The 30-Minute Pre-Term-Sheet Site Screen

A term sheet is the first expensive thing a lender does. Everything before it is conversation; everything after it is commitment, and the cost of withdrawing rises with every day the borrower spends believing the deal is real. The screen described here sits in that gap. It takes about thirty minutes, it uses only public records, and its purpose is not to approve the site or to clear it. Its purpose is to find, before anyone spends money, the facts that would change the structure, the pricing conversation, or the answer.

The thirty minutes is not a stunt. It is the amount of time it takes one analyst to open six public layers for a single parcel and write down what they say. The layers are the same every time, which is what makes the exercise a screen rather than a browse, and the discipline is that each layer produces a written finding with a source and a date, even when the finding is that the layer is silent.

What a screen is for, and what it is not

A site screen is not a feasibility study, not a market study, and not an appraisal. It does not produce an opinion of value, a conclusion about market support, or a recommendation. It answers a narrower question: does the public record contain anything about this location that a credit committee would want to know before the institution puts a term sheet in writing?

The distinction matters because the three formal documents each arrive later, cost real money, and are commissioned once the deal has momentum. By the time an appraisal is ordered, the institution has usually decided it wants the loan. A screen that runs before the term sheet is the last cheap moment to discover that the collateral sits in a floodway, that the parcel carries a recorded environmental lien, or that the use the borrower describes is not the use the zoning permits.

What the screen produces is a page. Six findings, each with the source and the date it was read, and a single line at the end that says proceed, proceed with a named condition, or stop. That page goes into the credit file, and it is the reason the file can later show that the institution looked before it committed. The way analytics enter the credit file is the subject of this pillar, and the screen is its earliest artifact.

The order of operations

The sequence is not arbitrary. The layers are ordered so that the cheapest disqualifying finding arrives first, which is what keeps the median screen well under thirty minutes.

Collateral arithmetic comes first because it needs no external lookup and can end the exercise on its own. Flood comes second because it is the fastest federal lookup that carries a mandatory consequence: a special flood hazard area designation triggers an insurance requirement by statute, and that requirement is not negotiable by either party. Environmental comes third because it is the layer most likely to produce a finding that changes the structure rather than the price, and because a hit here determines whether the fourth and fifth layers are worth running at all. Hazard history, zoning and access follow, in that order, because each is more likely to shape the terms than to end the discussion.

Two rules hold the sequence together. The first is that a layer is never skipped because an earlier layer was clean, since the point of a fixed procedure is that the analyst cannot quietly stop at the first comfortable answer. The second is that a silent layer is written down as silent, with the source and the date, because "the database returned no records for this parcel on this date" is a finding, and it is the finding that a later reviewer will most want to see stated rather than assumed.

Layer one: the collateral arithmetic that constrains everything after it

Before any hazard layer, the screen establishes what the loan can be. The Interagency Guidelines for Real Estate Lending Policies, codified for national banks at Appendix A to Subpart D of 12 CFR Part 34 and for state nonmember banks in identical text at 12 CFR Part 365, set supervisory loan-to-value limits by collateral type: 65% for raw land, 75% for land development, 80% for construction of commercial, multifamily and other nonresidential property, 85% for one-to-four family residential construction, and 85% for improved property.

These are supervisory limits, not underwriting targets, and the guidelines contemplate loans above them within an aggregate that the institution monitors. The screen's job is simply to know which line applies, because the answer changes with what the borrower intends to do. A parcel described as "land we will build on" is raw land at 65% until entitlements exist, land development at 75% while horizontal work proceeds, and construction at 80% once vertical work begins. Three different loans, three different advance rates, one parcel, and the difference is a matter of documented status rather than intention.

The second piece of arithmetic is coverage, and at screening depth it is done with the same definitions an examiner would use. The OCC's Comptroller's Handbook defines the debt-service coverage ratio in its glossary as cash flow or net operating income divided by the debt service, and debt yield as the ratio of net operating income to debt, expressed as a percent. Both can be computed from the borrower's own stated numbers in a few minutes, and the purpose at this stage is not to underwrite but to see whether the deal is close to the line or far from it.

The handbook is also explicit that a single point estimate is not the test. It directs that as real estate income and prices rise in periods of economic growth, capitalization rates, interest rates and debt-service coverage ratios should be stress-tested to determine whether a property will likely remain viable during a period of economic stress, and that cash-flow analysis should include stress testing for sensitivity to changing conditions under a variety of scenarios, naming absorption rates, interest rates and capitalization rates. It notes as well that a net operating income analysis may assume market vacancy rates above or below actual vacancy, and expenses that are not an actual or immediate cash expense, such as management fees and reserves for capital replacements. A screen that records the borrower's coverage at their assumptions, and again at a rate one or two hundred basis points higher, has produced the single most useful number on the page.

The arithmetic is worth doing in the first five minutes because it is the constraint that cannot be relaxed by a better story. If the borrower needs proceeds that imply 85% against raw land, the screen has already found the deal's binding problem, and no amount of environmental or flood clearance will move it.

Layer two: flood, and the three states a map can be in

Flood is the layer most often screened badly, because the question is usually asked as though a parcel is either in a flood zone or not. The National Flood Hazard Layer is FEMA's geospatial database of current effective flood hazard data, assembled from effective flood maps and the Letters of Map Change delivered to communities, and FEMA states that it covers over 90% of the United States population and can be used in place of the Flood Insurance Rate Map for National Flood Insurance Program purposes with appropriate care.

The regulatory definitions sit at 44 CFR 59.1, which defines an area of special flood hazard as land subject to a 1% or greater chance of flooding in any given year, with zone designations A, AO, AH, AE, AR, VO and V. The shallow flooding zones, designated AO, AH, AR/AO, AR/AH and VO, describe areas with that same annual chance where flooding averages one to three feet deep and a clearly defined channel does not exist. A coastal high hazard area, the V zone, extends from offshore to the inland limit of a primary frontal dune along an open coast and is subject to high velocity wave action from storms or seismic sources. Under 44 CFR 64.3, a Flood Hazard Boundary Map is issued before a detailed study and a Flood Insurance Rate Map after it, and mandatory flood insurance purchase attaches in the listed zones.

The part that a screen must get right is that an effective map is only one of three states. Preliminary flood hazard data gives an early view of projected risk and passes through a formal review period in which the determinations may be appealed with better data. Pending FIRM databases have already been designated final by a Letter of Final Determination and are scheduled to become effective within six months, but FEMA states plainly that they do not define the minimum requirements for NFIP purposes until they become effective at the end of that six-month adoption and compliance period. A screen that reports the pending zone as the governing zone has reported a map that is not yet in force. A screen that ignores the pending zone entirely has missed the fact that the borrower's insurance cost is about to change.

The mandatory purchase requirement itself comes from 42 U.S.C. 4012a, which bars federal agencies from approving financial assistance for acquisition or construction in a special flood hazard area unless the property carries flood insurance, and requires regulated lenders to prohibit loans secured by improved real estate in high-risk zones without adequate coverage for the loan term. The determination is documented on the Standard Flood Hazard Determination Form under 42 U.S.C. 4104b, and the lending rules sit at 12 CFR Part 22 and its sister codifications. A borrower or lessee who believes the map is wrong has a route: 44 CFR Part 70 gives the owner or lessee of property the right to submit scientific or technical data challenging the map's accuracy, and the Administrator may issue a Letter of Map Amendment in response.

What makes flood tractable in thirty minutes is that the federal record is unusually complete and unusually well documented. It is also unusually easy to misdate, which is the recurring theme of this screen and of the provenance standard the rest of this library argues for.

Layer three: the environmental record you can reach for free

The environmental layer is where a thirty-minute screen earns its keep, because the free federal record is far denser than most credit teams assume. EPA's Envirofacts is the agency's facility-focused gateway to program data, exposing the Superfund Enterprise Management System for sites addressed under CERCLA, RCRAInfo for hazardous waste handler activity, permits, corrective action and compliance, and the Toxics Release Inventory for facility and chemical release data.

Queried through the Facility Registry Service on 23 August 2026, the scale of that record is concrete. The registry held 1,607,599 facility records associated with RCRAInfo, 1,223,260 associated with the National Pollutant Discharge Elimination System, 75,629 associated with the Toxics Release Inventory, 18,796 associated with underground storage tank programs, and 16,239 associated with the Superfund Enterprise Management System. A screen does not read those files. It asks one question of them: does anything sit on this parcel, or close enough to it to matter?

State records frequently matter more than federal ones, because the liabilities that most often reach a small-balance commercial file are tanks rather than Superfund sites. In California, the State Water Resources Control Board's Underground Storage Tank Program publishes GeoTracker, the state's public system for leaking underground storage tank and cleanup case data, alongside the UST Cleanup Fund and the Low-Threat UST Case Closure Policy. Local overlays can be sharper still. San Francisco Health Code Article 22A, the Maher Ordinance, requires an applicant for a building or grading permit that disturbs 50 cubic yards or more of soil inside the mapped Maher Area to apply to the Department of Public Health's Site Assessment and Mitigation Program, and the Maher Area is drawn to include parcels with current or historical industrial use or zoning. A screen that checks only federal databases in a city with an ordinance like that one will report a clean parcel that is not clean of process.

One caution belongs in any current screen. EPA's EJScreen was not reachable on EPA's website when this research was run: both the tool page and its former subpage returned HTTP 404 on 22 August 2026. A screening procedure that names EJScreen as a live federal tool is describing something that did not answer that day, which is a reminder to check that a cited source still resolves rather than inheriting a checklist written when it did.

Layer four: hazard history, and what the storm record does not contain

Wind and hail exposure is a real credit question for single-story commercial buildings with large roof areas, and the public record for it is the NOAA National Centers for Environmental Information Storm Events Database. Its coverage history is the caveat that governs its use. Only tornado events were recorded from 1950 to 1954. Tornado, thunderstorm wind and hail events were added from 1955 to 1995, initially compiled from paper publications. The database expanded to its current 48 defined event types starting in 1996, under National Weather Service Directive 10-1605, and the permitted event types are the table reproduced in the Storm Data bulk format documentation.

The practical consequence is that any long-run frequency computed from this database is a frequency of reported events, and the reporting regime changed twice. A screen that says "this county has recorded 41 severe hail events since 1996" is making a defensible statement. A screen that says "hail frequency here has tripled since 1980" is describing a change in the recording rules at least as much as a change in the weather. The same discipline governs the honest use of state traffic counts, where coverage cycles and factoring rules shape the series as much as traffic does.

Layer five: zoning, access and the buildable envelope

The fifth layer asks whether the use the borrower describes is the use the site permits, and whether the building they describe fits. There is no federal zoning dataset, which is the single most important thing to know about this layer: zoning is municipal, published in municipal code, and inconsistently machine readable, so the screen reads the code rather than a national layer. What the parcel record supplies is the envelope arithmetic, the lot dimensions and coverage against the setbacks, floor area ratio and height limits the code sets, and the assessor's record of what is currently there.

Thirty minutes is enough to establish the zoning district, the permitted and conditional uses, and whether the proposal needs a discretionary approval. It is not enough to conclude that an approval will be granted. The finding a screen can support is a status and a route: by right, conditional, or requiring a variance, with the code section cited. That distinction changes the timeline, and a timeline change on a construction loan is a budget change. The mechanics of reading these records are covered in zoning data by parcel and the envelope arithmetic in buildable area analysis.

Access and demand context complete the picture at screening depth. Traffic counts establish whether the location is exposed to the volumes the borrower's plan assumes, and trade-area demographics establish whether the population the plan needs is present. Neither is a market study, and a screen should not pretend otherwise. Both are pulled here only to detect a contradiction between the record and the story, and the apportionment traps that make ring demographics unreliable at this depth are set out in the ACS trade-area piece.

The All Appropriate Inquiries clock, and what a Phase I actually buys

The environmental layer has a legal architecture that a screen should understand before it recommends spending money on a Phase I. All Appropriate Inquiries is the standard of pre-acquisition environmental investigation codified at 40 CFR Part 312. EPA published the final AAI rule on 1 November 2005 and it took effect on 1 November 2006. The current procedure reference sits at 40 CFR 312.11(a), which points to ASTM E1527-21 for the Phase I Environmental Site Assessment process, and to ASTM E2247-23 for forestland or rural property. EPA's amendment recognizing E1527-21 was published at 87 FR 76578 on 15 December 2022 and took effect on 13 February 2023, with the older E1527-13 recognized only until 13 February 2024 and E2247-16 only until 24 June 2025. Neither older standard is current.

The timing rules are what a screen must know. 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 of adjoining properties; and the declaration by the environmental professional.

A Phase I is therefore a perishable document with two different shelf lives running at once, and a report commissioned early in a slow deal can age out of the 180-day window while the one-year window is still open. That is a scheduling fact a screen can flag on day one, and it is cheaper to flag than to discover at closing.

Now the part that is routinely misunderstood, and the reason this screen exists at all. Completing All Appropriate Inquiries does not clean a site and does not protect the lender's collateral. It establishes a defense. The FDIC's Risk Management Manual of Examination Policies notes at page 3.2-45 that bona fide prospective purchaser protection is limited to CERCLA and does not extend to RCRA, including underground storage tank liability, or to other federal or state statutes. The OCC's Comptroller's Handbook makes the parallel point at page 66: neither the CERCLA section 101(20) secured creditor exemption nor an AAI study protects the bank from the decline in collateral value that contamination causes. The exemption itself is narrow, covering a lender that holds a security interest without participating in management, and permitting foreclosure only where the bank divests at the earliest practicable, commercially reasonable time.

Put those together and the counterintuitive conclusion follows. The document the industry treats as the environmental answer is a liability shield for the acquiring party, bounded by one statute, and it says nothing about whether the building is still worth what the loan assumes. The tank that most often impairs a small commercial property sits under RCRA, outside the shield entirely. A lender who orders a Phase I and stops has bought a defense against a claim, not an assurance about the asset. The screen that runs before the term sheet is the one that protects the loan, because its question is the one the Phase I never asks: given what the public record already says about this parcel, is the collateral worth what we are about to advance against it?

That is also why the screen belongs in the file. The FDIC manual directs examiners to verify that institutions maintain an environmental risk program consistent with Appendix A to 12 CFR Part 364 and with Part 365, including an initial environmental risk analysis during the application process that considers present and past uses of the property. A dated screening page is evidence that the analysis happened at application, which is precisely when the manual expects it. The deeper treatment of that sequence is in environmental and hazard screens before the Phase I.

If the file is going to SBA

Where the credit is heading for an SBA 7(a) or 504 guaranty, the screen should record two additional facts, because both change the shape of the deal rather than merely its diligence.

The first is environmental. SBA's SOP 50 10 sets a tiered environmental investigation rather than a blanket Phase I requirement. The lender first makes a good faith determination of the property's current and known prior uses, expressed as NAICS codes, and compares them to the list of environmentally sensitive industries carried in the SOP's Appendix 6. A match escalates the requirement. The SOP defines an Environmental Investigation as an investigation conducted by an independent Environmental Professional that begins with a Phase I Site Assessment in accordance with ASTM E1527-21, and its Appendix 4 carries the environmental definitions. A screen that has already identified the prior use of the parcel has done the first step of that determination before the file opens.

The second is structural. For 504, the SOP tabulates the typical structures directly: in the standard case the Third Party Lender provides 50%, the CDC debenture 40% and the borrower 10%. Where the business is new or the property is a limited or special purpose property, the borrower contributes 15% and the debenture drops to 35%. Where the project is both a new business and a limited or special purpose property, the borrower contributes 20% and the debenture falls to 30%. A Special Purpose Property is defined as a limited-market property with a unique physical design, special construction materials, or a layout that restricts its utility to the specific use for which it was built.

That definition is a screening question, not a closing question. Whether the collateral is special purpose determines the borrower's injection, and a borrower told at term sheet that they need 10% who learns at commitment that they need 20% has been failed by the screen, not by the SOP. Two thresholds are worth carrying on the same page, and the appraisal one is a trap: 13 CFR 120.160(b) sits behind every version of it, but the SOP writes a different rule in each product chapter, so a rule quoted without its chapter is not a citation. For Standard 7(a) loans, those above $350,000, Chapter 1 is categorical: for all Standard 7(a) loans secured by commercial real property the lender must obtain an appraisal by a State licensed or certified appraiser, USPAP compliant and dated within 12 months of the application for guaranty, with no dollar threshold at all. For 7(a) Small and SBA Express, Chapter 2 sets no dollar test either: an appraisal is required where the parties have a close relationship or where SBA or the lender concludes one is necessary, and otherwise an appropriate evaluation consistent with safe and sound banking practices. The familiar $500,000 loan-amount test belongs to CAPLines, in Chapter 3, where loans above $500,000 require the appraisal and loans at or below it fall to the same close-relationship carve-in and evaluation default. Export Trade Finance, Chapter 4, is categorical for all loans secured by commercial real estate. The 504 programme is the only one that keys on property rather than loan: an appraisal is required where the estimated value of the project property is greater than $500,000. One rule is common to all of them: where the property's estimated value is over $1,000,000 the appraiser must be State-certified rather than merely State-licensed. The second threshold is coverage: the SOP's credit standards look for projections supporting debt service coverage of at least 1.15 within two years of funding, or within two years of the end of construction for construction projects.

Writing the screen into the credit file

A screen that is not written down did not happen, and a screen written without dates cannot be relied on six months later. The output is one page with six findings. Each finding names the layer, states what was found, names the source and the date it was read, and classifies itself as clear, conditional or adverse. The final line is a recommendation with a named condition where there is one.

Dates are the load-bearing element, because every layer in this screen has a vintage and the vintages do not align. FEMA's flood record is refreshed on its own schedule, EPA's facility registry on another, the assessor's roll annually, and the municipal code whenever the legislative body acts. When the public flood datasets were queried on 23 August 2026, the National Flood Insurance Program policy file held 74,349,525 records last refreshed on 27 July 2026, the claims file held 2,724,656 records refreshed on 4 August 2026, the multiple-loss property file held 240,651 records, the community status book held 25,125 records refreshed on 20 August 2026, and the disaster declarations summary held 70,248 records last refreshed on 25 September 2025. Five layers, five different refresh dates, one of them nearly a year older than the others. A screen that reports all five as "current federal data" has said something untrue in a way that no reader can detect.

The habit that survives audit is simple: carry the source and the retrieval date on every displayed value. That is the standard MMCG Analytics is built on, a map-first commercial real estate analytics platform for lenders and investors built on federal, state and public data with source and vintage provenance carried on displayed values. It is also the standard that lets a credit file defend a number two years after the analyst who pulled it has moved on.

What the screen cannot do

The limits are as important as the findings, and stating them is what separates a screen from a claim.

A screen cannot clear a site. It reports what public records say and nothing about what they omit, and the absence of a listing is not evidence of absence. It cannot substitute for All Appropriate Inquiries, which has a defined procedure, a defined professional and a defined timing rule, none of which a thirty-minute review satisfies. It cannot value the collateral, and nothing in it is an appraisal or a market study. It cannot resolve zoning, only classify the route. It cannot establish jurisdictional wetlands, because the National Wetlands Inventory states in its own use constraints that there is no attempt to define the limits of proprietary jurisdiction of any government or the geographic scope of regulatory programs. And it cannot be inherited: a screen run for one borrower on one intended use does not transfer to a different use on the same parcel, because the questions change with the use.

What it can do is decide, cheaply and on the record, whether the next dollar should be spent. On most parcels the answer is yes and the page is short. On the parcels where the answer is no, the thirty minutes has saved an appraisal, a Phase I, a legal review and a conversation with a borrower who was told the deal was likely. That is the whole return, and it is a large one. The full public-records stack that follows this screen, once the deal survives it, is set out in the due diligence data stack.

Frequently asked questions

What is a site screen in commercial lending?

A short, fixed review of public records for a specific parcel, run before a term sheet is issued. It covers collateral limits, flood status, environmental records, hazard history, zoning and access, and it produces a dated page with a finding and a source for each layer. It is not an approval, a valuation or a clearance, and it does not replace any formal report.

How long should a pre-term-sheet site screen take?

About thirty minutes for one parcel by one analyst. The budget holds because the layers and their order are fixed, and because the screen records what each source says rather than investigating why. The order runs from the cheapest disqualifying finding to the most nuanced: collateral arithmetic, flood, environmental, hazard history, zoning and access.

Does a Phase I environmental site assessment protect the lender?

Not in the way it is usually assumed to. Completing All Appropriate Inquiries establishes a liability defense, and its reach is bounded. The FDIC's examination manual notes at page 3.2-45 that bona fide prospective purchaser protection is limited to CERCLA and does not extend to RCRA, including underground storage tank liability, or to other federal or state statutes. The OCC's handbook notes at page 66 that neither the secured creditor exemption nor an All Appropriate Inquiries study protects a bank from the decline in collateral value that contamination causes.

How long is a Phase I environmental site assessment valid?

Two windows run at once. Under 40 CFR 312.20(a) all appropriate inquiries must be conducted within one year prior to the date of acquisition. Under 40 CFR 312.20(b) five components must be conducted or updated within 180 days of and prior to acquisition: interviews, environmental cleanup lien searches, government records reviews, visual inspections, and the environmental professional's declaration. The current referenced standard at 40 CFR 312.11(a) is ASTM E1527-21.

What is the difference between a site screen, a market study and an appraisal?

Scope, cost and timing. A site screen is a free public-records review run before commitment to find disqualifying facts. A market study addresses demand, supply and absorption for a specific use. An appraisal produces an opinion of value under a professional standard. The screen precedes both and is designed to prevent spending on either where the public record already shows the deal will not work.

Does a special purpose property change an SBA 504 structure?

Yes, and it changes the borrower's cash. SBA's standard operating procedure tabulates the typical structures: standard financing places 50% with the Third Party Lender, 40% in the CDC debenture and 10% with the borrower. Where the business is new or the property is a limited or special purpose property, the borrower contributes 15% and the debenture falls to 35%. Where the project is both, the borrower contributes 20% and the debenture falls to 30%.

Sources

1. Interagency Guidelines for Real Estate Lending Policies, Appendix A to Subpart D of 12 CFR Part 34 (OCC), with identical text at 12 CFR Part 365 (FDIC), current Code of Federal Regulations text, 2026. https://www.law.cornell.edu/cfr/text/12/appendix-A_to_subpart_D_of_part_34
2. Office of the Comptroller of the Currency, Comptroller's Handbook, Commercial Real Estate Lending, Version 2.0, 2022 (glossary definitions of debt-service coverage ratio and debt yield; stress-testing guidance; secured creditor exemption at pages 65 to 67). https://www.occ.gov/publications-and-resources/publications/comptrollers-handbook/files/commercial-real-estate-lending/pub-ch-commercial-real-estate.pdf
3. Federal Deposit Insurance Corporation, Risk Management Manual of Examination Policies, Section 3.2 Loans, Environmental Risk Program, 2023 (pages 3.2-44 to 3.2-46). https://www.fdic.gov/regulations/safety/manual/section3-2.pdf
4. Federal Emergency Management Agency, Flood Data Viewers and Geospatial Data (National Flood Hazard Layer), page last updated 3 April 2025. https://www.fema.gov/flood-maps/national-flood-hazard-layer
5. Federal Emergency Management Agency, OpenFEMA API dataset metadata (v1 DataSets endpoint), record counts and last-refresh dates retrieved 23 August 2026. https://www.fema.gov/api/open/v1/DataSets
6. Code of Federal Regulations, 44 CFR 59.1, definitions of area of special flood hazard, coastal high hazard area and zones of shallow flooding, current text 2026. https://www.law.cornell.edu/cfr/text/44/59.1
7. Code of Federal Regulations, 44 CFR 64.3, Flood Hazard Boundary Maps and Flood Insurance Rate Maps, current text 2026. https://www.law.cornell.edu/cfr/text/44/64.3
8. Code of Federal Regulations, 44 CFR Part 70, Procedure for Map Correction, current text 2026. https://www.law.cornell.edu/cfr/text/44/part-70
9. United States Code, 42 U.S.C. 4012a, flood insurance purchase and compliance requirements, current text 2026. https://www.law.cornell.edu/uscode/text/42/4012a
10. United States Code, 42 U.S.C. 4104b, Standard Flood Hazard Determination Form, current text 2026. https://www.law.cornell.edu/uscode/text/42/4104b
11. United States Code, 42 U.S.C. 4101, flood hazard identification and map revision, current text 2026. https://www.law.cornell.edu/uscode/text/42/4101
12. Code of Federal Regulations, 12 CFR Part 22, Loans in Areas Having Special Flood Hazards (OCC), current text 2026. https://www.law.cornell.edu/cfr/text/12/part-22
13. U.S. Environmental Protection Agency, Envirofacts, About the Data, 2026. https://www.epa.gov/enviro/about-data
14. U.S. Environmental Protection Agency, Envirofacts data service, Facility Registry Service program facility counts by program system acronym, retrieved 23 August 2026. https://data.epa.gov/efservice/frs_program_facility/
15. Code of Federal Regulations, 40 CFR 312.20, timeliness of all appropriate inquiries, current text 2026. https://www.law.cornell.edu/cfr/text/40/312.20
16. Code of Federal Regulations, 40 CFR 312.11, references to industry standards, current text 2026. https://www.law.cornell.edu/cfr/text/40/312.11
17. U.S. Environmental Protection Agency, Standards and Practices for All Appropriate Inquiries, final rule, 87 FR 76578, 15 December 2022. https://www.govinfo.gov/content/pkg/FR-2022-12-15/pdf/2022-27044.pdf
18. U.S. Environmental Protection Agency, Brownfields All Appropriate Inquiries program page, 2026. https://www.epa.gov/brownfields/brownfields-all-appropriate-inquiries
19. NOAA National Centers for Environmental Information, Storm Events Database documentation, 2026. https://www.ncei.noaa.gov/stormevents/details.jsp
20. NOAA National Centers for Environmental Information, Storm Data bulk data format documentation, citing National Weather Service Directive 10-1605, 2020. https://www.ncei.noaa.gov/pub/data/swdi/stormevents/csvfiles/Storm-Data-Bulk-csv-Format.pdf
21. U.S. Small Business Administration, SOP 50 10 8, Lender and Development Company Loan Programs, effective 1 June 2025 (typical 504 structures; Special Purpose Property definition; appraisal requirement; credit standards); document text retrieved and read 23 August 2026. https://legacy.sba.gov/sites/default/files/2025-05/SOP%2050%2010%208%20Technical%20Updates%20effective%206.1.2025.docx
22. U.S. Small Business Administration, SOP 50 10 8.1, Lender and Development Company Loan Programs, effective 1 October 2026 (environmental policies and procedures; Appendix 4 definitions; Appendix 6 NAICS codes of environmentally sensitive industries). https://legacy.sba.gov/sites/default/files/2026-08/SOP%2050%2010%208.1%20effective%2010.1.2026_0.docx
23. Code of Federal Regulations, 13 CFR 120.160(b), appraisals and evaluations, current text 2026. https://www.law.cornell.edu/cfr/text/13/120.160
24. California State Water Resources Control Board, Underground Storage Tank Program (GeoTracker, UST Cleanup Fund, Low-Threat UST Case Closure Policy), 2026. https://www.waterboards.ca.gov/water_issues/programs/ust/
25. City and County of San Francisco, Frequently asked questions about the Maher Ordinance (San Francisco Health Code Article 22A), 2026. https://www.sf.gov/information--frequently-asked-questions-about-maher-ordinance
26. U.S. Fish and Wildlife Service, Federal Geographic Data Committee metadata for the USFWS National Wetlands Inventory dataset, Use Constraints, metadata publication date 1 May 2024. https://documentst.ecosphere.fws.gov/wetlands/data/metadata/FWS_Wetlands.xml
27. U.S. Environmental Protection Agency, observed HTTP status of the EJScreen pages, recorded 22 August 2026. https://www.epa.gov/ejscreen