Environmental and Hazard Screens Before the Phase I

The public environmental record of the United States is mostly a record of things that are no longer there. At the midpoint of fiscal year 2026, the U.S. Environmental Protection Agency counted 534,189 active underground storage tank systems in the country and 2,010,194 closed ones, and it counted 583,313 releases confirmed since the program began in 1984 (EPA Office of Underground Storage Tanks, 2026). There are more confirmed petroleum releases on file than there are regulated tanks in the ground.

That single comparison explains why a screen built before a Phase I is worth running and what it is worth. The question an environmental site assessment answers is a question about history: what was here, what was next door, and what did it leave behind. The largest public files bearing on that question are files of closures, releases and cleanups, and a lender can read them in minutes, for nothing, from a desk, before a consultant is engaged and before a term sheet is drafted.

What follows is that screen: which public layers exist, who publishes them, what each one reaches, what the federal rule says about how far around a parcel to look, and where the whole exercise stops. The stopping point matters more than any of the layers, so it goes at the top and it will be repeated. A public-data screen is not a Phase I Environmental Site Assessment. It does not become one by being thorough. The Phase I is defined by ASTM E1527-21 as referenced in 40 CFR part 312, the All Appropriate Inquiries rule, and it must be performed by an environmental professional as that rule defines one. Nothing described in this article delivers the landowner liability protections under the Comprehensive Environmental Response, Compensation, and Liability Act, and nothing here is a substitute for the judgment of the environmental professional or the lender. MMCG Analytics provides data and analytics with source and vintage provenance carried on every displayed value; the decisions rest with the lender and its environmental professional.

The screen has one job: to tell a credit officer, early and cheaply, what the assessment is likely to find, so that the engagement is scoped correctly the first time and priced into the timeline before anybody commits. It is the environmental and hazard layer of screening a site before the term sheet, taken deeper, and it belongs to the same discipline as the rest of analytics in the credit file: every figure carries a named source and a date, and every negative result carries the radius it was negative within.

What the screen is, and the line it never crosses

Start with the legal architecture, because the whole workflow hangs on it and most descriptions of it are loose.

CERCLA imposes strict, joint and several liability on owners and operators of contaminated property, and it provides a small set of defenses. Those defenses, the innocent landowner defense, the bona fide prospective purchaser protection and the contiguous property owner protection, are available only to a party that performed all appropriate inquiries into the previous ownership and uses of the property before acquiring it. EPA's rule at 40 CFR part 312 defines what all appropriate inquiries means. Section 312.11 references two ASTM standards as satisfying the rule: E1527-21, Standard Practice for Environmental Site Assessments, Phase I Environmental Site Assessment Process, and E2247-23 for forestland and rural property. The transition clauses that permitted the older E1527-13 and E2247-16 expired on 13 February 2024 and 24 June 2025 respectively, so neither older standard satisfies the rule now (40 CFR 312.11, read 24 August 2026).

Section 312.10(b) defines who may perform the inquiry. An environmental professional holds a professional engineer's or professional geologist's license plus three years of full-time relevant experience, or a government license or certification to perform environmental inquiries plus three years, or a baccalaureate or higher degree in engineering or science plus five years, or ten years of full-time relevant experience without a degree. Relevant experience is defined in the same section as participation in the performance of all appropriate inquiries investigations, environmental site assessments or other site investigations involving an understanding of surface and subsurface environmental conditions.

Two consequences follow, and they are the boundary conditions of everything below.

First, no amount of public-records work performed by anyone other than an environmental professional advances the AAI defense by a single step. A screen is not a partial Phase I, not a Phase I in progress, and not evidence of inquiry within the meaning of the rule. It is management information. When it is written into a file it should say so on its face.

Second, even a completed AAI-compliant Phase I buys less than most credit memoranda imply. The FDIC's examination manual states that a bona fide prospective purchaser protection reached through an AAI-compliant evaluation applies under CERCLA only, and not under the Resource Conservation and Recovery Act, which includes underground storage tank liability, nor under other federal or state statutes (FDIC, Risk Management Manual of Examination Policies, Section 3.2, 2023). The OCC's handbook makes the parallel point about the lender's own position: the CERCLA section 101(20) secured creditor exemption protects a lender that holds a security interest without participating in management, and it does not protect the bank from the decline in collateral value that contamination causes (OCC, Comptroller's Handbook, Commercial Real Estate Lending, version 2.0, 2022). A clean Phase I is not a clean site, and a defense against a cleanup claim is not a defense against a valuation loss.

The SBA chain starts with a data question

For any file that will carry an SBA guaranty, the environmental workflow is written down, and the first step in it is a records question rather than a site question. SOP 50 10 8, effective 1 June 2025, requires an environmental investigation of all commercial property offered as security for a loan or debenture. The investigation begins with a good faith determination of the NAICS codes for the property's current and known prior uses, compared against Appendix 6, the list of environmentally sensitive industries (SBA, SOP 50 10 8, Section A, Chapter 5, Paragraph E).

What happens next is decided by that comparison. A match requires the investigation to begin with a Phase I regardless of loan amount, and codes beginning 457, gasoline stations and fuel vendors, additionally carry the gas station requirements at Appendix 7. With no match, a loan up to and including $250,000 may begin with an environmental questionnaire, and a loan above $250,000 must begin at minimum with a questionnaire plus a records search with risk assessment. A records search with risk assessment that rates the property anything other than low risk escalates to a Phase I. A transaction screen or a Phase I whose environmental professional finds further investigation warranted escalates to the next step, typically a Phase II. Any deviation requires an exception from SBA's environmental committee.

The definitions sit in Appendix 4. A Phase I is an AAI-compliant assessment by an environmental professional under the most recently adopted ASTM standard, currently E1527-21, citing 40 CFR 312.20. A transaction screen is limited due diligence under ASTM E1528-22: an interview with the owner or operator, a site visit, questionnaires A and B, and a review of government records and historical sources. A records search with risk assessment is a search of the government databases identified in 40 CFR 312.26 plus historical use sources, aerial photography, city directories, reverse directories and fire insurance maps, for the property and adjoining properties, followed by an environmental professional's risk assessment. Every transaction screen and every Phase I and Phase II must be performed by an environmental professional and accompanied by the Appendix 5 reliance letter, and every report must be dated within one year of issuance of the SBA loan number.

Read that chain again with a lender's eye. Three of its four rungs, the questionnaire aside, are the work of an environmental professional. The one rung that is not, the NAICS determination against Appendix 6, is a pure data task, it decides everything downstream, and it can be done before a single consultant is contacted. That is the screen's point of maximum leverage in an SBA file.

Appendix 6 rewards close reading. MMCG counted 55 entries in the list as printed in SOP 50 10 8, at three to six digits, and the appendix states that a shorter code covers every industry beginning with those digits. Thirty-two of the 55 entries are three-digit codes, which means they reach an entire NAICS subsector rather than a named industry. Twenty of the 55 sit in manufacturing, NAICS 31 to 33, which is 36.4% of the list from one part of the economy. Twenty-nine entries carry a stated condition or exclusion, so the code alone does not settle them: food manufacturing is on the list if fuel tanks are present, textile mills are not required if the operation is sewing, weaving or hemming only, motor vehicle dealers are on the list if service bays are present, death care services are on it unless no embalming or cremation occurs at the property, and laundry and dry cleaning services are on it if dry cleaning operations have ever existed on site. That last clause is the whole discipline of a screen in one line: what matters is what has ever been there.

One footnote overrides the list entirely. A Phase I should always be obtained if the business sells, supplies or dispenses fuel, gasoline or heating oil, even where the NAICS code is not on the list. A screen that stops at code matching and does not read the fuel footnote will clear files it should have flagged, which is one reason the trigger determination belongs to a documented process rather than to a lookup. The wider question of what evidence the SBA file actually demands is covered in the evidence SOP 50 10 8 actually asks for, and the collateral consequences of a limited or special purpose finding run through special-purpose property risk in the loan data, since several of the Appendix 6 industries, gas stations and car washes among them, sit on both lists.

Two version notes belong in any current file. SOP 50 10 8 applies to applications issued an SBA loan number on or after 1 June 2025. A successor, SOP 50 10 8.1, carries a stated effective date of 1 October 2026 and keeps the same tiered structure, the same $250,000 threshold and the same Appendix 6 mechanism. Delegated lenders under PLP, SBA Express, Export Express and PCLP, and all 7(a) small loan lenders, cannot seek SBA approval of an environmental investigation and must document the file themselves, which raises rather than lowers the value of getting the trigger determination right at the front.

Outside the SBA file, the same duty arrives from the examiner

Most commercial files are not SBA files, and the duty is no lighter. It simply arrives through supervision instead of through a program manual.

The OCC's commercial real estate handbook directs banks to maintain an environmental risk management program reviewed and approved annually by the board, with risk thresholds based on property type, use and loan amount that determine when and what type of due diligence is required. The methods it contemplates run from borrower questionnaires, site visits and government records reviews at the light end to AAI-compliant evaluations by qualified professionals at the heavy end. It expects environmental reports to be received and evaluated before the bank finally commits to lend, and loan documents to carry rights of access sufficient to facilitate AAI-compliant evaluations (OCC, 2022).

The FDIC's examination manual is more specific about sequence. It directs examiners to verify that an institution performs an initial environmental risk analysis during the application process, covering present and past uses of the property, any agency contacts and, often, a documented site visit, and that it commissions a structured investigation by a qualified individual whenever the application, the interview or the visit indicates a possible concern. It expects review of public records and of federal and state lists of contaminated property. It states that failure to establish or comply with an appropriate environmental program is to be criticized (FDIC, 2023).

Read those two passages together and the public-records screen is not an optional efficiency. It is the documented first stage that both agencies describe, and the thing that distinguishes a defensible program from an undocumented habit is that the screen is written down: the layers queried, the radius searched, the date, and the result including the negative ones.

The radius is in the rule, and it is not one mile

Ask a credit team what radius an environmental database report covers and the answer is usually one mile. That is a vendor convention, not the rule. The rule is 40 CFR 312.26, and it assigns a different minimum search distance to each class of government record.

For the subject property itself, section 312.26(b) calls for records of reported releases and site investigation reports, records of activities or conditions likely to cause releases including landfill and disposal permits, storage tank records and permits, hazardous waste handler and generator records and permits, priority cleanup site listings and spill reporting records, CERCLIS records, public health records, Emergency Response Notification System records, and publicly available registries of engineering controls and of institutional controls including environmental land use restrictions.

For nearby and adjoining properties, section 312.26(c) sets the distances. NPL sites and tribal or state equivalents, one mile. RCRA facilities subject to corrective action, one mile. Federally registered or state permitted or registered hazardous waste sites identified for investigation or remediation, including state voluntary cleanup enrollees and listed brownfields sites, one half mile. Leaking underground storage tanks, one half mile. Delisted NPL sites, one half mile. Registries of engineering controls, one half mile. Former CERCLIS sites with no further remedial action notices, one half mile. Permitted landfills and solid waste management facilities, one half mile. RCRA small and large quantity generators, adjoining properties. Registered storage tanks, adjoining property.

The tiering is not arbitrary and it is not symmetric with the data. The two classes that reach a full mile are the rarest records in the federal system: EPA reported 1,343 sites on the National Priorities List and 37 proposed as of 13 March 2026. The classes that stop at the property next door are the most numerous: EPA's Facility Registry Service returned 1,607,599 facility records associated with RCRAInfo when queried on 24 August 2026. The record type with the widest required radius is scarcer than the record type with the narrowest by a factor of roughly 1,200 to 1.

That inversion is the rule working correctly. A National Priorities List site is a plume-scale problem that can plausibly reach a mile. A small quantity generator is a business that ships a drum of solvent twice a year, and its relevance to the parcel across the street is close to zero. A screen that applies one radius to every layer therefore fails in both directions at once: it under-searches the layers that migrate and floods the file with hits from the layers that do not. Section 312.26(d) lets the environmental professional modify a distance with a documented rationale, on the nature and extent of a release, geologic, hydrogeologic or topographic conditions, land use or development densities, property type, existing or past uses of surrounding properties, or potential migration pathways. That modification is the professional's call, not the screener's, which is another reason the screen produces an input to an engagement rather than a conclusion.

The tank record is the one that reaches a small-balance file

Superfund dominates the vocabulary of environmental risk and almost never touches a small-balance commercial loan. Tanks do. A dry cleaner, a filling station, a truck yard, a body shop, a rural motel with a heating oil tank: these are the properties that arrive in a community bank's pipeline, and their environmental history is a tank history.

EPA's Office of Underground Storage Tanks collects tank data from the states twice a year and publishes it by state and by EPA region. The mid-year fiscal 2026 report, covering 1 October 2025 through 31 March 2026, gives the national picture. There were 534,189 active UST systems at approximately 190,224 facilities as of March 2026, of which 532,628 were petroleum systems and 1,561 were hazardous substance systems. Against that, the file recorded 2,010,194 closed systems, 1,962,397 of them petroleum and 47,797 hazardous substance. For every active tank system in the country there are 3.8 closed ones.

The release history is on the same page and is the more arresting number. Cumulative confirmed releases since the program began in 1984 reached 583,313. Cumulative cleanups initiated reached 570,891 and cleanups completed 530,454, leaving 52,859 releases still to be cleaned up as of March 2026. In the six months of the reporting period alone, states confirmed 1,997 new releases and completed 2,947 cleanups.

Set those two facts side by side. There are 534,189 active tanks and 583,313 confirmed releases. The release count exceeds the active tank population, because it accumulates across four decades of tanks that have since been pulled out of the ground, while the active count is a snapshot. This is the structural reason a screen must query the closed and historical record rather than the current registry: a parcel that shows no active tank today may sit on top of three closed ones and a release that was closed under a state low-threat policy in 2003.

The state detail matters for portfolio work. California reported 37,000 active systems against 161,821 closed, a ratio of 4.4 to 1. New York reported 21,924 against 116,338, a ratio of 5.3 to 1. Texas, with the largest active population at 47,243, carries 132,882 closed. Georgia, with 30,445 active, carries only 55,282 closed, a ratio of 1.8 to 1. The spread reflects different closure histories, different reporting practices and different eras of commercial development, and it means the base rate a screen should expect for a hit is a state-level parameter rather than a national constant.

A listing is not a finding, and a compliant site is not a clean one

The single most common misreading of an environmental screen is treating a database hit as evidence of contamination. It is not. RCRAInfo carries hazardous waste handler activity, permits, corrective action and compliance records, and the overwhelming majority of its entries are notifications by businesses that generate small volumes of regulated waste in the ordinary course. A hit says a facility exists and filed a form. Whether anything was released is a separate question that the database does not answer.

The inverse error is subtler and more expensive. A facility that is registered, active and inspected is not thereby a facility being operated correctly. EPA's technical compliance rate measures the share of recently inspected federally regulated UST facilities in compliance with each requirement, and separately the share complying with all of them at once. Between April 2025 and March 2026 the national figures were 82.1% for spill prevention, 83.1% for overfill prevention, 90.7% for corrosion protection and 71.5% for release detection. The share of facilities in compliance with every technical category at the same time was 60.9%.

Nearly two in five recently inspected tank facilities were out of compliance with at least one technical requirement. Additional measures on the same report show 89.7% compliance with class A and B operator training, 90.4% with financial responsibility requirements and 83.2% with walk-through requirements. EPA notes on the table that states target inspections differently, some toward known non-compliant facilities and some at random, so the rate is a property of recently inspected facilities rather than a clean estimate of the whole population. Even read conservatively, it disposes of any inference that an operating, registered tank facility next door is a settled question.

The screening consequence is a rule about language. A screen reports what a record says and what its status field means, and it never converts a status into a conclusion. Listed is not contaminated. Closed is not remediated to current standards. No further action is a decision made under the standards and the risk assumptions in force on the date it was made. And absent from the registry means absent from that registry, on that day, at that address.

Where the federal record stops and the state record starts

Which brings us to the failure mode that costs the most and shows up in no error log: the layer that simply is not there.

EPA's Facility Registry Service is the natural starting point for a national screen because it links a facility to every program system that holds records about it, and EPA's Envirofacts gateway exposes it alongside the Superfund Enterprise Management System, RCRAInfo and the Toxics Release Inventory. Queried by program acronym on 24 August 2026, the registry held 1,607,599 facility records associated with RCRAInfo, 1,223,260 with the National Pollutant Discharge Elimination System, 46,677 with the brownfields ACRES system, 16,239 with the Superfund system, and 18,796 associated with underground storage tank programs.

Eighteen thousand tank records, in a country where EPA's own tank program counts 190,224 active tank facilities. The federal facility registry holds roughly 9.9% of the tank facilities the same agency reports through its state partners. Tank programs are implemented by the states, and the tank file lives in state systems.

A worked example makes the gap concrete. Queried by county name on 24 August 2026, the Facility Registry Service returned 18,868 San Francisco County records associated with RCRAInfo, 342 associated with the biennial hazardous waste report, 48 with water discharge permits, 25 with brownfields properties, 15 with Superfund, and zero associated with underground storage tank programs. A screen of a San Francisco parcel that queried only the federal registry would report no tanks in the county. California's own reporting for the same period counts 37,000 active and 161,821 closed tank systems statewide.

So the state layer is not a supplement to the federal layer for tanks. It is the layer. California runs its tank cleanup records through the State Water Resources Control Board's GeoTracker system alongside the UST cleanup fund and the low-threat case closure policy, with the Department of Toxic Substances Control maintaining the separate cleanup site database and the state maintaining the statutory list of hazardous waste sites that a property disclosure must reference. Every state has an equivalent set under different names, and their coverage, currency and search interfaces differ enough that a national screening procedure has to hold a per-state routing table rather than a single query.

Local overlays can bind harder than either. San Francisco Health Code Article 22A, the Maher Ordinance, is administered by the Department of Public Health's Site Mitigation Program, which regulates the characterization and mitigation of hazardous substances in soil, soil vapor and groundwater. The city states the criteria plainly: the ordinance applies where a project needs a building permit application, the project sits in the mapped Maher Area or is otherwise suspected to have contamination, and either the project will disturb 50 cubic yards of soil or it involves a change to a residential or sensitive use (San Francisco Department of Public Health, Site Mitigation Program, read 24 August 2026). Whether a given parcel falls inside the Maher Area is checked on the San Francisco Property Information Map, which also carries the state hazardous-sites layer alongside zoning, flood and seismic layers. A parcel can be clean in every federal database and still carry a mandatory soil investigation on the path to a building permit, which is a schedule risk and a budget risk on a construction file and is invisible to any national layer.

The hazard layers: flood, wetlands and wind

Environmental screening in a credit file rarely stops at contamination, because the same early pass is the natural place to read the physical hazard layers. They behave differently from the contamination layers in one important way: they are national, they are complete or nearly so, and their weakness is vintage rather than coverage.

Flood comes first because it is the only one with a statute attached. Under 42 U.S.C. 4012a(b)(1), a regulated lender may not make, increase, extend or renew a loan secured by improved real estate located in a FEMA-identified special flood hazard area where NFIP insurance is available unless the building and any personal property securing the loan carry flood insurance for the term of the loan in an amount at least equal to the outstanding principal balance or the maximum available coverage, whichever is less. The determination is recorded on FEMA's standard flood hazard determination form, identified as FEMA Form FF-206-FY-21-116 under OMB control number 1660-0040 (FEMA, 88 FR 66491, 2023), and 42 U.S.C. 4104b permits reliance on a previous determination not more than seven years old where its basis was set out on that form and no map revision has intervened.

The mapping layer behind it is FEMA's National Flood Hazard Layer, assembled from effective flood maps and Letters of Map Change, covering more than 90% of the U.S. population in digital form. Three data states have to be kept apart in any file: effective data governs, preliminary data is under review and appealable, and pending data carries a Letter of Final Determination but does not define NFIP minimum requirements until the six-month adoption and compliance period ends. The mechanics of the zone designations themselves, and the difference between a zone reading and a risk reading, sit in reading flood zone designations.

What a screen should not do is treat the flood zone as the flood risk. MMCG queried FEMA's OpenFEMA claim file on 24 August 2026: it held 2,724,656 National Flood Insurance Program claim records, of which 2,577,804 carried one of the enumerated rated flood zone codes. Grouping those records puts 1,795,908 in the A zones and 90,476 in the V zones, both inside the special flood hazard area, against 407,535 in zone X, 164,299 in zone C and 115,248 in zone B, all three of which sit outside it. That is 687,082 claims, 26.7% of the zoned file, paid on policies rated in zones where no mandatory purchase requirement applies.

Two caveats belong with that figure and neither erases it. The rated flood zone is the zone used to rate the policy, which can differ from the zone on the current effective map, and the file counts claims rather than buildings, so a repeatedly flooded property appears repeatedly. What survives both caveats is the underwriting point: a parcel outside the special flood hazard area is a parcel outside a purchase mandate, not a parcel outside flood risk, and the collateral in a zone X commercial file is frequently uninsured for a peril that the federal claim record shows arriving there in volume.

The same file carries a second warning for commercial work. Non-residential occupancy codes account for 231,424 of the 2,724,656 claim records, 8.5% of the file, and only 17,218 records carry the small business indicator. The public flood-loss record is overwhelmingly residential, so it is thin evidence for a commercial property's loss history and cannot be read as a commercial frequency estimate. Compounding this, 42 U.S.C. 4013(b)(4) caps NFIP coverage for a non-residential building at $500,000, with $500,000 more available for the owner's contents and $500,000 for tenant contents. A commercial building worth more than that carries uninsured exposure above the cap even when the mandatory purchase requirement is satisfied in full, which is a collateral question rather than a compliance question and belongs in the credit memorandum rather than in the flood file.

Wetlands are the second layer and the one most often over-read. The U.S. Fish and Wildlife Service's National Wetlands Inventory carries more than 37 million wetland and deepwater features and is updated twice a year. Its own metadata is explicit about what it is not: the data are intended for use with base maps at 1:12,000 or smaller, the primary intended use is regional and watershed display and analysis rather than specific project analysis, and the data should not be displayed at scales smaller than 1:250,000. The Service states that agencies with jurisdiction over wetlands may define them differently from the inventory and that no attempt is made to define the limits of anyone's regulatory jurisdiction. Jurisdiction narrowed in 2023 when the Supreme Court held in Sackett v. EPA that the Clean Water Act reaches adjacent wetlands only where they have a continuous surface connection to covered waters, and the Corps definitions at 33 CFR 328.3 were amended to conform. An approved jurisdictional determination under 33 CFR 331.2 is a Corps document stating the presence or absence of waters of the United States on a parcel and is an appealable action; a preliminary determination is advisory. An inventory polygon is a screening flag that says get a delineation, and nothing more, which is the discipline set out in NWI wetlands data in site diligence and extended to national scale in mapping wetlands constraints at national scale.

Wind and hail are the third, and the record has a shape problem rather than a scale problem. NOAA's Storm Events Database recorded only tornado events from 1950 to 1954, added thunderstorm wind and hail from 1955 to 1995, and expanded to its current 48 defined event types from 1996 under National Weather Service Directive 10-1605. Magnitude is populated only for wind speed, in knots, and hail size, in inches, with a type code distinguishing estimated gusts from measured ones. NOAA states plainly that the database is a partial record built on voluntary reporting and that the Weather Service does not guarantee its accuracy. A screen built on it counts reports, not events, and the defensible output is reported events per area per decade from 1996 forward, never a design wind speed and never a trend spanning the scope changes. The full treatment is in NOAA storm event data for hazard screening. Terrain and soils belong in the same early pass where earthwork or drainage will drive the budget, and the product-resolution trap there is set out in screening buildable land at national scale.

How a screen scopes the Phase I engagement

The output of all of this is not a conclusion. It is a scope of work, and it is worth being precise about what goes into it.

Four findings change an engagement. The first is a NAICS trigger: if the current or a known prior use matches Appendix 6, or if the business dispenses fuel in any form, the SBA file starts at a Phase I and there is nothing to decide. The second is a historical use that the current record does not show: a closed tank, a former dry cleaner two doors down, a filled watercourse under the parking field. Those are the facts that turn a routine assessment into one with a targeted historical component, and they are exactly what the closed-record layers are for. The third is an adjoining-property hit inside the tight radii, since RCRA generators and registered tanks are searched only to the adjoining property under 312.26(c) and a generator on the shared lot line is a materially different fact from one three blocks away. The fourth is a local process trigger like the Maher Area, which changes the construction timeline rather than the assessment scope.

What the screen hands the environmental professional is a documented starting point: the addresses and parcel identifiers used, the layers queried with their vintages, the radius applied to each, the hits with their status fields and dates, and the negatives with the radius they were negative within. That last item is the one most often omitted and the one that carries the most weight later, because a negative without a radius is not a finding at all.

Timing then follows the rule rather than the habit. Under 40 CFR 312.20(a), all appropriate inquiries must be conducted within one year prior to the date of acquisition. Under 312.20(b), five components must be conducted or updated within 180 days of and prior to acquisition: interviews with past and present owners, operators and occupants; searches for recorded environmental cleanup liens; reviews of federal, tribal, state and local government records; visual inspections of the facility and adjoining properties; and the environmental professional's declaration. Those two windows are the origin of the one-year and 180-day shelf-life conventions. Note what they are measured to: the acquisition date. A report ordered for a refinancing rather than an acquisition sits outside the rule's timing structure entirely, which is why SBA imposes its own clock, requiring environmental investigation reports to be dated within one year of issuance of the SBA loan number, and why a lender relying on a borrower's older report needs the Appendix 5 reliance letter to reach it at all.

Everything the screen produces then goes into the file the way any other analytic input does, with a source, a date and a method, which is the standard argued in why every number needs a source and a date. The wider public-records set that surrounds the environmental layer, zoning, assessor records, access and the rest, is laid out as a data-first due diligence checklist, and the physical constraint arithmetic that follows a clean environmental read is in estimating buildable area from a parcel.

Six ways the screen is wrong

A screening procedure earns trust by publishing its own failure modes. These are the six that matter.

The address is not the parcel. Every layer discussed here is joined by geography, and the join is where silent error enters. A rooftop geocode that lands on the wrong side of a street moves a parcel out of an adjoining-property search; a centroid geocode on a large industrial parcel can move it hundreds of feet. The error rates, and the difference between a match, a tie and a no-match, are set out in geocoding accuracy in CRE analysis. A screen that reports a clean result without reporting the geocode quality that produced it is reporting an untested claim.

The status field is stale. Registry status changes at the pace of state administrative process, not at the pace of a transaction. A case shown as open may have closed; a case shown as closed may have reopened on new data. Every status carries the date the record was refreshed, and that date belongs in the file next to the status.

Listed is not contaminated, and unlisted is not clean. This runs in both directions and the second direction is worse. A property that never triggered a reporting obligation appears in no registry no matter what happened on it. Pre-regulatory disposal, farm dumps, small solvent operations that closed before RCRA notification, and heating oil tanks that were never federally regulated leave no federal trace at all.

Vapor and offsite migration are invisible to a radius search. A radius search finds records at locations. It does not model subsurface transport, and vapor intrusion in particular is a pathway that the presence or absence of a nearby record does not settle. Groundwater flow direction can put a plume under a parcel from a source that a distance-ranked list buries on page four. Section 312.26(d) exists precisely because distance alone is a poor proxy for pathway, and the person authorised to make that call is the environmental professional.

The layer may not exist for that state. The tank example above is the clearest case: 9.9% federal coverage of active tank facilities means the national layer is close to useless for the single most common small-balance exposure, and the state layer has to be reached directly. Any national screening claim that does not name its per-state sources is claiming coverage it does not have.

The source itself can disappear. Federal screening tools move, are retired or return errors, and a checklist written two years ago can name a tool that no longer answers. During the research for this library, EPA's EJScreen returned HTTP 404 at both its primary and former addresses, and several federal hosts refused automated requests entirely on the days they were tried. A screening procedure should verify that each cited source resolves on the day it is run, and record that it did.

Above all of these sits the limit that is not a failure mode but a boundary: no public-records screen, however carefully built, supports an All Appropriate Inquiries defense. The defense comes from an AAI-compliant Phase I performed by an environmental professional as defined at 40 CFR 312.10, and from nothing else.

What negative information is worth

The counterintuitive part of this work is that its product is mostly absence. A good screen usually ends with a list of layers queried, a set of radii, a date, and no hits. That reads like nothing happened, and it is the most useful output the exercise produces.

It is useful for three reasons. It scopes the engagement: an assessment ordered against a documented clean screen is a routine assessment with a known timeline, while one ordered blind carries an open-ended tail. It dates the knowledge: a negative with a radius and a date can be re-run later and compared, and a difference between two runs is a signal, whereas an undocumented negative can never be re-tested. And it sizes the residual: knowing that the tank layer for a given state carries a 4 to 1 ratio of closed to active systems tells a credit officer what a clean current-registry result actually rules out, which is much less than it appears to.

The distribution of the data carries a lesson of its own. The environmental record is densest exactly where regulation has been most active, which is to say where risk is best documented rather than where it is highest. A parcel in a state with a mature tank program and a public cleanup database will produce more hits than an identical parcel in a state with thinner reporting, and the difference measures the programs, not the ground. That asymmetry is the strongest argument for treating the screen as scoping information rather than as a risk score, and for keeping the environmental professional, whose judgment the rule actually vests, at the center of the decision.

None of this replaces the assessment. The screen tells a lender what to expect, what to ask for and what to pay attention to. The Phase I, performed to ASTM E1527-21 by an environmental professional under 40 CFR part 312, is what the file eventually rests on, and the decision rests with the lender and that professional. The screen's contribution is that by the time the report is commissioned, nobody is guessing about its scope.

Frequently asked questions

What is an environmental screening for a commercial property?

It is a review of public environmental and hazard records for a parcel and its surroundings, run from published federal, state and local data before any consultant is engaged. A working screen reads EPA facility and cleanup records, the state tank and cleanup registries, FEMA flood mapping, wetlands mapping and storm records, and it records for each layer the source, the vintage, the radius searched and the result. Its purpose is to tell a lender what a formal assessment is likely to find so the engagement can be scoped, not to reach a conclusion about contamination.

Is an environmental screen the same as a Phase I environmental site assessment?

No, and treating it as one is the most expensive mistake in this workflow. A Phase I is defined by ASTM E1527-21 as referenced at 40 CFR 312.11, and it must be performed by an environmental professional as defined at 40 CFR 312.10, which requires a professional engineer's or geologist's license plus three years of relevant experience, a government license or certification plus three years, a science or engineering degree plus five years, or ten years of relevant experience. A public-records screen performed by anyone else contributes nothing to the All Appropriate Inquiries defense under CERCLA. The screen precedes the Phase I and scopes it. It never replaces it.

What triggers a Phase I environmental site assessment on an SBA loan?

Under SOP 50 10 8, the lender first makes a good faith determination of the NAICS codes for the property's current and known prior uses and compares them with Appendix 6, the list of environmentally sensitive industries. A match requires the investigation to begin with a Phase I regardless of loan amount. With no match, a loan up to and including $250,000 may begin with an environmental questionnaire, and a loan above $250,000 must begin with a questionnaire plus a records search with risk assessment. Any risk assessment result other than low risk escalates to a Phase I, as does a transaction screen or questionnaire indicating further inquiry is warranted. A footnote to the appendix requires a Phase I whenever the business sells, supplies or dispenses fuel, gasoline or heating oil, whatever its NAICS code.

How far around a property does an environmental records search reach?

It depends on the record, and 40 CFR 312.26(c) sets the minimums. NPL sites and tribal or state equivalents and RCRA facilities subject to corrective action are searched to one mile. Hazardous waste sites identified for investigation or remediation, leaking underground storage tanks, delisted NPL sites, engineering control registries, former CERCLIS sites with no further remedial action notices, and permitted landfills and solid waste management facilities are searched to one half mile. RCRA small and large quantity generators and registered storage tanks are searched only to adjoining properties. Section 312.26(d) allows the environmental professional to modify a distance with a documented rationale based on the nature of a release, geology, land use density, property type, surrounding uses or migration pathways.

Can you check environmental records for a commercial property for free?

Yes, for a large share of the federal record. EPA's Envirofacts gateway and its Facility Registry Service expose facility records linked to Superfund, RCRAInfo, the Toxics Release Inventory and other program systems, and EPA's Office of Underground Storage Tanks publishes national and state tank and release measures twice a year. FEMA publishes the National Flood Hazard Layer and the redacted National Flood Insurance Program policy and claim files, the Fish and Wildlife Service publishes the National Wetlands Inventory, and NOAA publishes the Storm Events Database. What is not free of effort is the state layer, because tank and cleanup programs are state-run and each state publishes them differently.

How long is a Phase I environmental site assessment valid?

The rule does not give a report an expiry date; it puts two clocks on the inquiry. Under 40 CFR 312.20(a), all appropriate inquiries must be conducted within one year prior to the date of acquisition. Under 312.20(b), five components must be conducted or updated within 180 days of and prior to acquisition: interviews with past and present owners, operators and occupants, searches for recorded environmental cleanup liens, reviews of government records, visual inspections of the property and adjoining properties, and the environmental professional's declaration. Both windows are measured to an acquisition date, so a report ordered for a refinancing sits outside the structure entirely. SBA imposes its own clock instead, requiring environmental investigation reports to be dated within one year of issuance of the SBA loan number.

Does a flood zone determination tell you a property's flood risk?

It tells you whether a purchase mandate applies, which is a narrower question. MMCG queried FEMA's redacted National Flood Insurance Program claim file on 24 August 2026 and grouped the 2,577,804 records carrying a rated flood zone: 687,082 of them, 26.7%, were rated in zones B, C or X, which sit outside the special flood hazard area where no mandatory purchase requirement applies. The rated zone can differ from the zone on the current effective map and the file counts claims rather than buildings, but the direction is unambiguous. A commercial property outside the mapped flood zone is outside a rule, not outside a risk, and NFIP coverage for a non-residential building is capped at $500,000 in any case.

What does it mean when a property appears in an EPA database?

Usually that a facility exists there and filed a form. RCRAInfo carries hazardous waste handler activity, permits, corrective action and compliance records, and most entries are routine notifications by businesses that generate small volumes of regulated waste. A listing is not evidence of a release, and the absence of a listing is not evidence of a clean site, since anything that closed before a reporting obligation attached leaves no federal trace. The status field and its refresh date matter more than the hit itself, and the interpretation belongs to an environmental professional rather than to the screen.

Sources

1. Code of Federal Regulations, Title 40, Part 312, Section 312.10, Definitions (Cornell Legal Information Institute), read 24 August 2026: the definition of environmental professional and its four qualification pathways, the definition of relevant experience, and the definitions of abandoned property and adjoining properties. https://www.law.cornell.edu/cfr/text/40/312.10
2. Code of Federal Regulations, Title 40, Part 312, Section 312.11, References (Cornell Legal Information Institute), read 24 August 2026: ASTM E1527-21 and ASTM E2247-23 as the standards satisfying All Appropriate Inquiries, with the transition clauses expiring 13 February 2024 and 24 June 2025. https://www.law.cornell.edu/cfr/text/40/312.11
3. Code of Federal Regulations, Title 40, Part 312, Section 312.20, All appropriate inquiries (Cornell Legal Information Institute), 2026 current text: the one-year window before acquisition and the five components requiring conduct or update within 180 days. https://www.law.cornell.edu/cfr/text/40/312.20
4. Code of Federal Regulations, Title 40, Part 312, Section 312.26, Reviews of Federal, State, Tribal, and local government records (GovInfo, 2025 annual edition XML, revised as of 1 July 2025), read 24 August 2026: the subject-property record classes at paragraph (b), the ten nearby and adjoining record classes with their minimum search distances at paragraph (c), and the modification provision at paragraph (d). https://www.govinfo.gov/content/pkg/CFR-2025-title40-vol30/xml/CFR-2025-title40-vol30-sec312-26.xml
5. U.S. Small Business Administration, SOP 50 10 8, Lender and Development Company Loan Programs with Technical Updates, effective 1 June 2025: Section A, Chapter 5, Paragraph E, Environmental Policies and Procedures; Appendix 4, Environmental Definitions; Appendix 5, Reliance Letter; Appendix 6, NAICS Codes of Environmentally Sensitive Industries; Appendix 7, Gas Station Loans. Document text extracted locally after fetch and read 24 August 2026. https://legacy.sba.gov/sites/default/files/2025-05/SOP%2050%2010%208%20Technical%20Updates%20effective%206.1.2025.docx
6. U.S. Small Business Administration, SOP 50 10 8.1, Lender and Development Company Loan Programs, with a stated effective date of 1 October 2026, posted August 2026: the same tiered environmental investigation, the $250,000 threshold and the Appendix 6 mechanism carried forward. https://legacy.sba.gov/sites/default/files/2026-08/SOP%2050%2010%208.1%20effective%2010.1.2026_0.docx
7. U.S. Small Business Administration, Procedural Notice 5000-866054, Update to SOP 50 10 7.1 Section A, Chapter 5, Paragraph E, Environmental Policies and Procedures, effective 20 March 2025: the one-year report dating rule, the E-Tran certification step, and the statement that delegated and 7(a) small loan lenders cannot seek SBA approval of an environmental investigation. https://legacy.sba.gov/sites/default/files/2025-03/Procedural%20Notice%20-%20Environmental%20Review%20Process%20%281%29.pdf
8. U.S. Environmental Protection Agency, Office of Underground Storage Tanks, Semiannual Report of UST Performance Measures, Mid Fiscal Year 2026 (1 October 2025 to 31 March 2026), published May 2026, read 24 August 2026: 534,189 active UST systems at approximately 190,224 facilities; 2,010,194 closed systems; 583,313 cumulative confirmed releases, 570,891 cleanups initiated and 530,454 cleanups completed since 1984; 52,859 releases remaining; the technical compliance rate components and the additional compliance measures; and the state and EPA region tables. https://www.epa.gov/system/files/documents/2026-06/performance-measures-mid-year-fy-26.pdf
9. U.S. Environmental Protection Agency, UST Performance Measures program page, read 24 August 2026: the current semiannual report, the definitions of the measures, and the archive of reports back to fiscal year 2005. https://www.epa.gov/ust/ust-performance-measures
10. U.S. Environmental Protection Agency, Envirofacts data service, Facility Registry Service program facility counts by program system acronym and by county name, retrieved 24 August 2026: national counts of 1,607,599 records associated with RCRAInfo, 1,223,260 with the National Pollutant Discharge Elimination System, 46,677 with the brownfields ACRES system, 18,796 with underground storage tank programs and 16,239 with the Superfund system, and the San Francisco County counts. https://data.epa.gov/efservice/
11. U.S. Environmental Protection Agency, Envirofacts, About the Data, 2026: the program systems the gateway exposes, including the Superfund Enterprise Management System, RCRAInfo and the Toxics Release Inventory. https://www.epa.gov/enviro/about-data
12. U.S. Environmental Protection Agency, Current NPL Updates: New Proposed NPL Sites and New NPL Sites, read 24 August 2026: 1,343 sites listed on the National Priorities List and 37 proposed as of 13 March 2026. https://www.epa.gov/superfund/current-npl-updates-new-proposed-npl-sites-and-new-npl-sites
13. U.S. Environmental Protection Agency, Brownfields All Appropriate Inquiries program page, read 24 August 2026: ASTM E1527-21 and E2247-23 as currently accepted standards, the 2005 final rule effective 1 November 2006, and the one-year and 180-day timing provisions. https://www.epa.gov/brownfields/brownfields-all-appropriate-inquiries
14. U.S. Environmental Protection Agency, EJScreen page, HTTP status observed 24 August 2026: the address returned HTTP 404, as it did on 22 August 2026, so the tool cannot be cited as a live federal screening resource. https://www.epa.gov/ejscreen
15. Federal Deposit Insurance Corporation, Risk Management Manual of Examination Policies, Section 3.2 Loans, dated May 2023, pages 3.2-44 to 3.2-46: the environmental risk program expectation, the initial environmental risk analysis during the application process, and the statement that a bona fide prospective purchaser protection reached through an AAI-compliant evaluation applies under CERCLA only and not under RCRA including underground storage tank liability. https://www.fdic.gov/regulations/safety/manual/section3-2.pdf
16. Office of the Comptroller of the Currency, Comptroller's Handbook, Commercial Real Estate Lending, version 2.0, 2022, pages 65 to 67: the CERCLA section 101(20) secured creditor exemption, the statement that the exemption does not protect the bank from the decline in collateral value contamination causes, and the elements of an environmental risk management program approved annually by the board. https://www.occ.gov/publications-and-resources/publications/comptrollers-handbook/files/commercial-real-estate-lending/pub-ch-commercial-real-estate.pdf
17. 42 U.S.C. 4012a, Flood Disaster Protection Act mandatory purchase provision (Cornell Legal Information Institute), 2026 current text: the requirement that a regulated lender not make, increase, extend or renew a loan secured by improved real estate in a special flood hazard area without coverage at least equal to the outstanding principal balance or the maximum available coverage, whichever is less. https://www.law.cornell.edu/uscode/text/42/4012a
18. 42 U.S.C. 4013, Limits of coverage (Cornell Legal Information Institute), 2026 current text: section 4013(b)(4) sets $500,000 aggregate liability for a non-residential building, with $500,000 for the building owner's contents and $500,000 for tenant contents. https://www.law.cornell.edu/uscode/text/42/4013
19. 42 U.S.C. 4104b, Standard Flood Hazard Determination Form and reliance on previous determinations (Cornell Legal Information Institute), 2026 current text: reliance on a previous determination not more than seven years old whose basis was set out on the standard form, absent an intervening map revision. https://www.law.cornell.edu/uscode/text/42/4104b
20. Federal Emergency Management Agency, National Flood Hazard Layer, page last updated 3 April 2025 (browser read; the host refuses automated clients): the layer is built from effective flood maps and Letters of Map Change and covers more than 90 percent of the U.S. population in digital form, with effective, preliminary and pending data published as distinct states. https://www.fema.gov/flood-maps/national-flood-hazard-layer
21. Federal Emergency Management Agency, Agency Information Collection Activities: Standard Flood Hazard Determination Form, 88 FR 66491, 27 September 2023: the form is FEMA Form FF-206-FY-21-116, formerly 086-0-32, under OMB control number 1660-0040. https://www.govinfo.gov/content/pkg/FR-2023-09-27/pdf/2023-20982.pdf
22. Federal Emergency Management Agency, OpenFEMA, NFIP Redacted Claims v3 and NFIP Redacted Policies v3, record counts retrieved from the OpenFEMA API on 24 August 2026: 2,724,656 claim records and 74,349,528 policy records; claim counts by rated flood zone, by occupancy type and by small business indicator; field descriptions from the OpenFEMA data set field metadata. https://www.fema.gov/api/open/v3/NfipClaims
23. U.S. Fish and Wildlife Service, National Wetlands Inventory, Wetlands Data Layer and dataset metadata, 2026: more than 37 million wetland and deepwater features released twice a year; the 1:12,000 intended use scale, the 1:250,000 display floor, and the statement that the inventory does not define the limits of any government's regulatory jurisdiction. https://www.fws.gov/program/national-wetlands-inventory/wetlands-data
24. Sackett v. Environmental Protection Agency, No. 21-454, decided 25 May 2023, slip opinion syllabus: adjacent wetlands are covered by the Clean Water Act only where they have a continuous surface connection to covered waters. https://www.supremecourt.gov/opinions/22pdf/21-454_4g15.pdf
25. Code of Federal Regulations, Title 33, Sections 328.3 and 331.2, Corps of Engineers regulatory definitions (Cornell Legal Information Institute), 2023 conforming amendment at 88 FR 61968, current text: the definition of waters of the United States and the distinction between an approved jurisdictional determination, which is appealable, and a preliminary determination, which is advisory. https://www.law.cornell.edu/cfr/text/33/328.3
26. National Oceanic and Atmospheric Administration, National Centers for Environmental Information, Storm Events Database documentation and Storm Data Bulk CSV Format citing NWS Directive 10-1605, read 24 August 2026: tornado events only from 1950 to 1954, tornado, thunderstorm wind and hail from 1955 to 1995, and 48 defined event types from 1996; magnitude recorded only for wind speed in knots and hail size in inches with an estimated or measured type code; and NOAA's statement that the database is a partial record built on voluntary reporting. https://www.ncei.noaa.gov/stormevents/
27. California State Water Resources Control Board, Underground Storage Tank Program, read 24 August 2026: the GeoTracker public system for leaking underground storage tank and cleanup case records, the UST cleanup fund and the low-threat UST case closure policy. https://www.waterboards.ca.gov/water_issues/programs/ust/
28. San Francisco Department of Public Health, Site Mitigation Program, Check your conditions, read in a browser 24 August 2026: the Maher Ordinance under Health Code Article 22A applies where a project needs a building permit application, sits in the Maher Area or is otherwise suspected to have contamination, and either disturbs 50 cubic yards of soil or involves a change to a residential or sensitive use. https://www.sf.gov/departments--site-mitigation/find-out-if-you-should-submit-an-application/
29. San Francisco Planning Department, Property Information Map, 2026: per-parcel zoning, height and bulk districts, assessor summary and building records, permits, flood risk and seismic hazard layers, and the state hazardous sites layer. https://sfplanninggis.org/pim/
30. MMCG Research, Environmental and Hazard Screening Series, 2026: the tabulation of the 55 entries in SOP 50 10 8 Appendix 6 by NAICS sector, code length and condition; the closed-to-active tank ratios computed from the EPA state tables; the circle areas computed from the 40 CFR 312.26 search distances; the Facility Registry Service counts retrieved by program acronym and by county name; and the NFIP claim groupings by rated flood zone and occupancy class. All computed 24 August 2026 from sources 1 to 29; MMCG database, 2026. https://mmcganalytics.com/methodology/