Dooley, J.
Safety Medical Systems, Inc. (SMS) appeals from a decision of the Chittenden Superior Court denying it an air pollution
control permit to operate a medical waste incinerator because the operation will emit excessive levels of certain toxic substances. The parties opposing the permit are Conservation Law Foundation (CLF), Raymond Gonda, Timothy Fasten, Coca-Cola Bottling Company of New England, Inc., Seventh Generation, Inc., Little Red Wagon Daycare, Inc. and the City of Winooski.
Safety Medical Systems argues that the action of the Vermont Agency of Natural Resources (ANR), which granted the permit, was supported by the record and the applicable regulations and should have been affirmed.
Specifically, SMS argues that the ANR decision must be affirmed on theories of “de minimis” impact or offsetting benefits.
We affirm in part, and reverse in part.
In September 1989, Safety Medical Systems applied for an air pollution control permit to operate a medical waste incineration facility in Colchester. The facility includes two incinerators, each capable of incinerating 1,000 pounds of medical waste per hour. After further submissions, the application was deemed complete in July 1990. After a public hearing, ANR granted Safety Medical Systems’ •application and issued a permit.
The ANR decision was subsequently appealed to the superior court. While the appeal was pending, Safety Medical Systems completed the facility and gained all other required approvals. Because the permit contained a ninety-day waiting period between the time the facility was ready to operate and the date of first operation, Safety Medical Systems requested a remand to ANR to seek a permit modification. The remand was granted, and ANR held another public hearing, primarily on modifications to the permit. ANR reaffirmed its original decision but made three amendments to the permit, including elimination of the ninety-day waiting period. The matter again went to superior court, which, on Conservation Law Foundation’s motion for summary judgment, held that ANR acted in violation of the applicable regulations in granting the permit.
I.
A.
The case is controlled by air pollution control regulations adopted by ANR pursuant to 10 V.SA,. § 554(2).
These regulations employ a three-step process to determine whether the emission of hazardous air contaminants will be authorized. For each hazardous air contaminant, ANR must first determine whether the emission rate will exceed the action level specified in the regulations for that contaminant. Vermont Agency of Natural Resources, Environmental Protection Regulations (EPR) § 5-261(1). If the emission does not exceed the action level, there is no further regulation of that contaminant.
Id.
If the emission does exceed the action level, the second regulatory step requires the source, in this case Safety Medical Systems, to use control technologies “to achieve the hazardous most stringent emission rate (HMSER).”
Id
§ 5-261(2). Finally, even if the second step is met, ANR may not issue a permit allowing the discharge of hazardous air contaminants “which cause or contribute to ambient air concentrations in excess of any Hazard Limiting Value.”
Id
§ 5-261(3). The regulations contain hazard limiting values for each of the hazardous air contaminants.
Id
Ch. 5 Appendix C.
B.
According to the data supplied by Safety Medical Systems, the medical waste incinerator will emit cadmium, chromium, dioxin and nitric oxide at a rate that exceeds the action level for each contaminant. For example, the action level for chromium is 0.0000071 pounds per eight-hour period; the SMS incinerator will emit 0.000018 pounds of chromium in the same period. Because the SMS incinerator met the first step for regulation, ANR then found that Safety Medical Systems must use control technologies to achieve the most stringent emission rate (HMSER) for each of the contaminants. It further found that Safety Medical Systems proposed to use sufficient control technologies to meet this requirement.
C.
The issue in this case arises with respect to ANR’s handling of the final regulatory step dealing with the impact of the SMS incinerator
output on ambient air concentrations of cadmium, chromium, dioxin and nitric oxide. As authorized by its regulations, ANR gave Safety Medical Systems the option of determining the actual ambient air concentrations at the site through a year of testing or of using preexisting data generated off-site for the four contaminants. Safety Medical Systems chose the latter option.
The only data for cadmium and chromium were collected at Randolph, Vermont between 1980 and 1985. For each contaminant, the data showed that the average level in the air exceeded the applicable hazard limiting value (HLV). No data were available for levels of dioxin or nitric oxide in the air in Vermont. The nearest place surveyed to determine ambient air levels of dioxin was in Connecticut. The Connecticut data registered levels of dioxin above the Vermont hazard limiting value. As for the lack of data on nitric oxide, ANR assumed that nitric oxide converts rapidly to nitrogen dioxide, so that existing levels of nitric oxide should be considered to be negligible.
ANR discounted the ambient air level findings for the four contaminants and issued the permit to Safety Medical Systems despite indications from the available data that the SMS incinerator would emit contaminants in excess of ANR’s hazard limiting values. ANR did so because it found the amount of the emissions to be de minimis and because the addition of contaminants from the SMS incinerator would be offset by a greater reduction in emissions of those contaminants in other incinerator facilities. On remand, it added that the Randolph and Connecticut data were unreliable. The superior court rejected each reason. We agree with respect to the first two reasons but conclude that analysis of the third reason is flawed because of inadequacies in the record. We now treat these three points of contention in succession.
II.
The superior court first found fault with ANR’s application of a “de minimis” test when judging the impact of the increased ambient air concentration of the four contaminants against each contaminant’s hazard limiting value. As part of the permitting process, ANR required Safety Medical Systems to show the extent to which its discharge would increase the ambient air level for each of the pollutants within 200 meters of the point of discharge. The results
showed that the SMS incinerator discharges would increase the ambient air level of cadmium 3.3%; of chromium, 1.4%; of dioxin, 3.6%; and of nitric oxide, 0.6%. As noted previously, these increases were over and above existing concentrations which the data used showed to be in excess of the hazard limiting values. Nonetheless, based on the discharge test findings, ANR concluded that Safety Medical Systems’s contribution to concentrations of contaminants would be de minimis so that “human health and safety is adequately protected.” Then, as a first basis for its decision to grant the permit, ANR ruled that a de minimis violation of the prohibition on contributing to ambient air concentrations above a hazard limiting value was allowable.
Conservation Law Foundation argues, and the superior court accepted, that EPR § 5-261 can not be read to allow a de minimis violation. The crux of this position is that in the first step of the analysis, the comparison of the proposed discharge amounts to predetermined action levels, the regulations already determine when a discharge is significant enough to be considered. The superior court agreed and held that the regulations do not contain such a “de minimis” exception and that ANR therefore could not sustain granting the permit on this basis.
This issue involves a question of the proper interpretation of an agency’s regulations. In interpreting regulations, our overall goal is to discern the intent of the drafters.
In re Verburg,
159 Vt. 161, 164, 616 A.2d 237, 239 (1992). We employ a deferential standard of review of an agency’s interpretation of its own regulations. See
id.
at 165, 616 A.2d at 239. Its interpretation may be overcome only by compelling indications of error.
Id.
(quoting
Rogers v. Watson,
156 Vt. 483, 489, 594 A.2d 409, 412 (1991)).
We must view the hazardous air contaminant regulations as a whole.
Id.
at 165-66, 616 A.2d at 239. If they require that no discharge of a hazardous air contaminant shall cause or contribute to ambient air levels above a hazard limiting value, the ANR position would not be sustainable. To consider whether an emission is de minimis after it is known that the level of a contaminant in the ambient air exceeds a hazard limiting value is inconsistent with the structure and obvious intent of the regulation. Similarly, we have declined to read a de minimis exception into a zoning ordinance or the zoning enabling statute.
In re Cumberland Farms, Inc.,
151 Vt. 59, 64, 557 A.2d 486, 489 (1989). If the Agency wishes to include an additional de minimis exception, it must do so explicitly.
B.
The superior court also found fault with ANR’s reasoning that the discharge from the SMS incinerator was more than offset by reductions in emissions from existing uncontrolled incinerators that would send their waste to the Safety Medical Systems facility. ANR determined, again from studies SMS supplied, that the replacement of older medical waste incinerators in Vermont with the SMS facility would cause a net reduction in the amount of chromium, cadmium and dioxin emitted into Vermont air. On that basis, it concluded that the SMS emissions would not “cause or contribute to an exceedence of any HLV.” The superior court held that the ANR regulations do not allow for this type of offset analysis.
This issue is again a matter of proper interpretation of the ANR regulations. Despite our deference to the Agency’s interpretation of this regulation, the structure of the whole regulation again precludes us from accepting it. The hazardous air contaminant regulation is silent on the subject of offset authorization. By contrast, the Agency’s nonhazardous air contaminant regulations contain a specific authorization for considering “binding offsetting emission reductions.” EPR § 5-502(6)(a). The effect of considering offsets may be to trade a degradation of air quality in one place in order to achieve an improvement in another. We decline to read an offset authorization into a silent rule, particularly when ANR has been explicit in authorizing this analysis for other air pollution standards. The superior court was correct in rejecting this basis.
Finally, the superior court found that ANR had granted the permit on the basis of an inadequate record. Upon remand from the superior court, ANR added a third reason that justified its decision to grant the SMS permit even though the available data showed an increase in ambient air concentrations above the hazard limiting values. ANR stated that the data from the Randolph, Vermont study were flawed because the instruments were incapable of measuring at the low discharge rates of the hazard limiting value. ANR reasoned that whenever the level was below that measurable by the instruments, the lowest level readable was recorded, artificially increasing the average. Apparently, ANR believed that the Connecticut data on dioxin were not the most accurate. Essentially ANR’s reasoning is that no valid data showed the amount of chromium, cadmium or dioxin in the air to be above the relevant hazard limiting value.
The superior court rejected this reasoning because ANR specifically stated, even after the remand, that the decision to award the permit was made on the record created before the remand. Because the court found that this record contained none of the technical analysis to support rejection of the Randolph and Connecticut data, it held that ANR could not grant the permit on this basis. The court concluded that the rejection of the data was a post hoc justification for the ANR decision. It held that ANR’s decision to award the permit was “void as a matter of law” and that the only way Safety Medical Systems could obtain a permit was through a variance. See 10 V.S.A. § 561.
Although this issue was decided by the superior court based primarily on its determination of the content of the record, a determination we examine in detail below, there is an underlying conflict about the requirements of the regulation. CLF’s position is that the permit applicant has the burden to show the level of contaminants in the ambient air and that these levels do not exceed the relevant hazard limiting values. Safety Medical Systems’ position is that the hazard limiting values for chromium, cadmium and dioxin are so low that it is impossible to determine through existing technology whether the levels currently existing in the air exceed them. From this, SMS argues that it cannot be assigned the burden to prove something that is technologically impossible to prove.
The hazardous air contaminant rules prohibit the discharge of any contaminant that causes or contributes to ambient air concentrations above the relevant hazard limiting value, EPR § 5-261(3), but the Secretary is allowed discretion in the administration of this requirement. The regulation states that the Secretary
“may
require ... an air quality impact evaluation,
id.
(emphasis added), and the use of the term “may” allows the Secretary to dispense with such an evaluation in appropriate circumstances.
Thus, we do not accept the rigid burden of proof argument made by CLF. The permit may be sustainable if Safety Medical Systems prevails on the record deficiency issue decided by the superior court.
An analysis of the superior court’s decision requires further background about the nature of this proceeding in superior court and the state of the record. As stated above, ANR found the Randolph and Connecticut data flawed, but the superior court rejected this rationale as unsupported by the record. Specifically, the court held:
This “explanation” of the ANR’s decision to grant SMS a permit does not harmonize with its findings of fact in the original permit which the ANR expressly affirmed on remand without reopening the record. It was SMS’ responsibility in the permit process to provide ambient air quality data applicable to the proposed site. SMS chose to rely on the Randolph and Connecticut incinerator studies to develop this data. The ANR recites the data in its findings of fact as the impact of the proposed facility, but then out of hand refutes them as unreliable with no further explanation.
We recognize that it is possible that the Randolph and Connecticut data submitted by SMS may be reliable for sound scientific reasons with respect to its applicability to the SMS facility. The record appears to be void of any such scientific rationale for the agency’s discarding the data that SMS chose to rely on.
ANR’s decision on remand, to which the court refers, came after a hearing in which members of the public, including CLF and those represented by it, could comment orally and in writing on the SMS permit. SMS was not expected to, and did not, present any evidence at that hearing. Following the hearing, the Secretary of ANR stated in her decision:
After carefully reviewing the prior record in this matter and considering the public written and oral arguments, the Agency has concluded that its original September 14, 1990 decision to issue the permit was proper and based on adequate evidence in the record. For purposes of clarification, the Agency has responded to the comments received at and after the May 3 public hearing. However, the Agency has decided not to reopen the record in this matter or to take additional evidence on the issues. Information presented at the May 3 hearing, and subsequently, was either available to and considered by the Agency in its permit decision or is not significant enough to warrant reconsideration of the decision to issue the permit. The Agency has decided to reaffirm and reissue the permit with the findings of
fact and conclusions of law as stated in the September 14, 1990 decision.
The Secretary attached two documents that respond to claims made at the hearing to her decision; in one is the explanation about the unreliability of the underlying data.
Early in the case, the superior court made two important and correct rulings, that place critical significance on the record developed in the Agency. The first is that this proceeding is governed by Rule 74 of the Vermont Rules of Civil Procedure, which governs appeals from governmental agencies when the right to appeal is given by statute. As originally adopted, Rule 74 applied only in appeals from decisions in “contested cases” governed by the Vermont Administrative Procedure Act. Reporter’s Notes to 1981 Amendment, V.R.C.P. 74. A contested case is a proceeding “in which the legal rights, duties, or privileges of a party are required by law to be determined by an agency after an opportunity for hearing.” 3 V.S.A. § 801(b)(2). A proceeding is not a contested case if there is no legal obligation to hold a hearing, even if the agency chooses to do so. See
In re Marble Sav. Bank,
137 Vt. 123, 125, 400 A.2d 1022, 1023 (1979). The proceeding in ANR was not a contested case because there is no legal obligation to hold a contested case hearing, although ANR twice chose to hold a form of a hearing.
In 1981, Rule 74 was amended to extend its coverage to all cases when review of agency action is provided by law in the superior court, whether or not involving a contested case. See Reporter’s Notes to 1981 Amendment, V.R.C.P. 74. The main effect was to apply appellate procedures to such cases, including the use of the notice of appeal to commence the action in superior court, even though only informal adjudication
occurred in the administrative agency. See
id.
The 1981 amendment to Rule 74 covers this case because review is specifically
authorized by statute, 10 V.S.A. § 562(c), although a contested case is not involved.
The court’s second ruling was that review of an agency decision would occur based on the record in the Agency, with some deference given to the Agency’s determination. Although Rule 74 is structured with the expectation that review will proceed on the record, see V.R.C.P. 74(d), the rule can be applicable if review is de novo. See V.R.C.P. 74(e) (jury trial available when there is right to have question determined by jury). The nature of review is determined by the Legislature, but we presume that review will be on the record and not de novo. See
Department of Taxes v. Tri-State Indus. Laundries, Inc.,
138 Vt. 292, 296, 415 A.2d 216, 219 (1980). Although we have applied this principle in contested cases, based partially on the ability of parties to develop facts before the agency,
id.
at 295, 415 A.2d at 219, it is rooted in the separation of powers between the executive branch agencies and the judiciary.
Id.
at 294-95, 415 A.2d at 218-19. The superior court has only a narrow role in ensuring that the decisions of ANR are made in accordance with law. In particular, the superior court is not a higher environmental agency entrusted with the power to make environmental law and policy de novo or with the power to apply the policy it develops to the facts it finds.
Id.
at 297, 415 A.2d at 220; see also
Chioffi v. Winooski Zoning Bd.,
151 Vt. 9, 13, 556 A.2d 103, 106 (1989) (in zoning appeals, “court must resist the impulse to view itself as a super planning commission”).
The statute under which review proceeded in this case, 10 V.S.A. § 562(c), does not specifically state that review would proceed de novo. In fact, it describes review as occurring “by appeal,” indicating that the Legislature intended that on-the-record review would be used.
Id.
The trial court was correct in holding that review in this case would be based on the record below.
To facilitate on-the-record review, Rule 74(d) specifies that the record on appeal, when no contested case is involved, consists of “all writings and exhibits in the agency proceeding” and a transcript of any oral proceedings. The rule requires the Agency to transmit this record to the superior court clerk within thirty days of the filing of the notice of appeal. V.R.C.P. 74(d).
We are not clear on the extent to which Rule 74(d) was followed in this case, although ANR did purport to file a record in superior court. Our attempt to establish the content of the record in this case is described in Appendix A to this opinion. In any event, in response to motions by the parties, the superior court designated the record for
purposes of review. It designated the record filed by ANR, the ANR remand decision with the attachments responding to the public hearing comments, the ANR regulations, and an internal memorandum to the ANR Secretary on the proper interpretation of the regulations. Although the court included the record filed by ANR, the only documents considered were the permit, including the findings of fact, and a technical analysis prepared by an ANR staff member.
In order for judicial review to proceed on the record, it is critical that the court have before it the full agency record “that was before the Secretary at the time he made his decision.”
Citizens to Preserve Overton Park, Inc. v. Volpe,
401 U.S. 402, 420 (1971),
overruled on other grounds by Califano v. Sanders,
430 U.S. 99, 105
(1977);
see also
Camp v. Pitts,
411 U.S. 138, 142 (1973) (“focal point for judicial review should be the administrative record already in existence”). The record for review is not necessarily limited to that submitted by the agency; it “‘consists of all documents and materials directly or
indirectly
considered by agency decision-makers.’” See
Thompson v. United States Department of Labor,
885 F.2d 551, 555 (9th Cir. 1989) (quoting
Exxon Corp. v. Department of Energy,
91 F.R.D. 26, 32 (N.D. Tex. 1981) (emphasis in original)). Thus, “if the agency decisionmaker’s decision is based on the work and recommendations of subordinates, the record should include all documents considered by the agency employees whose input reached the decisionmaker.”
McMillan & Peterson,
The Permissible Scope of Hearings, Discovery, and Additional Factfinding During Judicial Review of Informal Agency Action,
1982 Duke L.J. 333, 342.
We find that the record before the court in this case was incomplete. It appears that the “record” supplied by ANR contains no document earlier than the proposed permit. For example, it does not include the permit application or any of the information Safety Medical Systems submitted in support of the application. More
important, it does not contain the Randolph or Connecticut studies or the data derived from them.
The trial court has a number of remedies when the record supplied is inadequate. For example, it may call for the missing documents or it may have the agency supply the needed material through testimony or other evidentiary methods. See
Franklin Sav. Ass’n v. Director, Office of Thrift Supervision,
934 F.2d 1127, 1139 (10th Cir. 1991). Here, the trial court used the inadequacy of the record to reverse the ANR decision as unsupported by the record supplied. Although such action might be appropriate in extreme cases, it should be accompanied by a remand to allow the agency to cure the deficiency. See
Florida Power & Light Co. v. Lorion,
470 U.S. 729, 744 (1985) (“if the reviewing court simply cannot evaluate the challenged agency action on the basis of the record before it, the proper course, except in rare circumstances, is to remand to the agency for additional investigation or explanation”). Here, the initial remedy should have been to deny summary judgment and to seek the full record.
Alternatively, the trial court rejected ANR’s rationale because the unreliability of the Randolph and Connecticut data appeared to be a post hoc justification and could not be considered in light of ANR’s refusal to reopen the record. We concur that the Agency decision must stand or fall on the reasons given contemporaneously with the decision and not a later revision of those reasons. See
Camp v. Pitts,
411 U.S. at 143. The documents in the record supplied to this Court, however, see Appendix A, indicate that the Agency always harbored concerns about the Randolph and Connecticut data. The Secretary’s decision not to reopen the record
is not inconsistent with this rationale because the underlying data were already before the Agency. Thus, it does not appear this is an after-the-fact reason.
Even if we concluded that the measurement and data quality rationale was added too late and could not support the permit, we do not agree that the remedy is for the court to deny the permit. As the
United States Supreme Court has held, the proper remedy is to vacate the Secretary’s decision and remand the matter for further consideration.
Id.
Because of its nature, it will be true that the rationale for agency action based on informal adjudication will often be inadequately explained. Thus, the “court may require the administrative officials who participated in the decision to give testimony explaining their action.”
Citizens to Preserve Overton Park,
401 U.S. at 420. The purpose of this evidence is to determine whether the agency considered “all relevant factors or fully explicated its course of conduct or grounds of decision.”
Friends of the Earth v. Hintz,
800 F.2d 822, 829 (9th Cir. 1986). Such an exploration of agency reasoning might have been beneficial here.
The superior court’s decision that the Environmental Protection Regulations do not contain a de minimis exception or allow offset analysis is affirmed; the remainder of the decision is reversed and remanded for proceedings not inconsistent with this opinion.
APPENDIX A
— THE
RECORD
Our experience in attempting to determine the record before the superior court requires some explanation. Conservation Law Foundation filed both a Rule 75 action, by filing a complaint with the superior court, and a Rule 74 appeal, by filing a notice with the Agency. The Agency transmitted the notice of appeal to the clerk of the superior court, and Safety Medical Systems has included this document in its printed case. There is no docket entry reflecting that the notice was filed by the clerk, nor does the notice appear in the file as we received it from the court. ANR took the position that Rule 74 represented the appropriate route for judicial review and filed a motion to dismiss the complaint. The superior court agreed, and the complaint was dismissed by stipulation.
After receiving the notice of appeal, ANR transmitted the “record” to the superior court. Again, there is no docket entry showing its receipt and filing; what was transmitted to the superior court was not transmitted to this Court in accordance with V.R.A.E 11. We know it was received and considered by the superior court because it is described in the court’s opinion as “the contents of two folders filed with the court by the Vermont Department of Environmental Conservation.”
We asked the parties to supply us with the two folders referenced in the superior court’s decision, and the Agency sent us what
purported to be a copy of the contents of those folders. Over 90% of the documents transmitted dealt with interactions between ANR and Safety Medical Systems after the permit was issued and after the record was transmitted to the superior court. These documents are obviously not part of the “two folders” transmitted to the superior court; we have sent them back to ANR and not considered them.
The documents sent to us include a number that were generated before November 9, 1990, the date the “record” was sent to the Chittenden Superior Court. Most deal with staff answers to those who complained about issuance of the permit. The only documents that precede the issuance of the permit are proposed permit conditions, and the letter of transmittal of these conditions to Safety Medical Systems (but no response from Safety Medical Systems); a notice of a public hearing (but no indication of what occurred at that hearing); a proposed permit with a transmittal letter to Safety Medical Systems; correspondence from Conservation Law Foundation opposing the proposed permit and seeking information; and a response to the CLF position by Safety Medical Systems. The SMS response states, “The issues [CLF]. . . raise[s] were all debated at length and fully resolved during the long application review process.” Obviously, none of the documents generated in the long application review process made it into the record considered by the trial court or transmitted to us.
We have assumed for purposes of this opinion that all parts of the “record” in the two folders were made available to us. Nothing in the superior court decision indicates that it had access to additional information.
The present case illustrates the difficulty of establishing the “record” in appeals from complex regulatory decisions under Rule 74. The frequency and importance of these cases is likely to increase in Vermont, as similar appeals have done under federal regulatory schemes over the last two decades or more. Private, administrative and judicial resources will be used most efficiently if parties, agencies, and judges focus on the question of the record at the earliest moment and emulate as far as possible the care reflected in creating the record in contested cases.