Build a tank-facility directory from imagery and public records

What does a tank-facility directory actually contain?

For each mapped site, record the facility identity, visible tank candidates, source records, and unresolved questions. Explain why the observation belongs to that facility so another researcher can check the match.

Use one stable internal ID for each facility and separate IDs for its tank observations. Link parcel and account records underneath or alongside that structure. A site with six visible vessels is one candidate facility with six observations. A company with three separate terminals has three facility records linked to one account. Neither arrangement justifies six or three independent purchasing opportunities.

Use the terms tank farm locations and aboveground storage tank map as research starting points. Define your own inclusion rule before counting. In this workflow, tank farm is a descriptive grouping of visible tank candidates, not a legal classification. Keep isolated tanks eligible when they fit your service scope. A directory for coating contractors may need a different asset mix from one for fuel-system technicians.

Territory and counting rules

Write a one-page scope before searching: geography, intended work, included tank classes, excluded lookalikes, acceptable evidence age, and the unit being counted. Use county boundaries, a drawn service area, or named industrial districts. Save the boundary version so two researchers cover the same ground and overlapping searches can be reconciled.

Choose whether the directory includes only visible aboveground storage tank candidates, abbreviated AST candidates here, or also document-only records for underground storage tanks, abbreviated USTs. Keep those categories separate. For a broad industrial inventory, give water storage, process vessels, petroleum storage, and unresolved vessels different research labels. Do not force an uncertain object into the service category you hope to sell.

Set acceptance before collection. For this method, an accepted desktop facility match needs a located observation, a matching facility record, a documented reason for the association, and no unresolved identity conflict. Unknown contents or maintenance needs can remain explicit. Acceptance means the research relationship meets your rules; it does not certify the equipment or confirm that the site is operating.

Where can you find public tank and facility records?

Start with the responsible state's storage tank program, then add environmental facility records and local property resources. Read each dataset's scope before downloading it. A search of registered petroleum tanks cannot answer every question about industrial vessels or water storage. Save the query, filters, extraction date, and the agency's description of what the file covers.

Florida's tank information page distinguishes all-facility reports from regulated-facility reports and provides facility and tank identifiers. Its broader reports include active equipment, history, and cleanup activity; closed-facility owner information may be outdated [1]. Pennsylvania separately offers active and inactive searches and explains that certain tank information is omitted from public displays [2]. Those are concrete reasons to record coverage limits before reporting a territory total.

EPA's Facility Registry Service, or FRS, links reported site information from several program systems under Registry IDs. EPA describes attempted site matching and periodic refreshes [3]. Use that cross-reference to investigate aliases and program records. A registry match alone does not identify every vessel at the location. Returning the same underlying record through two portals also does not create two independent confirmations.

What can imagery establish about a tank site?

Use imagery to record observable geometry and context: cylindrical forms, horizontal vessels, grouped footprints, apparent containment boundaries, and their relationship to buildings. Describe an uncertain shape as an uncertain shape. A round structure can warrant a closer records check without becoming a confirmed storage tank in the inventory.

Record the image capture date separately from the day you viewed it. If only a date range is available, retain that range. If no capture date is provided, write unknown and apply your evidence-age rule. Zoom level, a current map copyright notice, or the day you downloaded a view does not supply a capture date.

Mark obscured areas and ambiguous objects. Use at least six visible candidates when six are clear and another area is hidden; do not silently turn that observation into six total tanks. Keep the original view reference and your interpretation together so a reviewer can repeat the count. Describe size as estimated only when you record a measurement method and its limitations.

Leave contents, usable capacity, wall thickness, coating condition, contamination, compliance, and demand for service unresolved unless separate evidence supports them. A footprint does not supply a verified height or operating volume. Before using any dimensional estimate in a scope, reconcile it with equipment records or authorized field measurements. These are acceptance boundaries for this research method, not a remote inspection procedure.

Minimum directory record

Keep a facility sheet, a linked observation sheet, and a source log. A spreadsheet is enough for a pilot if the identifiers and field definitions remain consistent. Store identifiers as text so an import preserves leading zeros. Keep original names and addresses alongside normalized versions rather than overwriting the evidence used to make the match.

Give each field its own source and verification status. Use observed, document reported, operator confirmed, unknown, or conflicting. These labels describe the basis of a field; they are not numerical probabilities. One row can have a supported facility match, document-reported capacity, and unknown current contents. A single verified flag would hide those differences.

Distinguish not searched, searched without a result, unavailable, and conflicting. Blank cells erase that work. For unresolved fields, add a specific next action and responsible role. Keep credentials and private access instructions out of the shared research export; the person arranging an authorized visit can manage those separately.

For coordinates, retain the coordinate system and what the point represents. For reported quantities, retain the units. In the source log, save the URL or file reference, source date, effective date where available, and retrieval date. The compact table below names the field groups; those details belong in the accompanying field dictionary.

How do you match a visible tank to a facility record?

First, preserve the original observation and source rows. Standardize addresses in separate fields, including alternate street names and unit information. Compare agency identifiers before names. Within one program, a facility ID can organize the related tank records; use the agency, facility ID, and tank ID together when linking equipment. Do not merge identical tank numbers from different sites.

Second, inspect the location. Record whether coordinates represent the vessel, facility reference point, parcel center, or an address estimate. Compare the site boundary, adjacent parcels, access roads, and neighboring operators. Proximity creates a candidate match. A fixed distance rule alone cannot decide which of two neighboring terminals owns an observed tank group.

Third, corroborate the association with a different kind of evidence: a site plan, facility address history, parcel relationship, or operator-supplied inventory. Seattle's research guidance specifically discusses changed addresses, divided parcels, historical records, and construction plans that require a request [5]. Search aliases and related addresses before concluding that the permit file is empty. Record the archive and period searched.

Fourth, review the dates. Compare what each source says at its own point in time. An older image and a newer tank status are not necessarily contradictory. Preserve both, identify the possible change, and request evidence that connects the dates. A newer database extraction does not automatically mean every underlying field was recently verified.

Finally, assign a decision and a reason. Accept the facility relationship only when the evidence meets your written rule. Keep competing matches unresolved until the boundary or responsible operator is established. Reject a duplicate or wrong asset class with its original ID intact. Have another reviewer check ambiguous boundaries and the accepted matches before handing the directory to sales or estimating.

Worked match: one facility, conflicting tank counts

Site F-014 has an image captured May 14, 2025 showing three clear circular vessels inside one apparent yard. A February 2026 registry extract lists two active aboveground tanks and one closed tank under a facility address. The parcel record names Example Land Company while the facility file names Example Terminal Operations. The record reconciles names, dates, documents, and counts.

The first defensible entry is three visible candidates on the image date, two tanks reported active on the registry extract, and one reported closed. Do not subtract the closed tank from the visible count or decide which circle it is. Closed status does not by itself tell this researcher whether the vessel remained in place when the image was captured. Keep the tank-to-observation links unresolved.

Now the operator supplies a dated site plan linking tank tags T1, T2, and T3 to the three positions, confirms that T3 remains physically present but out of service, and identifies its company as operator. Preserve the plan date and confirmation date. Accept the facility match and document the tag associations. Record operating status as operator confirmed, while capacity, current contents, inspection condition, and purchasing authority remain unresolved unless supplied separately.

The handoff is one facility, three located observations, and three linked equipment records with different statuses. It is not three active service accounts. If the plan instead places one vessel across the neighboring boundary, reopen that association and retain the correction history.

Who owns the land, the tank, and the service decision?

Research those roles separately. King County's official property research page, for example, connects parcel searches with assessment reports and recorded-property resources [4]. Use the equivalent official resources in your territory to establish the property identity. Treat the resulting entity as a research contact path until someone confirms responsibility for the equipment.

Ask the facility's public business contact who operates the tank system, who maintains its records, and who approves outside service. Record the answer, role, business channel, and date. The property owner, operating company, maintenance manager, access coordinator, and purchaser can occupy different fields without creating a conflict.

Keep commercial qualification in a separate stage. A supported facility match can move to operator research. A confirmed operator can discuss a defined service question. A visit needs an agreed purpose, scope, access arrangement, and report recipient. Record declined contact and restrictions so another person does not restart the same inquiry from an older export.

What must field verification settle before you price work?

Arrange an authorized external walkdown with the person responsible for the site. Match permitted nameplate photographs, asset tags, and locations to the directory. Confirm which tanks belong to the operator and which are outside the proposed scope. Record inaccessible areas and identify the document or specialist needed to resolve them.

For an inspection inquiry, request the tank register, latest inspection report, applicable program, and next required task. EPA's bulk storage container fact sheet distinguishes routine exterior observations from integrity inspection and testing and describes the importance of methods, personnel qualifications, and the container's configuration [6]. Its oil-storage guidance supports the scope distinction here; it does not classify every tank in your directory.

For coating work, obtain the service, substrate, previous system, outage constraints, and requested assessment before estimating quantities. For connected fuel or process equipment, request the system drawings and maintenance boundaries. Use the related Anvilfield guides for those detailed scopes after the operator confirms which work is relevant.

Keep this directory-verification visit external and observational. Tank entry, sampling, isolation, testing, climbing, or operating equipment requires a separately agreed work scope and the appropriate site controls. A successful match does not authorize any of those tasks. Return confirmed facts and unresolved questions to the original IDs so estimating receives the corrected record.

Worked budget: research 60 candidate facilities

A typical pilot starts with 80 raw facility rows collected for an agreed territory. Setup removes 12 duplicate rows, then eight outside the boundary, leaving 60 distinct candidates. Those exclusion groups are sequential and do not overlap. Review ends with 40 accepted desktop facility matches, 12 unresolved candidates, and eight rejected for the wrong asset class.

The researcher spends four hours setting up the scope and cleaning the input, 12 minutes reviewing observations per candidate, 18 minutes matching public records per candidate, 15 additional minutes on each of the 12 unresolved cases, and three hours preparing the export and handoff. A second reviewer spends six minutes on every candidate, including accepted and rejected cases. All work is performed once within these allowances.

Anvilfield research uses an internal researcher cost of $60 per hour and reviewer cost of $80 per hour. Both are fully loaded allocation rates including wages, payroll burden, ordinary software, equipment, and office overhead. Add $120 for the pilot's data access and record-copy charges, including any applicable purchase taxes. No other fees, subscriptions, or charges are needed.

The scope is desk research and a reviewed export. It excludes operator calls, travel, field visits, inspections, engineering, outreach, proposals, and future monitoring. The needed records are accessible within the allotted time. Add a reserve equal to 10% of base cost for research overruns; that percentage is a management choice. This is an internal budget, with no sales price, margin target, or sales tax on a customer invoice.

What does the pilot cost per usable facility match?

The example uses 46 person-hours: 40 researcher hours and six reviewer hours. Its $3,300 budget divided by 40 accepted matches is $82.50 per accepted desktop facility match. Dividing by all 60 candidates gives $55 per candidate processed. Label the denominator every time; those figures answer different management questions.

Only 30 matches qualify while the same budget is spent. The cost becomes $110 per accepted match. That sensitivity changes only the accepted count and holds every cost constant. It establishes no expected yield, contract value, or payback. A documented facility may still have no relevant service need.

Approve the next batch from actual hours, acceptance reasons, and remaining uncertainty. If most time went into neighboring-site conflicts, improve the matching evidence before expanding geography. If operator research is the bottleneck, fund that separate scope. Complete those checks before expanding the list.

How do you keep the directory current?

Set a refresh schedule based on the decision the directory supports and the available evidence. A proposed quarterly desk review is a workflow choice, not a promise that new imagery or agency updates arrive quarterly. Record last checked separately from last changed. Reopening an unchanged record must not make its underlying observation appear newer.

Compare compatible views and retain both dates. If a tank first appears in a later image, describe the observed change interval. Do not assign an installation date from that interval. If a vessel disappears, check visibility, image coverage, and records before calling it removed. Keep the earlier observation and the reason for any status change.

Review facility names, status, owner roles, and open tasks as well as imagery. EPA notes that source updates may take time to reach FRS [3]. A copied downstream record can therefore lag the originating system. Resolve material disagreements at the source and preserve who changed the directory, when, and why.

What to document and hand off

Deliver the facility table, linked observations, source log, field dictionary, coverage statement, and exception register together. State the territory version and acceptance rule. Reconcile raw rows, duplicates, exclusions, accepted matches, unresolved cases, and rejections. Report visible observations separately from document-reported active tanks and operator-confirmed equipment.

A weak note reads: Old tank farm, three tanks, call owner for painting. A useful sample note reads: F-014 has three visible candidates in the dated view; a supplied plan links the tags; the operator reports one out of service; coating condition and purchasing authority remain unknown; request the maintenance contact and latest assessment. Attach the site plan and assign the maintenance-contact check.

Assign every exception a responsible role and next action. Test the export by tracing one accepted match back to its evidence and one rejected row back to its reason. If either chain is missing, finish the handoff before treating the file as a reusable directory.

Common mistakes

The expensive matching error is accepting the nearest facility point without checking its meaning or the neighboring site. Preserve competing candidates and obtain boundary evidence. The expensive counting error is mixing tanks, parcels, facilities, and accounts in one total. Give each entity its own identifier and count.

The common evidence error is treating an old source copied into a fresh export as current confirmation. Keep source, effective, capture, and retrieval dates distinct where available. The commercial error is turning an unresolved maintenance question into a claim that the owner needs work. Move from evidence to a useful question, then to an agreed scope.

Directory release checklist

Standards and references

Sources checked November 4, 2026. State and local examples describe their own records systems; they do not establish nationwide coverage. EPA's inspection fact sheet is dated August 2013 and is used here for the distinction between inventory research and inspection scope. Confirm current requirements with the responsible program before defining regulated work.

The matching labels, acceptance rules, pilot outcomes, and costs are original planning examples. The related guides provide the next technical reading once tank service and responsibility are established.

PorchRocket's Asset Search offers custom searches and monitoring for assets visible from above [7]. Prepare the territory, visible inclusion rules, required dates, and unresolved-field policy before requesting a tank search. Carry the resulting candidates through the same facility and operator checks used for your own research.

Units and terms

Facility match

A documented association between a physical site and a facility record under the directory's stated acceptance rules.

Observation date

The date represented by the image or observation, kept separately from the date a researcher retrieved or reviewed it.

Source lineage

The originating record and the systems that copied or transformed it, used to distinguish independent evidence from repeated data.

Reported capacity

A capacity value attributed to a specific record, retaining its units and evidence date; it does not establish current contents or usable inventory.

Methodology

Counts, prices, work rates, schedules and outcomes in typical cases are inputs to the Anvilfield research model. Calculations use the stated scope, formulas, cost inclusions and exclusions. Cited public records and technical documents support their attributed facts. Named case records, dates and outcomes are scenario inputs; cited public records retain their source attribution. Local tank counts, facility identities, market size and conversion rates were not independently measured.

Related tools

FAQ

People also ask

• What is the difference between a UST and an AST?
• What is a tank lining?
• How much fuel does a generator need?
• How does a wastewater treatment plant work?
• What is industrial process piping?