical manufacturing and contains several facilities for the storage, treatment, and disposal of manufacturing waste — known as “Solid Waste Management Units” — including a wastewater treatment unit, an active hazardous waste landfill, and two inactive waste-disposal areas. These are known as the North Inactive Site and the South Inactive Site. During its operation of the Site, UCC used hundreds of thousands of pounds of polychlorinated biphenyls (“PCBs”) in its manufacturing processes. Those processes created substantial PCB-laden chemical wastes. UCC disposed of that waste by burning it, or by depositing it into neutralization tanks or lime pits at the Site. After the lime pits filled with sludge, UCC drained them and dug them out, and used the PCB-laden sludge to backfill other areas of the Site. Some of the PCB-laden waste was deposited in the North Inactive Site, a 5.5 acre landfill located uphill and northwest of the Site’s waste water treatment facility and a creek that runs through the Site, known as Sugar Camp Run. See Dist. Ct. Dkt. No. 77-1 at 48 (map of the Site); see also Dist. Ct. Dkt. No. 80-17 at 82-85 (ENVIRON Map of Site). The precise locations of all the backfill areas are unknown. UCC used PCBs at Sistersville until 1972, when environmental concerns regarding PCB toxicity began to emerge.2 In 1979, after soil sample analysis reported the presence of PCBs in Sugar Camp Run, UCC began to investigate its past waste disposal practices. Clem Schubert, a UCC employee, was tasked with that investigation. Schubert conducted interviews with individuals knowledgeable about UCC’s historical waste disposal practices and learned about UCC’s practice of disposing of PCB-laden wastes in lime ponds and neutralization tanks, and backfilling areas of the Site with sludge from the lime ponds. Schubert identified several areas in the North Inactive Site and several locations south of the wastewater treatment plant where he believed UCC had dumped PCB wastes and further determined that UCC had shifted around previously buried waste during several construction projects. Schubert summarized his findings in a January 1980 memorandum, in which he stated that “it is reasonable to conclude that in the area of the Sistersville Plant site there are buried…heavies with up to 250,000 pounds of PCB[s].” Dist. Ct. Dkt. 238 at 71. That memorandum was circulated to UCC employees, including UCC’s Plant Manager. UCC conducted several rounds of testing for various contaminants (including PCBs) at the Site between 1979 and 1981. While some tests showed low levels of PCBs in groundwater, the results overall were inconsistent and revealed only small amounts of PCBs.3 In 1984, a UCC memorandum regarding contamination at the North Inactive Site recognized that “[a]lthough no definitive evidence can be found, it is possible that up to 250,000 pounds of PCB[s] are buried at this site…However, wells at this site when last monitored in 1981 showed no significant leaching or contamination and no PCB[s] were found.” Dist. Ct. Dkt. 78-10 at 1. A 1992 memorandum summarizing all prior UCC investigations into PCBs at the Site — including Schubert’s 1980 memorandum — concluded that “it is not possible to state unequivocally that PCB[s] were not placed in the north inactive site” but that “[i]nformation suggesting disposal is purely speculative” because “[m]onitoring data to date do not substantiate the speculation.” App’x 382-83. At various times during the 1980s and 1990s, UCC employees speculated that the low levels of PCBs detected by UCC’s site monitoring tests were the result of spillage from PCB-containing electrical equipment that UCC had used at the site. In 1981, as required by the recently enacted Resource Conservation and Recovery Act (“RCRA”), see 42 U.S.C. §§6901-6992k, UCC submitted an application to the Environmental Protection Agency (“EPA”) for a permit to treat, store, or destroy hazardous waste. RCRA established a framework to “ensure the proper treatment, storage, and disposal of [hazardous] waste.” Prisco v. A&D Carting Corp., 168 F.3d 593, 608 (2d Cir. 1999) (citation omitted). It requires the owner or operator of any “facility that treats, stores, or disposes of hazardous wastes” to “seek[] and obtain[] a permit from the [EPA].” Owen Elec. Steel Co. of S.C., Inc. v. Browner, 37 F.3d 146, 147 (4th Cir. 1994); see also 42 U.S.C. §6925(a). The statute requires the EPA to evaluate the environmental risks of any such facility and to issue permits requiring the owner or operator of a facility to undertake “corrective action for all releases of hazardous waste…from any solid waste management unit.” 42 U.S.C. §6924(u). UCC’s permit application required its disclosure of any “hazardous substances” that had been “stored, treated, or disposed of” at the Site. See Hazardous Substances: Notification of Treatment, Storage and Disposal Facilities, 46 Fed. Reg. 22144, 22145 (Apr. 15, 1981); see also 42 U.S.C. §9603(c) (requiring disclosure of certain facility owners of the “stor[age], treat[ment], or dispo[sal]” of any “ hazardous substances”).4 While UCC indicated on the application that it was aware of several types of hazardous waste at the Site, it did not disclose its prior use of PCBs, leaving a check box for “PCBs” unchecked.5 In its written correspondence with the EPA regarding its RCRA permit for Sistersville, UCC made no mention of its prior use of PCBs or its dumping of PCB-laden waste. In response to the EPA’s 1985 request for information about all solid waste management units at the Site that had “the potential to release hazardous wastes or hazardous constituents to the environment,” Sp. App’x 87, UCC disclosed general information about some potential contaminants at the North and South Inactive Sites but did not mention PCBs.6 In response to the EPA’s 1986 further request for information regarding the solid waste management units, UCC disclosed the results of its site testing, which showed that, if any PCBs were present in the tested samples, that presence was below detection limits. UCC did not communicate its history of PCB use and disposal at Sistersville. EPA representatives visited the Site on May 15, 1986, and while the parties dispute what happened during that visit, Fred Dailey, a UCC employee, wrote a memorandum in 1987 — one year after the visit — stating that, during the visit, he informed the EPA orally that “it was speculated that as much as 500,000 lbs of contaminated PCB heat fluid was generated during the plant’s previous activities,” that “its disposition was never determined,” and that “it could have been placed in an inactive site.” Sp. App’x 88. Dailey’s memo then stated that he told the EPA representatives that UCC had conducted groundwater testing and examination of the inactive sites at Sistersville, and that “[a]ll the results of the tests and analyses conducted showed no PCBs.” Id. According to Dailey’s 1987 memorandum, the EPA representatives did not further inquire about PCBs during the meeting. In 1988, the West Virginia Division of Natural Resources (“WVDNR”) — which was authorized by the EPA to administer certain provisions of RCRA — issued UCC a permit covering Sistersville’s active hazardous waste management units. That permit included a “Remedial Action Plan” requiring UCC to treat the groundwater near a copper sludge removal pond at Sistersville. In 1991, UCC installed a recovery well, which was part of a larger project to close the sludge pond. Also in 1988, the EPA issued a RCRA permit to UCC for Sistersville. The EPA determined that eight of the solid waste management units at the Site — including the North Inactive Site and the wastewater treatment system — required further investigation and required UCC to submit a workplan to remediate hazardous waste issues at each unit. The EPA’s RCRA permit did not mention PCBs. UCC submitted a workplan for the North Inactive Site in 1992, which concluded that the area was a source of contamination and recommended construction of an earthen cap to reduce surface water, sediment, and groundwater contamination. UCC’s 1992 submission to the EPA did not mention PCBs because of what UCC later referred to as “an inadvertent failure to perform PCB analysis.” Dist. Ct. Dkt. 89-1 at 25 62. The EPA accepted UCC’s proposal, and UCC undertook construction of the earthen cap in the summer and fall of 1992.7 The same year, UCC also constructed a diversion ditch to intercept surface water from the hillside and divert it away from the disposal area. In 1993, UCC wrote a letter to the EPA advising that it inadvertently failed to disclose PCBs in its 1992 report and disclosing sampling results from Sugar Camp Run showing low levels of PCBs. The EPA did not require any further action in response to this additional disclosure. None of the corrective actions undertaken by UCC were designed to investigate or address the possibility that large amounts of PCB-laden wastes were buried at Sistersville. Instead, the construction of the earthen cap and the diversion ditch for the North Inactive Site were designed to address the presence of other contaminants in groundwater and soil in various parts of the north end of the Site — contaminants UCC believed to have originated at the North Inactive Site.8 As more recent events make clear, a full investigation into UCC’s historical PCB waste dumping would have required a comprehensive sampling program in the vicinity of the North Inactive Site and other potential dumping sites and in Sugar Camp Run. UCC did not conduct such a study, nor did the EPA require it to.9 Documents produced in the 1980s and 1990s show that UCC’s position was that any PCBs found during site monitoring were likely the result of spillage from its PCB-containing electrical equipment. UCC removed all PCB-containing electrical equipment from the Site by mid-1991, and subsequently declared that the Site had attained “PCB ‘free’ status.” Sp. App’x 90. i. Subsequent ownership of the Site In 1993, UCC entered into an agreement to sell the Site to OSi Specialties, Inc. (“OSi”). Prior to closing, OSi retained Environment Strategies Corporation (“ESC”) to conduct environmental due diligence. ESC requested information from UCC about materials disposed of at the North Inactive Site, and UCC directed ESC to the 1992 workplan it had submitted to the EPA, which purported to detail the known contaminants at the site. ESC issued a report which included five paragraphs on PCBs, three of which address PCB-containing electrical equipment. The other two paragraphs discuss soil samples collected in 1992 and 1993 which showed the presence of PCBs, but not at a level requiring remediation under EPA guidelines. After the sale, the RCRA permit for the Site transferred to OSi. In 1994, OSi submitted to the EPA a Corrective Measures/Stabilization Proposal for the North Inactive Site, which summarized the corrective actions that had previously been taken to control surface water infiltration and runoff from the North Inactive Site — including the earthen cap and diversion ditch UCC had built in 1992 — and proposing the construction of an interceptor trench to divert water away from the North Inactive Site and continued monitoring and maintenance of the site. EPA approved the proposal in June 1994, and OSi constructed the interceptor trench in the fall of 1994. In 1997, EPA modified the Site’s permits to recognize that it was satisfied with the measures being taken to remedy contamination at the North Inactive Site and to require continued inspection and maintenance of the earthen cap and the diversion ditches. In 2003, OSi, which had become Crompton Corporation, sold the Site to an affiliate of General Electric, Inc., (“GE”), then named GE Advanced Materials, which would later become Plaintiff MPM Silicones. The purchase agreement included indemnity provisions relating to environmental contamination at the Site. Prior to the sale to GE, OSi/Crompton again retained ESC to perform an environmental site assessment. After conducting site visits and interviews, ESC issued a report stating that the Site had obtained “PCB free” status following the removal of all PCB-containing electrical equipment in 1991. Sp. App’x 93. The report also acknowledged a previous UCC memo recognizing the possibility that PCBs had been disposed of at the North Inactive Site but noting that UCC’s site monitoring could not substantiate the existence of buried PCB wastes. GE hired ENVIRON, an environmental consulting firm, to conduct a review of the site prior to the sale. John Wood, a GE employee who supervised ENVIRON’s environmental due diligence, testified (at his deposition during the proceedings in the district court) that UCC employees told him about the investigation in the 1980s that had raised the possibility of PCB disposal at the site, but had assured him that UCC’s site monitoring could not substantiate that theory. Wood also testified that UCC employees had told him that UCC had used PCB-containing electrical equipment at the Site — which had since been removed — and that they believed that “the source of PCBs…found at the site was leaks from or maintenance [of that electrical equipment].” Dist. Ct. Dkt. 89-9 at 5-6. Wood further testified that UCC did not reveal “anything in [the] documents or [the] interview process that would have led [GE] to know that PCBs were used in the manufacturing process,” and that, had he and his team known about this history, they would have “tailored the investigation and sampling program in a way that we would have collected more samples, specifically trying to address that issue or that area of concern.” Id. ENVIRON conducted extensive soil and groundwater testing at the Site in May 2003. ENVIRON’s final report, issued in May 2004 (after the sale of the property closed in July 2003), noted that PCBs had been found in concentrations exceeding federal regulatory thresholds in groundwater and soil samples from various areas in the north end of Sistersville — including Sugar Camp Run — and that these positive PCB results were “unexpected.” Dist. Ct. Dkt. 80-17 at 74.10 The report explained that prior reports of PCB-laden waste dumping at the Site had been dismissed based on a 1992 UCC memorandum determining — without explanation — that positive PCB results had been “determined not to be valid,” and based on statements from UCC representatives that UCC’s “distillation process would not produce PCBs.” Id. at 11-12. The report, however, further stated: Although Facility representatives are no longer concerned about PCBs in the North Inactive Area, unexpected levels of PCBs were found in some Facility soils and selected ground water samples during the Phase II investigation. As there is no obvious source for these PCBs, resampling of ground water, and further investigation of soil, is suggested. Id.; see also id. at 74. In July 2004, GE Advanced Materials sent the May 2004 ENVIRON report to OSi/Crompton and asserted that under the indemnity provisions in the purchase agreement OSi/Crompton was responsible for the costs associated with the necessary investigation and remediation of PCBs at the Site. OSi/Crompton disclaimed financial responsibility, but nonetheless engaged ESC to investigate PCB contamination in two areas of the Site: the Waste Incineration Area and the Permitted Hazardous Storage Waste Area — both of which are in close proximity to the North Inactive Site. ESC issued a final report on May 30, 2006, concluding that PCB levels in certain areas of the site exceeded federal regulatory levels, and recommended “a second phase of soil characterization to delineate the vertical and horizontal extent of PCBs in soil” in this area. Dist. Ct. Dkt. 80-7 at 13. On August 28, 2007, WSP Environmental Strategies delivered a plan for additional sampling, which GE Advanced Materials (now named MPM Silicones, LLC) did not implement. ii. MPM discovers buried PCBs In June 2008, during construction of a project to upgrade the site’s wastewater treatment unit, MPM uncovered discolored soils which, when analyzed, were revealed to be contaminated with PCBs. MPM suspended its construction project and spent $123,195.05 in response to its discovery. This included $30,385.18 disposing of the contaminated materials it had uncovered; $8,714.71 decontaminating rented construction equipment; and $84,095.16 on steel shoring it needed to prevent the excavated area from caving in. In August 2008, MPM contacted OSi/Crompton (then known as Chemtura) informing it of the recently discovered contamination and again asserting that OSi/Crompton was responsible for the costs of remediation. OSi/Crompton again disclaimed responsibility under the indemnity provisions of the purchase agreement. In June 2009, MPM hired Clean Harbors, an environmental consulting firm, to conduct further sampling in the wastewater treatment unit area. Clean Harbors uncovered additional contamination and reported that it could not yet determine the geographical extent of the contamination. MPM incurred costs of $251,345.20 in connection with Clean Harbors’ sampling activities, bringing its total outlays in response to its discovery of PCBs to $374,540.25. After receiving the results from Clean Harbors, MPM searched the Site’s records and found UCC documents dating back to 1970 that indicated that UCC had purchased and used “significant quantities” of PCBs at the Site. Sp. App’x 105. MPM has not conducted any further soil sampling or any further construction work on its construction of the wastewater treatment upgrade since Clean Harbor concluded its sampling in September 2009. MPM, however, plans to resume the wastewater treatment upgrade by installing a pump station in the location where it encountered PCBs in 2008, which will require MPM to excavate and dispose of the PCB-containing soils. MPM contacted UCC regarding the PCB contamination in September 2010, and the parties entered into a tolling agreement which tolled the statute of limitations from May 19, 2011 to December 30, 2011. MPM reported its discovery of the PCB contamination to the West Virginia Department of Environmental Protection (“WVDEP”) in March 2012. As of September 2017, neither WVDEP nor the EPA had required any further remediation of the PCB contamination.11 iii. Procedural History MPM filed this suit on December 30, 2011 seeking, inter alia, reimbursement under CERCLA from UCC for $374,540.25 in past cleanup costs, and a declaratory judgment of UCC’s liability for any future cleanup costs required to deal with contamination at Sistersville.12 UCC brought crossclaims seeking contribution under CERCLA, arguing that MPM should bear a portion of the response costs.13 Like RCRA, CERCLA was enacted to “address the risks associated with the improper storage and disposal of hazardous and toxic substances.” Schaefer v. Town of Victor, 457 F.3d 188, 190 (2d Cir. 2006). CERCLA’s “dual goals [are] cleaning up hazardous waste and holding polluters responsible for their actions.” New York v. Next Millennium Realty, LLC, 732 F.3d 117, 124 (2d Cir. 2013).14 Among other measures, CERCLA “authoriz[es] private parties to pursue contribution or indemnification from potentially responsible parties for expenses incurred responding to environmental threats.” Commander Oil Corp. v. Barlo Equip. Corp., 215 F.3d 321, 326 (2d Cir. 2000).15 Section 107(a) of CERCLA provides a cause of action for private parties to seek recovery of costs from a “potentially responsible party”16 of “any [] necessary costs of response” to a release of hazardous substances. Schaefer, 457 F.3d at 195; see also 42 U.S.C. §9607(a). Section 113(g)(2) permits a party to seek a declaratory judgment on a potentially responsible party’s liability for any necessary future response costs. 42 U.S.C. §9613(g)(2). Section 113(f)(1) permits a party which has been sued for cost reimbursement under Section 107 of CERCLA to sue other potentially responsible parties for contribution. See id. §9613(f)(1); see also Cooper Indus., Inc. v. Aviall Servs., Inc., 543 U.S. 157, 162-63 (2004). CERCLA’s cost-recovery provision distinguishes between two kinds of response projects. “Remedial actions” are “generally long-term or permanent containment or disposal programs,” State of New York v. Shore Realty Corp., 759 F.2d 1032, 1040 (2d Cir. 1985), which are taken “in the event of a release or threatened release of a hazardous substance into the environment” and are “consistent with [a] permanent remedy.” 42 U.S.C. §9601(24) (defining “remedy” or “remedial action”). “Removal actions,” by contrast, are “typically short-term cleanup arrangements,” Shore Realty, 759 F.2d at 1040, which respond to immediate threats to the environment. 42 U.S.C. §9601(23) (defining “removal”); New York State Elec. and Gas Corp. v. FirstEnergy Corp., 766 F.3d 212, 230 (2d Cir. 2014) (“NYSEG”) (“Removal actions are generally clean-up measures taken in response to immediate threats to public health and safety.”). Section 113(g)(2) of CERCLA provides different statutes of limitations for lawsuits brought to recover costs for removals and remediations. See 42 U.S.C. §9613(g)(2). A cost recovery suit for a removal must be brought “within 3 years after completion of the removal action,” while a cost recovery suit for a remediation must be brought “within 6 years after initiation of physical on-site construction of the remedial action.” Id. The parties cross-moved for summary judgment. UCC argued, inter alia, that MPM’s Section 107 claim for reimbursement of past remediation costs (and its Section 113(g)(2) claim for a declaratory judgment on UCC’s liability for future remediation costs) were time-barred because previous corrective actions taken at Sistersville — specifically, UCC’s installation of a recovery well in 1991 and its construction of the earthen cap and diversion ditch in 1992, and OSi/Crompton’s construction of the interceptor trench in 1994 — were remediations, so that the commencement of construction of these measures in the early 1990s triggered the six-year statute of limitations for remediations. iv. The district court’s July 2016 order The district court’s July 7, 2016 order concluded that UCC’s actions in the early 1990s (and OSi’s construction of the interceptor trench in 1994) were remediation (rather than removal), based in part on the fact that the EPA had incorporated these measures into the “final remedy” for dealing with contamination at Sistersville. See MPM I, 2016 WL 3962630, at *11-14. (MPM challenges this conclusion on appeal, arguing that “UCC’s activities were more analogous to removal action than to remedial action.” MPM Br. at 46-50.) The district court further found that MPM’s claims for reimbursement of costs of its past remediations (under Section 107(a)) and its request for a declaratory judgment on UCC’s liability for future remediation costs (under Section 113(g)(2)) were time-barred. MPM I, 2016 WL 3962630, at *14-19. In so holding, the district court relied on a statement in our decision in NYSEG that “there can only be one remedial action at any given site.” 766 F.3d at 235. The district court characterized the multiple remediation projects — i.e., (i) UCC and OSi’s corrective actions in the early-1990s, (ii) any efforts undertaken by MPM to dispose of the PCB-contaminated soil in 2008 that were remedial in nature, and (iii) any future remediation efforts MPM would undertake to decontaminate the Site — as part of the Site’s single remediation and held that “the statute of limitations [to recover costs for that remediation] began to run when [UCC] commenced physical on-site construction of the remedial action [in the early 1990s].” MPM I, 2016 WL 3962630, at *14. Because the six-year statute of limitations began to run in the early-1990s, the court held that the limitations “had expired long before the parties’ May 19, 2011 tolling agreement,” and that MPM’s claims were time-barred under Section 113(g)(2)(B). Id.17 (MPM’s principal argument on appeal is that the district court erred in holding that these claims were barred by the statute of limitations.) On that basis, the district court granted UCC’s motion for summary judgment to the extent that MPM’s claims sought recovery of costs related to any remediation at the Site. Id. at *35. The district court, however, held that UCC was liable as a matter of law for past and future costs of removal that MPM had undertaken or would undertake at Sistersville. Id.18 Accordingly, the court entered a declaratory judgment as to UCC’s liability for the costs of future “removal actions” at Sistersville. Id. at *31.19 In a later order, the district court granted UCC’s motion for summary judgment that MPM was liable under Section 113(f)(1) in contribution for an equitable share of the response costs — leaving open the issue of cost allocation between MPM and UCC.20 The parties subsequently reached a settlement of MPM’s claim for past removal costs. After the settlement, the only remaining issue was the allocation of financial responsibility for any future removal costs. v. The district court’s September 2017 order From January 31, 2017 to February 2, 2017, the district court held a bench trial on the remaining issues: (a) whether future response costs should be allocated and (b) what the parties’ equitable shares of those response costs should be. In a September 22, 2017 order, the district court held that the issue of allocation of costs was ripe for adjudication because MPM had established that “it is likely to incur future costs in responding to PCBs at the Site because MPM plans to upgrade its wastewater treatment facilities” which will require that “MPM respond appropriately to PCBs in the soils.” MPM III, 2017 WL 6408611, at *15-18. UCC argues on appeal that the district court erred in so holding because the possibility of MPM incurring future removal costs is too remote. In considering how to allocate future removal costs, the district court noted that “UCC was the only entity to bring PCBs onto the Site” and “benefitted from disposal of the [PCB] waste” without having to engage in any corrective action (factors which weighed in favor of allocating costs to UCC), but concluded that there was insufficient evidence to conclude that “UCC’s failure to disclose PCBs [to regulators] was driven by an intention to conceal information.” MPM III, 2017 WL 6408611, at *19. The court also rejected UCC’s argument that MPM should bear a majority of future removal costs because MPM “had notice of the PCBs and assumed the risk of future PCB-related response costs when it acquired the Site,” instead crediting Wood’s testimony that, during the due diligence process, UCC employees led him to believe that “any PCB contamination was due to PCB electrical equipment.” Id. at *20. Further, the court rejected UCC’s argument that the indemnity agreement between MPM and Crompton warranted allocating responsibility to MPM, finding that “MPM has not been indemnified, nor is there any reason to conclude that it will be” because “Crompton has disclaimed responsibility under the indemnity agreement.” Id. at *21. The court also rejected the argument that MPM should bear the majority of the responsibility because it would “derive commercial benefit from any future removal action when it upgrades the wastewater treatment unit,” holding that “there is no evidence that MPM would realize a meaningful increase in property value as a result of a PCB cleanup.” Id. The court, however, faulted MPM for failing to contact government regulators after discovering PCB-contaminated soil in 2008, and for waiting until 2012 to inform WVDEP of this development, noting that the Site’s RCRA permit required MPM to notify regulators of any “relevant facts” regarding newly-discovered contamination within seven days of discovery. Id. at *20-21. Accordingly, the court allocated five percent of future removal costs to MPM and allocated the remaining ninety-five percent of future removal costs to UCC. Id. at *1.21 (On appeal, UCC argues that the district court abused its discretion in weighing the relevant allocation factors, and that the district court should have allocated more of the cost burden to MPM.) B. DISCUSSION MPM argues in this appeal that the district court erred in concluding that MPM’s claims for reimbursement of remediation costs are time-barred by §113(g)(2)(B). UCC contends that the district court erred in determining that future allocation of removal costs was ripe for review, and further erred in allocating 95 percent of those costs against it. i. Statute of Limitations MPM directs its arguments against two aspects of the district court’s finding of untimeliness: (1) the court’s conclusion that UCC’s prior response projects in the 1990s were in the nature of remediation, rather than removal; and (2) its conclusion that, if UCC’s prior response projects constituted remediation, the six years allowed by §113(g)(2)(B) for recovery of the costs of any further remediation at the Site began to run upon UCC’s initiation of its remediation in 1992 and expired in 1998, long before MPM purchased the Site. The district court reached the latter conclusion based on its understanding of our NYSEG opinion to mean that, once any remediation is undertaken at a site, all future remedial activity at the site, regardless of circumstances, is deemed to be part of the same initial remediation, so that, under §113(g)(2)(B), the opportunity for timely suit to recover the costs of future remediation at the site necessarily expires, absent tolling, six years from the start of the initial remediation. We agree with the district court that UCC’s corrective actions undertaken in the 1990s were remediation, but we do not understand NYSEG to mean that, for purposes of determining the timeliness of a cost recovery action, all remediation activity at a site regardless of circumstances is deemed to be part of a single remediation, so that the six year limitations period necessarily begins to run at the start of the first remedial activity. Accordingly, we believe the district court’s conclusion that MPM’s suit was untimely was based on an inaccurate premise. We do not decide whether MPM’s suit to recovery remediation costs was in fact timely, but instead vacate the district court’s decision and remand for reconsideration consistent with this opinion. The distinction between “removal” actions and “remedial action[s]” arises from the first iteration of the CERCLA statute enacted in 1980. See Pub. L. 96-510, §101(a)(23)-(24); 94 Stat. 2767, 2770-71 (1980) (defining “ remove” or “removal” and “remedy” or “remedial action”). The statute defines “remove” or “removal” as: the cleanup or removal of released hazardous substances from the environment, such actions as may be necessary [] in the event of the threat of release of hazardous substances into the environment, such actions as may be necessary to monitor, assess, and evaluate the release or threat of release of hazardous substances, the disposal of removed material, or the taking of such other actions as may be necessary to prevent, minimize, or mitigate damages to the public health or welfare or to the environment, which may otherwise result from a release or threat of release. Further, the statute provides several examples of “removal” actions, including: security fencing or other measures to limit access, provision of alternative water supplies, temporary evacuation and housing of threatened individuals not otherwise provided for. 42 U.S.C. §9601(23). The statute defines “remedy” or “remedial action” as: those actions consistent with permanent remedy taken instead of or in addition to removal actions in the event of a release or threatened release of a hazardous substance into the environment, to prevent or minimize the release of hazardous substances so that they do not migrate to cause substantial danger to present or future public health or welfare or the environment. The statute also provides an illustrative list of “remedial action[s],” including: such actions at the location of the release as storage, confinement, perimeter protection using dikes, trenches, or ditches, clay cover, neutralization, cleanup of released hazardous substances and associated contaminated materials, recycling or reuse, diversion, destruction, segregation of reactive wastes, dredging or excavations, repair or replacement of leaking containers, collection of leachate and runoff, onsite treatment or incineration, provision of alternative water supplies, and any monitoring reasonably required to assure that such actions protect the public health and welfare and the environment. The term [also] includes the costs of permanent relocation of residents and businesses and community facilities [where] such relocation is more cost-effective [than other remedial action]…[and also] includes offsite transport and offsite storage, treatment, destruction, or secure disposition of hazardous substances and associated contaminated materials. Id. §9601(24). The statutory definitions do not provide clear insight as to the boundary between removals and remediations. The definitions of each type of action overlap substantially: certain corrective actions — like covering contaminated soil or diverting water away from contaminated areas with drainage controls,22 the provision of alternative water supplies to replace contaminated water,23 and related monitoring activities24 — may be classified as either “removal” or “remedial” actions. Over several decades of CERCLA litigation, courts have agreed on a general principle to distinguish the two: “[r]emoval actions are generally clean-up measures taken in response to immediate threats to public health and safety” that “address contamination at its endpoint,” while “[r]emedial actions are typically actions designed to permanently remediate hazardous waste” that address contamination at its source. NYSEG, 766 F.3d at 230-31. The key distinction between the two terms is immediacy and comprehensiveness. United States v. W.R. Grace & Co., 429 F.3d 1224, 1244 (9th Cir. 2005). Removal actions are undertaken to deal with “threat[s] to human health or the environment which require[] an immediate response.” Id. (citations omitted).25 Because removal actions are generally targeted at reducing the deleterious effects of contamination — e.g. the removal from downstream water supply wells of contaminants that have migrated from a nearby landfill, see Next Millenium, 732 F.3d at 126-27 — they are generally not so ambitious as to seek to eliminate or permanently contain the source of contamination. They do not generally aim to serve as permanent solutions. See, e.g., NYSEG, 766 F.3d at 231-33 (characterizing efforts to “clean[] up coal tar that had already migrated from [the source of contamination]” and to “ remove [coal tar] from [a] river” as “removal actions” because those efforts were “not designed to clean up contamination at the source”); id. at 233 (“[M]easures taken to minimize and mitigate contamination, but not to permanently eliminate it, are properly classified as removal actions.”) (citation omitted). Removals are often planned and executed relatively quickly in order to immediately abate public health hazards, such as contaminated drinking water.26 Accordingly, removals are often undertaken to secure prompt relief from a danger even though the action is not deemed a step toward permanent elimination of the threat. See, e.g., Colorado v. Sunoco, Inc., 337 F.3d 1233, 1244 (10th Cir. 2003) (holding that the plugging of a mine was a “removal” action in part because, when the plug was installed, “it was uncertain whether the plugging would achieve the goal of eliminating [contamination]“).27 In essence, because removals “can be initiated promptly after notification of a threat,” they provide “flexibility to tailor prompt and effective responses to immediate threats to human health and the environment.” W.R. Grace, 429 F.3d at 1226, 1240. Remediations, by contrast, include only actions “consistent with [a] permanent remedy.” 42 U.S.C. §9601(24). Remediations are undertaken to permanently remediate contamination, generally after months (if not years) of correspondence with regulators, soil testing, data collection, safety and quality control analysis, and feasibility studies. See 40 C.F.R. §300.430(a)(2) (describing the process of compiling a remedial investigation/feasibility study, which “includes the following activities: project scoping, data collection, risk assessment, treatability studies, and analysis of alternatives”). See generally id. §300.430 (describing the process a regulatory agency must undertake before approving or implementing a remedial action program). Unlike removals, remediations generally presuppose full disclosure has been made to the regulator of the “scope and complexity of the site problems being addressed,” id. §300.430(a)(ii)(C); see also 42 U.S.C. §9603(c), so that the remediation may serve as a “final, once-and-for-all cleanup of a site” designed to cost-effectively remediate the full scope of those known problems. NYSEG, 766 F.3d at 236. Unlike removals, remediations typically address “the underlying source of the contamination.” Id. at 233 (citation omitted) (characterizing a 1980s cleanup to prevent coal tar that had migrated from its source from “further migrating into [a] river” — taken as “an immediate response to a health concern” about water contamination — as a removal action, but characterizing later work to excavate the source of the coal tar contamination as a remediation because it was “designed to remediate the pollution at its source”). Moreover, CERCLA and its implementing regulations provide a pathway to “transition from a removal to a remedial action if it [is] ‘determine[d] that the removal action will not fully address the threat posed by the release.’” W.R. Grace, 429 F.3d at 1242-43 (quoting 40 C.F.R. §300.415(g)). In other words, removal actions initially undertaken to respond to immediate threats may be adopted as part of a permanent remedial solution. See, e.g. Next Millenium, 732 F.3d at 128-29 (noting that two removal measures “were ultimately adopted as part of a permanent remedial solution”).28 As passed in 1980, CERCLA did not contain a statute of limitations applicable to cost recovery actions under Section 107. See Merry v. Westinghouse Elec. Corp., 684 F. Supp. 852, 856 (M.D. Pa. 1988); United States v. Mottolo, 605 F. Supp. 898, 909-10 (D.N.H. 1985).29 In 1986, Congress passed the Superfund Amendments and Reauthorization Act of 1986 (“SARA”), Pub. L. 99-499, 100 Stat. 1613, which implemented separate statutes of limitations for removals and remediations. See 42 U.S.C. §9613(g)(2).30 Section 113(g)(2) of CERCLA, as amended by SARA, now provides: An initial action for recovery of the costs referred to in [Section 107] must be commenced — (A) for a removal action, within 3 years after completion of the removal action… (B) for a remedial action, within 6 years after initiation of physical on-site construction of the remedial action…. 42 U.S.C. §9613(g)(2). The legislative history does not explain why Congress took a different approach to limitation periods as between removals and remediations.31 a. UCC’s construction of the earthen cap and the diversion ditch in 1992 were remedial MPM argues on appeal that UCC’s construction of an earthen cap and diversion ditch for the North Inactive Site (and OSi/Crompton’s construction of a different interceptor trench in 1994) were removals rather than remediations because those actions are similar to actions that this court held in Next Millenium and NYSEG to be removals.32 In Next Millenium, however, we classified two measures — a “granulated activated carbon adsorption system” and an “air stripper tower,” both designed to remove volatile organic compounds from water supply wells in order to decontaminate drinking water, 732 F.3d at 122 — as removals in part because “both systems were installed in response to an imminent public health hazard,” which we noted was “a defining characteristic of removal actions.” Id. at 126. Moreover, the Next Millenium court noted that both measures “were designed…to address water contamination at the endpoint — the wells — and not to permanently remediate the problem by ‘prevent[ing] or minimiz[ing] the release of hazardous substances so that they do not migrate’ from the underlying source of contamination.” Id. at 127 (quoting 42 U.S.C. §9601(24)). The measures were, in other words, a “respon[se] to a water-supply problem, not an environmental cleanup concern” because they dealt with mitigating the effects of contamination, rather than eliminating or containing the source of contamination (which, in Next Millenium, was contamination at a nearby hazardous waste site). Id. In contrast, each of the measures taken by UCC and OSi/Crompton to deal with the North Inactive Site in the 1990s were clearly remedial in nature. Neither the 1992 earthen cap, nor the 1992 diversion ditch, nor the 1994 interceptor trench were efforts to deal with any “imminent” hazard or threat to public safety by neutralizing contamination “at its endpoint.” See NYSEG, 766 F.3d at 231. There is no suggestion in the record that the contamination at Sistersville had caused any kind of time-sensitive threat that “required an immediate response.” W.R. Grace & Co., 429 F.3d at 1244 (citation omitted); see id. at 1245 (concluding that “removal actions encompass interim, partial time-sensitive responses taken to counter serious threats to public health”). Rather, these actions were taken as steps to permanently prevent contaminants known to be buried at the North Inactive Site from migrating away from their source — i.e., the location of their burial. Such “permanent containment” measures are remedial in nature. Schaefer, 457 F.3d at 195; id. at 204 (holding that the construction of an earthen cover to close a landfill was a remediation); see also 42 U.S.C. §9601(24) (defining “remedial action[s]” as those taken “ to prevent or minimize the release of hazardous substances so that they do not migrate,” and listing “trenches,” “ditches” and “clay cover” as examples of “remedial action[s]“); W.R. Grace & Co., 429 F.3d at 1238-39 (noting that the definitions of “remedial action” and “removal” overlap, but that the “definition of ‘remedial action’…can be distinguished from ‘removal’ because it refers to ‘permanent’ remedies…[while] ‘removal’ is focused on temporary and emergency activities”).33 We conclude that the district court did not err in holding that UCC’s construction of the earthen cap and the diversion ditch in 1992, and OSi/Crompton’s construction of the interceptor trench in 1994, qualified as remediations under CERCLA. b. NYSEG’s single-remediation principle was not intended to govern all circumstances The district court concluded that MPM’s claims for reimbursement of the costs of its remediation are time-barred by reason of UCC’s prior remediation, which began more than six years before MPM brought suit. The district court believed this conclusion was compelled by a statement in our NYSEG opinion, which signed on to a proposition asserted by the Tenth Circuit (and some other courts) that “there can only be one remedial action at a site.” 766 F.3d at 236.34 We believe for reasons explained below that this misinterpreted our NYSEG ruling. The NYSEG opinion, as relevant here, considered whether the plaintiffs’ cost-recovery claims, brought in 2003, were barred by 42 U.S.C. §9613(g)(2), the same statute of limitations provision that applies here. NYSEG, 766 F.3d at 221. The plaintiff was the operator of two sites, Norwich and Owego, which had been contaminated by the defendant, a prior operator of the site.35 At the Norwich site, the plaintiff had undertaken a three-phase “Interim Remedial Measure,” planned in consultation with the governmental regulatory agency to take place in 1993 (Phase I), 1996 (Phase II), and April 1997 (Phase III). Id. at 234. The 2003 suit sought to recover the costs of Phase III only, presumably because a suit to recover the costs of Phases I and II would clearly have been barred as more than six years had elapsed from the beginning of those phases. Id.; 42 U.S.C. §9613(g)(2); see also New York State Elec. & Gas Corp. v. FirstEnergy Corp., 808 F. Supp. 2d 417, 427 (N.D.N.Y. 2011) (“NYSEG Dist. Ct. Op.”). The defendant contended the suit was barred, even as to the 1997 phase, because the six years allowed began to run at the start of the remediation in 1993. The plaintiff argued that its suit was not barred because the 1997 phase of the cleanup was a discrete project, distinct from the prior phases, so that its inception in 1997 started a new allowable six years for a cost-recovery suit.36 NYSEG, 766 F.3d at 234. Rejecting that argument, this court affirmed the district court’s finding that the three-phase cleanup was “one remedial cleanup,” relying, in part on the plaintiff’s having stipulated that the three stages of the project were a “single action comprised of three phases.” Id. At the one-acre Owego site, the plaintiff undertook to eliminate coal tar contamination between September 1994 and 1995 (the “1994 response”). Id. at 235. After discovering additional coal tar contamination and the pipe that was the source of that contamination, the plaintiff undertook additional response activity in 2003 (the “2003 response”). See NYSEG Dist. Ct. Op, 808 F. Supp. 2d at 479; NYSEG, 766 F.3d at 235. The defendant argued that work done in the 2003 response was a “continuation of the remedial work begun in 1994,” NYSEG, 766 F.3d at 235, so that the 2003 suit was time-barred by the passage of six years. See 42 U.S.C. §9613(g)(2). The district court agreed that the claim was time-barred, reasoning that the 1994 response and the 2003 response were part of the same “remedial” project, in part because both actions addressed “the same source and constituent contamination.” NYSEG Dist. Ct. Op., 808 F. Supp. 2d at 511.37 On review, this court affirmed the district court’s finding of untimeliness. NYSEG, 766 F.3d at 235-36. We rejected the plaintiff’s argument that the work done in 2003 was a separate and distinct remediation from the remedial work done in 1994, see id. at 235; see also Br. for Plaintiff at 45-47 in NYSEG, 766 F.3d 212 (No. 11-4143) (2d Cir. 2014), characterizing it rather as a continuation of the earlier project. What we have described above fully explained and justified the NYSEG court’s conclusion that the suit was time-barred. As further support, however, the NYSEG court cited with approval the Tenth Circuit’s statement in Colorado v. Sunoco that the CERCLA statute envisages only one “‘remedial action’ per site.” 337 F.3d at 1242. The district court in the case now before us understood that proposition as categorically applicable in all circumstances. MPM I, 2016 WL 3962630, at *14. For reasons explained below, we believe the NYSEG opinion cannot have intended this proposition to govern every different remediation circumstance. The proposition, although altogether reasonable in the conventional circumstances to which it was applied in NYSEG, Sunoco, and other cases, cannot have been intended to apply in very different circumstances, in which application of that proposition would make little sense. In NYSEG, as reviewed above, the plaintiff had undertaken a series of remedial steps directed towards remediation of a contamination problem perceived at the outset; the subsequent steps were either explicitly foreseen at the start of the remediation (Norwich), see NYSEG, 766 F.3d at 234, or at least contemplated (Owego), id. at 235; see also NYSEG Dist. Ct. Op., 808 F. Supp. 2d at 478 (noting that the regulatory agency had defined the Susquehanna River, the site of the subsequent response project, as an operational unit in March 1994, and that the operator begun an investigation of that operational unit in 1996). And nothing would have prevented the plaintiff from suing the contaminator, both to recover remediation costs already expended, and for a declaratory judgment as to liability and allocation of future costs, within the six years allowed for such a suit. See 42 U.S.C. §9613(g)(2) (allowing for a forward-looking claim for a “declaratory judgment on liability for…further response costs”). As applied to such circumstances, the single-remediation principle means simply and logically that the plaintiff cannot escape the six-year limitation period and endlessly postpone the bringing of suit by characterizing subsequent phases of the initial project as new remediations. A plaintiff whose suit is time-barred in such circumstances has suffered no unfairness as the preclusion was simply the result of the plaintiff’s needless delay. The several precedents that NYSEG followed in uttering the single-remediation principle similarly involved the same conventional circumstance, for which that principle is fair and sensible. The plaintiffs — with at least a general awareness of the contamination problems — had undertaken at the outset to remedy them. The subsequent stages of response were either (1) further steps towards remediating the original problems, see, e.g., California ex rel. Cal. Dep’t of Toxic Substances Control v. Hyampom Lumber Co., 903 F. Supp. 1389, 1390-91 (E.D. Cal. 1995), or (2) steps to remediate different aspects of the originally known problem, see, e.g., Sunoco, 337 F.3d at 1237-38. And there was no impediment that would have prevented the plaintiffs from suing the contaminator within six years of initiating the remediation as to both past and future remediation work. In Hyampom, for example, the District Court for the Eastern District of California considered a series of steps undertaken by California’s regulatory agency to remove contaminated soil from the site of a lumber mill. 903 F. Supp. at 1390. Those steps included: (a) the installation, in 1987 and 1988, of fences around the contaminated areas; (b) the construction, in September 1988, of temporary water and electrical infrastructure in preparation for the excavation of the contaminated areas; and (c) the excavation of the contaminated areas beginning in October 1988. Id. at 1390-91. There was no question that each of these steps were undertaken to remediate the same problem, and to bring about the same desired remedy, which the California Department of Toxic Substances (the regulator in charge of overseeing the cleanup) had identified in its Remedial Action Plan as the “excavat[ion] and remov[al] [of] the contaminated soil from the site.” Id. at 1390. There was nothing to prevent the State of California, which sought to recover the costs incurred by its regulator (under the same CERCLA provision as asserted here by MPM) from bringing a cost-recovery suit within six years of the “initiation of physical on-site construction” of its overall remediation effort. Because California failed to bring the suit within six years, it was not timely. The situation considered in Colorado v. Sunoco involved cleanup activities at a former gold mine site that contained contaminated water. 337 F.3d at 1237-38. The EPA, together with the State of Colorado, took control of the site in 1992 to deal with multiple environmental threats, including (a) several adits (i.e. horizontal mine openings for drainage) in the gold mine through which “metal-contaminated, acidic water” drained out of the mine and into the environment and (b) a “heap leach pad” containing waste treatment sludge leftover from mining and waste treatment operations. Id. at 1236-38. The EPA’s initial cleanup operations — undertaken in 1994 — included the improvement of the water treatment systems that dealt with the contaminated water on the “heap leach pad” and the plugging of the mine adits to prevent contaminated water from leaking out of the mine. Id. at 1237. The EPA later determined that a “permanent sludge disposal area” was required for disposal of the sludge from the heap leach pad; it undertook construction of that disposal area sometime between 1994 and 1996. Id. at 1238. It is clear that EPA undertook each of these steps to deal with the set of issues it set out to remediate when it took control of the site in 1992, and that each step was a phase in the regulator’s effort to bring about its desired “permanent remedy.” 42 U.S.C. §9601(24) (defining “remedial action”).38 There was no reason why the regulator could not have brought its suit within six years of the commencement of its response project in 1994; its failure to do so until 2001 meant that, as the Tenth Circuit held, it would have lost the opportunity to take advantage of CERCLA’s cost-recovery provisions for remediation.39 In the circumstances of NYSEG (and in the precedents it cites), the single-remediation principle was logical and fair, and it served the designs and objectives of RCRA and CERCLA. Although it is a reliable prescription in the great majority of cases, we do not believe that our NYSEG panel intended the principle to control if the circumstances of a case would render it illogical and unfair, and would defeat the statutory design or objectives. The most obvious example of inappropriate application of the single-remediation principle would arise when the subsequent remediation undertakes to remediate a problem that did not exist at the time of the prior remedial act. If, following a remediation, the operator of a site creates a different contamination, caused by different processes and a different contaminating substance, and the burden of remediating that contamination falls on a new owner that acquires the site many years later, there would never be a moment when the new owner could sue to recover its response costs. When such future contaminations occurred and became known, future owners or operators would be compelled by CERCLA and RCRA to report the discovery to government regulators and undertake any remediation that regulators may deem necessary, see 42 U.S.C. §9603(a) (requiring disclosure); id. §6924(u) (instructing regulators to “require[] corrective action for all releases of hazardous waste”); see also MPM III, 2017 WL 6408611, at *21 (noting that the RCRA permit for the Site required notification to regulators of newly-discovered contamination within seven days), but any suit to recover the cost of such remediation from the contaminator would be untimely by reason of a prior remediation decades in the past that had nothing to do with the present contamination, then as yet non-existent. We do not believe the NYSEG panel intended its single-remediation principle to apply in such a circumstance. A second less obvious but equally illogical and inappropriate example would arise when a site operator discovers a previously unsuspected contamination that was unrelated to, and perhaps far distant from, a previously remediated contamination. Suppose, for example, that the owner of a 10,000 acre site in 1990 completely remediated a chemical spill that it had caused on a tiny corner of a site. Suppose further that another operator acquired the site many years later and discovered a fifty-year-old contamination, far distant from, and unrelated to the contamination that had been remedied in 1990, and unsuspected at that time. If the single-remediation principle were to be categorically applied, the new owner, although compelled by law to report and remediate, would be unfairly precluded from utilizing CERCLA’s cost-recovery mechanism to recover the necessary costs of cleanup from the party responsible. If, in a third scenario analogous to the present case, the original polluter implemented an inadequate remediation pursuant to a regulatory approval that was procured by inadequate disclosure, and then held the site for six years after the misguided project began, a later owner who discovers the contamination that the prior remediation failed to address would nevertheless be obliged to disclose and remediate it, but could not recover its costs if the single-remediation principle were applied. The original polluter would enjoy an undeserved immunity as a result of its inadequate disclosure of what needed to be remediated, and its retention of the site for six years after initiating its incomplete remediation. A categorical single-remediation principle, while logical for the conventional pattern of cost-recovery suits, such as NYSEG, would frustrate and defeat the efficacy of cost-recovery as an essential motivator in the RCRA/CERCLA framework. Without access to cost-recovery suits against contaminators, neither private parties nor regulators would be incentivized to ensure that cleanups are conscientious and thorough. Cf. Key Tronic Corp. v. United States, 511 U.S. 809, 819 n.13 (1994) (“CERCLA is designed to encourage private parties to assume the financial responsibility of cleanup by allowing them to seek recovery from others.” (internal quotation marks omitted) (quoting FMC Corp. v. Aero Indus., Inc., 998 F.2d 842, 847 (10th Cir. 1993))). Worse still, an operator who, after taking ownership of a site, discovers preexisting contamination, if deprived of access to cost recovery, might conceal its discovery from regulators; if it disclosed the contamination, as CERCLA and the site’s RCRA permit would require, see 42 U.S.C. §9603(a); MPM III, 2017 WL 6408611, at *21, it may be required to undertake catastrophically costly remediation at its own expense while the contaminator, which should bear the costs, enjoys wholly unjustified immunity that results from overbroad application of a notion invented by the courts. Overbroad application of the single-remediation principle would accordingly undercut CERCLA’s manifest purpose to “encourag[e] the timely cleanup of hazardous waste sites” by private parties by “placing the cost of that cleanup on those responsible for creating or maintaining the hazardous condition.” Consolidated Edison Co. of New York, Inc. v. UGI Utils., Inc., 423 F.3d 90, 94 (2d Cir. 2005) (alteration, brackets, internal quotation marks, and citation omitted); see also H.R. Rep. No. 99-253(III), 1986 U.S.C.C.A.N. 3038, 3038 (explaining that one of CERCLA’s two primary goals is to “hold responsible parties liable for the cost of [necessary environmental] cleanups”). There is nothing in the words of the statute that invites a categorical single-remediation principle. Although Sunoco (which NYSEG cited) tried to find support within the statute’s wording, see 337 F.3d at 1241, the grammatical principle that the court invoked would, under careful examination, support the opposite conclusion.40 Moreover, the fact that neither the statute, nor the implementing regulations define the term “site” means that it can be applied to large areas, exacerbating the problems that could result from overly broad application of the concept. For the reasons explained above, we understand the single-remediation principle stated in NYSEG to mean that, when a remediation is undertaken under a remediation plan based on full disclosure of the known problem, successive remedial steps undertaken in furtherance of the original objective are part of a single remediation for purpose of the statute of limitations, so that the remediator may not delay suit by classifying subsequent stages of remediation of the original problem as new remediations. See NYSEG, 766 F.3d at 236. In that sense, there can be “only be one remedial action at a site” with respect to the particular contamination addressed and remedy conceived. Id. When, as in NYSEG, the contamination to be addressed arises from a single source, and the operator undertakes to remedy that “underlying source of [] contamination,” 766 F.3d at 233 (citation omitted), the distinct steps taken in furtherance of that objective will constitute a single remediation. But a subsequent remediation, undertaken to address a different source of contamination outside the scope of the prior remediation, may constitute a separate and distinct remediation for statute of limitations purposes. This interpretation comports more closely to the language and structure of the statute. CERCLA defines a “remedial action” as “those actions consistent with [a] permanent remedy…in the event of a release or threatened release of a hazardous substance into the environment.” 42 U.S.C. §9601(24). This language indicates that the term “remedial action” refers not to any remedial activity undertaken on a particular site, but instead to remedial steps intended to deal with a particular “release or threatened release of a hazardous substance.” Id; see also S. Rep. No. 96-848 at 54 (July 11, 1980) (“Remedial action may be taken in response to a discharge, release, or a significant threat of discharge[,] release, or disposal of a hazardous substance.”). CERCLA’s implementing regulations confirm this understanding: typically, remediations are only undertaken after a lengthy process of data collection, the goal of which is to produce a comprehensive description of the “history/nature of waste handling” at the site and a “description of known contaminants.” See 40 C.F.R. §300.420(c); id. 300.430(b). After the operator and the regulator obtain a comprehensive view of the “scope and complexity of the site problems being addressed,” id. §300.430(a)(1)(ii)(C), the parties analyze alternative courses of action, conduct feasibility studies, and select a remedy. See generally id. §300.430. It follows that the remediation that emerges from this process, while designed to be a “final, once-and-for-all cleanup of a site,” NYSEG, 766 F.3d at 236, is necessarily limited in scope to those problems revealed during the remedial investigation. A subsequent remediation that seeks to address a different set of problems — e.g. problems that were non-existent, unknown, elsewhere, or undisclosed to the regulators and unrevealed in an earlier remediation plan — should not be considered part of the same remediation.41 In deciding whether to characterize a later remedial activity, for statute of limitations purposes, as a continuation of previously conducted remediation, or as a separate and distinct remediation entitled to a new six-year period for cost recovery, a helpful inquiry would be to examine whether the recent action (sought by the remediator to be characterized as a new remediation) falls within the remedial scope of the previous remediation as revealed in the record before the regulatory agency. In many cases, the record of the correspondence between the regulatory agencies and the original remediators will reveal much about the nature and scope of the problem to be remediated as it was initially conceived. See 40 C.F.R. §300.430(f)(5)(i) (“To support the selection of a remedial action, all facts, analyses of facts, and site-specific policy determinations considered in the course of carrying out activities in this section shall be documented, as appropriate, in a record of decision…. “). CERCLA and its implementing regulations make clear that remediation requires “characteriz[ing] the nature of and threat posed by the hazardous substances,” including the “general characteristics of the waste, including quantities, state, concentration, toxicity, propensity to bioaccumulate, persistence, and mobility,” in addition to analyzing various site characteristics, to “conduct a site-specific baseline risk assessment.” Id. §300.430(d)(2), 300.430(d)(4). An assessment of the documentary record associated with the prior remediation will often show whether subsequent action is a continuation of it or distinct in its remedial purpose and scope. When, for example, the record reveals that the more recent remediation addresses a different problem than the previous remediation, e.g., a “release or threatened release of a hazardous substance” that was unrecognized or had not even occurred at the time of the previous remedial activity, that would suggest that the recent remediation should be treated as a new and distinct remediation for statute of limitations purposes.42 On the other hand, the more the agency record shows that the recent remedial activity seeks to bring about essentially the same “permanent remedy” of the same problem as was the goal of the prior remediation, the more appropriate to consider the recent remediation as a continuation of the prior remediation. NYSEG, 766 F.3d at 235. (As we recognized in NYSEG, the possibility that technological advancement may reveal new methods of remediating a particular contamination problem would not necessarily create the opportunity to pursue such methods as a distinct “remedial action,” particularly if those methods undertake to solve the same problems already addressed by the prior remediation. NYSEG, 766 F.3d at 236 (“[W]e recognize that what seems final at a given point in time might come to appear inadequate at a later date as scientific knowledge progresses.”).) We conclude that the district court’s reliance on the single-remediation principle asserted in NYSEG did not necessarily support its conclusion that MPM’s cost recovery action was untimely. We vacate the judgment and remand for further consideration. Consistent with our discussion above, the district court’s analysis on remand should address whether MPM’s remedial activity addressed to buried PCBs is part of the remediation begun by UCC in the 1990s, such that the instant suit filed in 2011 is untimtely, or should be deemed a separate and distinct remediation, entitled to a new six-year limitation period. ii. Future Removal Costs With respect to removal costs MPM may incur in the future, UCC challenges the district court’s declaratory judgment that UCC is liable for at least a share, as well as its allocation of 95 percent of those costs against UCC. UCC argues first that the possibility of future removal costs is too remote to present a constitutionally ripe controversy as to its liability for those costs, UCC Br. at 44-48; second, that the circumstances are “too speculative” to present a ripe controversy over the allocation of future costs, id. at 48-55; and, third, that cost allocation is not “prudentially ripe” because the issue would “benefit from [] further factual development” and because “[d]eferring allocation of hypothetical future removal costs would impose no real hardship on MPM,” id. at 55-56 (citation omitted).43 We disagree. a. Constitutional ripeness First, we find neither error nor abuse of discretion in the district court’s conclusion that the liability and allocation issues were constitutionally ripe for review due to the very high likelihood that MPM will incur future response costs. MPM III, 2017 WL 6408611, at *16-17. The doctrine of constitutional ripeness is “drawn from Article III limitations on judicial power” and “prevents a federal court from entangling itself in abstract disagreements over matters that are premature for review because the injury is merely speculative and may never occur.” In re Methyl Tertiary Butyl Ether (MTBE) Prods. Liability Litig, 725 F.3d 65, 110 (2d Cir. 2013) (citations omitted).44 There is no merit to UCC’s contention that it is speculative whether MPM will incur costs to deal with the PCB contamination it has uncovered at the Site. To the contrary, it is undisputed that MPM plans to continue with its wastewater treatment facility upgrade, which will “require that MPM respond appropriately to PCBs in the soils” in that area. MPM III, 2017 WL 6408611, at *17.45 MPM submitted a plan for PCB management during its wastewater treatment upgrade, which anticipates the necessity of the disposal of contaminated soil, along with other measures intended to “minimize the risk of spreading contamination,” and the WVDEP has approved that plan. There is a substantial likelihood that, depending on how these plans progress, at least some actions MPM will undertake to deal with UCC’s contamination of the Site — the extent of which is currently “unknown” — will be “clean-up measures taken in response to immediate threats to public health and safety,” which are properly characterized as removals. NYSEG, 766 F.3d at 230-31.46 Accordingly, we conclude that the district court did not exceed its Article III jurisdiction in adjudicating UCC’s liability for the future removal costs that MPM is likely to incur. See Kelley v. E.I. DuPont du Nemours and Co., 17 F.3d 836, 844-45 (6th Cir. 1994) (rejecting the argument that the district court’s declaration of future liability for cleanup costs was improper because “the wide-ranging contamination [at the relevant site] makes it more certain than speculative that the [claimant] will have to expend resources in the future”).47 b. Prudential ripeness The district court also correctly concluded that the issue of UCC’s liability and responsibility for MPM’s future response costs was prudentially ripe. The doctrine of prudential ripeness requires a court to ask “whether the claim is fit for judicial resolution” and “whether and to what extent the parties will endure hardship if decision is withheld,” and permits a court to decline to exercise its jurisdiction upon determining that “the case will be better decided later.” MTBE, 725 F.3d at 110 (brackets and citations omitted). The district court determined that the issue was fit for judicial resolution because the evidence necessary to determine UCC’s responsibility for MPM’s future response costs was “before the Court and unlikely to change.” MPM III, 2017 WL 6408611, at *18. In particular, the court noted that “the two factors that weigh most heavily in [its] allocation of responsibility in this case — UCC’s status as the sole entity responsible for using and disposing of PCBs at the Site and MPM’s delay in reporting its discovery of PCBs to environmental regulators — are settled.” Id. Further, the district court determined that it would be “unjust” to “require MPM to wait for allocation until MPM has ascertained [the extent of its] future PCB response costs,” given the “considerable” cost and time the parties had spent litigating this case, and that determining the parties’ respective financial responsibility for future cleanup costs without delay — as CERCLA requires — would “advance CERCLA’s ‘dual goals of cleaning up hazardous waste and holding polluters responsible for their actions.’” Id. (quoting NYSEG, 766 F.3d at 220). UCC argues that the issue is not fit for resolution because “many facts relevant to the parties’ relative responsibility for future removal costs are inherently unknowable and subject to substantial uncertainty” as “many of the circumstances of a future cleanup are unknown.” UCC Br. at 49-51.48 UCC further points to several specific uncertainties which, in its view, should preclude a determination of cost allocation, including (a) the “extent or impact” of MPM’s delay in addressing the PCB contamination, (b) the extent of MPM’s future cooperation with regulators, (c) the extent to which MPM will benefit economically from its future cleanup efforts, (d) the extent to which MPM is and will be responsible for “secondary disposal” of excavated contaminated material at the site, and (e) whether OSi/Crompton will indemnify MPM for any of its future response costs. UCC Br. at 51-54. UCC further asserts that MPM’s “future removal actions [] may well never occur,” arguing that resolution of its responsibility should wait “at least until (if ever) a concrete plan exists to conduct a removal action and more is known about the attendant circumstances.” Id. at 56. We are not persuaded. CERCLA provides a district court tasked with allocating responsibility for cleanup costs between potential responsible parties with “broad discretion to balance the equities in the interests of justice,” and “does not limit courts to any particular list of factors” to consider when making that determination. Bedford Affiliates v. Sills, 156 F.3d 416, 429 (2d Cir. 1998), overruled on other grounds by W.R. Grace, 559 F.3d at 89-90. The district court was clear that the two factors on which it primarily relied were (a) UCC’s sole responsibility for the PCB contamination at the Site, and (b) MPM’s delay in cooperating with regulators after its discovery of PCB-contaminated soil. MPM III, 2017 WL 6408611, at *18. The district court found that the evidence supporting these two determinative factors was “unlikely to change.” Id. The district court had a sound basis for concluding that UCC’s insistence on uncertainties was exaggerated and did not call for delay in adjudicating allocation. Id.49 As to the “hardship” element of prudential ripeness, MTBE, 725 F.3d at 110 (instructing courts to determine “whether and to what extent the parties will endure hardship if decision is withheld”), UCC argues that “[d]eferring allocation of hypothetical future removal costs would impose no real hardship on MPM” because “[i]f MPM ever does develop a concrete plan beyond [its Soil Management Plan], it can seek an allocation then.” UCC Br. at 56-57. This argument does not answer the district court’s determination that delaying determination of the allocation issue “until after MPM is again engaged in PCB cleanup” would be “unjust” and “wasteful” because of the “considerable” cost and time that have been spent in litigating the case. MPM III, 2017 WL 6408611, at *18 (“Determining allocation will save the litigants in this case substantial time and money…. “). Moreover, UCC does not challenge the district court’s determination that finalizing the issue of cost allocation — which, in its sensible view, it was fully capable of doing after the evidence presented at trial — would advance CERCLA’s ultimate goals. Id.50 As the district court pointed out, its declaration of liability for and allocation of MPM’s future response costs does not directly establish any financial liability. The court’s order “[left] open [] the issue of whether any costs MPM incurs in responding to PCBs are recoverable under CERCLA,” so that “if and when MPM brings an action to recover future removal costs, UCC will be entitled to raise appropriate objections.” MPM III, 2017 WL 6408611, at *18.51 What the district court’s order did finalize was that, by reason of UCC’s sole responsibility for the substantial levels of PCB contamination at the Site, it is responsible (notwithstanding MPM’s delay) for 95 percent of the cost of any appropriate, CERCLA-compliant removals designed to address that contamination. As explained below, that conclusion was well supported by the record, and the district court was under no obligation to delay decision and require the parties to come back to court at some future date to litigate the allocation of financial responsibility for UCC’s pollution. c. Future removal cost allocation UCC further argues that the district court’s decision to allocate 95 percent of MPM’s future removal costs to UCC was an abuse of discretion. UCC Br. at 58-70. The thrust of UCC’s argument is that the district court did not afford sufficient weight to several factors on which other courts have relied when making similar allocation determinations. UCC argues that the district court failed to take into account (a) that MPM assumed the risk of future cleanup costs by purchasing the Site without fully investigating potential contamination there, id. at 59-61, (b) that, by virtue of the documents MPM received before executing its purchase of the site, it was on objective notice of the PCB contamination, which should warrant a higher allocation to MPM under the doctrine of caveat emptor, id. at 62-67, and (c) the possibility that “future PCB cleanups…could result in substantial economic benefits to MPM,” id. at 68-69.52 A district court’s allocation of response costs under CERCLA is reviewed for abuse of discretion. Goodrich Corp. v. Town of Middlebury, 311 F.3d 154, 168-69 (2d Cir. 2002); see also United States v. Consolidation Coal Co., 345 F.3d 409, 412 (6th Cir. 2003) (same). The allocation of CERCLA response costs is an “equitable determination based on the district court’s discretionary selection of the appropriate equitable factors in a given case.” Goodrich, 311 F.3d at 170. The statute does not require district courts to consider or give weight to any particular allocation factor, but rather permits the court to determine which factors are most relevant to a given case. NCR Corp. v. George A. Whiting Paper Co., 768 F.3d 682, 700 (7th Cir. 2014) (“[W]e have stressed that the district court’s discretion [in allocating CERCLA costs] is broad, both when it determines how much weight to place on any given equitable factor before the court, and also when it chooses which factors are pertinent at all for the case before it.”).53 UCC’s argument that the district court should have assigned more weight to certain allocation factors is therefore unavailing. UCC does not argue that the district court’s cost allocation “rests on an error of law (such as application of the wrong legal principle) or a clearly erroneous factual finding,” and it is clear that the court’s decision was “within the range of permissible decisions.” Goodrich, 311 F.3d at 169. Accordingly, we conclude that the district court did not abuse its discretion in allocating 95 percent of MPM’s future removal costs to UCC. By affirming the district court’s grant of declaratory judgment on future removal cost allocation, we do not imply, much less rule, that the district court would be barred from reconsidering its allocation on the basis of new events or discoveries that would make the present anticipatory allocation inappropriate. Nor does this discussion constitute a ruling that such reconsideration of the allocation would be permitted. Our purpose is to leave the question open, expressing no view either way.54 C. CONCLUSION For the foregoing reasons, we VACATE the district court’s July 7, 2016 grant of partial summary judgment holding that MPM’s claim for recovery of remediation costs is time-barred under 42 U.S.C. §9613(g)(2), AFFIRM the district court’s September 22, 2017 order holding that UCC is liable to MPM for 95 percent of future removal costs, and REMAND for further proceedings consistent with this order.