How to map data center construction from permits and aerial changes

What belongs on a data center construction map?

Build the map by joining permits, parcels, and dated aerial observations at the building-phase level. Give each campus a parent record, each building or expansion a child record, and each shared utility yard its own linked feature. That structure lets you show an operating hall beside an active foundation without assigning one misleading status to the entire property.

Start with the commercial question. An earthwork contractor needs a different watchlist from a generator service manager or a commissioning firm. Write the service area, the work you perform, and the stage at which you can still help. Then set the inclusion rule: a named data center use in an official record, an operator confirmation, or an unresolved candidate that stays outside the confirmed count.

A broad roof and equipment yard justify research. Confirm data center use from an official record or the operator before including the site in the confirmed count. Keep unresolved industrial sites in a separate map layer.

Where do you find permits and project records?

Start with the jurisdiction that controls the parcel. Search its planning cases, approved site plans, building permits, inspection history, and public meeting attachments. Use the parcel number, street address, applicant name, and known project aliases. Search each separately. A campus marketing name may never appear on the permit application.

Loudoun County's LandMARC page provides access to public permit and land development records. It is a useful example of the record system to locate in your own market. Save the case identifier and the specific document you used, because a portal's search results can change when a case advances or an address is corrected.

Loudoun's data center overview separates existing buildings, including those under construction, from approved but unbuilt space. It also warns that not every approved project will be built. Keep the page's reporting dates with its figures; today's download does not make an older count current.

The table is a research sequence, not a universal approval sequence. Check the actual document status and scope. An application, an approval, an issued permit, a revision, and a closed record are different events. Where access is restricted or an attachment is missing, mark the missing evidence and request the record through the agency's published process.

How do you match permits to the right parcel?

Use the jurisdiction's parcel identifier as the first join key, with the jurisdiction name attached. Preserve leading zeros and punctuation in the source value. Make a separate normalized value if your spreadsheet needs one. Parcel identifiers from different counties can look alike, and a street address alone can put a pin at a mailing office or the wrong campus entrance.

Compare the permit's parcel list with the site-plan boundary and the mapped footprint. Record every parcel the phase occupies. If a parcel was split or consolidated, keep the earlier identifier as an alias and record the effective source date. Do not delete the older relationship just because the current map displays a new number.

Store a campus polygon, building footprints where supported, and a representative point for the list view. Mark approximate geometry as approximate. Keep the map's coordinate reference information with any export, and spot-check a known road intersection after import. A misplaced coordinate can silently join a real permit to the adjacent warehouse.

Separate owner of record, applicant, developer, operator, and general contractor. Give each role its own source and date. The property owner is a starting point for research; it does not identify the equipment buyer or the person authorized to invite your bid.

What can aerial changes actually prove?

Aerial comparisons can support observations about visible change between usable views. Compare the same boundary, similar scale, and recognizable fixed features. Record clearing, grading, new access, foundations, roof area, external equipment pads, and visible installations separately. Write the observation before the interpretation: rectangular pads appeared east of Building A; equipment type unresolved.

Loudoun County's aerial archive supports searching by address or parcel number and comparing years with a swipe tool. That makes it useful for establishing a baseline. It does not make an older image a present-day construction inspection. Use the capture date or stated capture range for the exact area, not the date you downloaded it or the year printed in a copyright footer.

Visibility has limits. The Centers for Disease Control and Prevention's cooling tower identification guidance combines imagery, ground scouting, and records, and notes that side-mounted fans can be difficult to see from above. Apply that discipline to equipment yards: record an uncertain object as uncertain, then specify the ground observation or equipment document needed to settle it.

The Department of Energy's design guide treats information technology equipment, cooling, and electrical systems as related parts of a facility. Use that relationship to decide what to investigate together. Do not turn it into a capacity estimate from a roof. Equipment capacity, manufacturer, tank contents, internal cooling architecture, commissioning, and available utility power need their own evidence.

Before accepting a change, inspect shadows, tree cover, image alignment, and obstruction. A darker roof may reflect lighting. A missing object may be hidden by a viewing angle. If the views are not comparable, record no reliable comparison rather than removed equipment. Save the evidence reference and an annotated crop only where your access rights allow it.

Which project status should you assign?

Assign the narrowest status the evidence supports. Use separate fields for administrative status, observed physical stage, power readiness, and procurement. A permit can be issued while the latest usable view still shows bare land. That is two compatible facts, not a reason to force the pin into a single advanced category.

The legend below is a recommended research convention, not an agency classification. Preserve the agency's original wording alongside it. Add unresolved and stale flags independently of the stage. A roof observed last year remains a recorded roof observation; it does not automatically become current fit-out or operating capacity because time passed.

Use plain confidence reasons: corroborated by permit and dated view; single dated source; conflicting evidence; location unresolved. These describe your evidence. They are not statistical probabilities. Reserve operational for explicit confirmation covering the relevant building or hall, and keep construction activity elsewhere on the campus visible.

How do you handle conflicting dates and stale evidence?

Keep four dates: the event or document date, the image capture date, your access date, and the date you last reviewed the record. They answer different questions. A permit issued in June, photographed work in September, and an October download do not establish an October construction start.

When a clear earlier image lacks a feature and a later comparable image shows it, report an observation interval. Say foundation first observed between May 12 and September 18. Do not assign the midpoint as the start date. If the earlier view is obscured, even that interval is unsupported; you only know the foundation was visible by the later capture.

If a fresh permit entry conflicts with an older view, preserve both and identify what would resolve the gap. If an agency says cancelled while a recent image shows earthwork, check whether the cancellation affects a building permit, an application revision, or the whole development. Similar project names and partial cancellations deserve a manual review.

Set an internal rule for rechecking stale records. For example, require a fresh status check before assigning takeoff hours. Label that as office policy, not an expiration date for public evidence.

Worked example: two phases on one campus

A typical campus record keeps each phase separate. Example Campus X occupies two adjoining parcels, EX-101 and EX-102, and an approved site plan identifies Buildings A and B with a shared equipment yard. The working map is reviewed for November 4, 2026.

A June 3 shell permit, EX-A, covers Building A only. A separate issued land-disturbance record, EX-LD, covers both parcels. No Building B construction permit has been located in the searched records. That last statement describes the search result; it does not prove that no permit exists.

Comparable, unobstructed images captured May 12 and September 18 show undisturbed ground in the first view, then a foundation at A, grading at B, and rectangular pads in the shared yard. A second reviewer matches the foundation to A's site-plan footprint. Neither reviewer has an equipment schedule, utility energization confirmation, or procurement contact.

Publish one campus with two building phases and one linked yard. Building A has a permitted shell and an observed foundation. Building B has observed site work with its building-permit status unresolved. The yard has observed pads of unconfirmed purpose. Count neither building as operational, and count no megawatts or equipment units from those pads. The next useful task is to resolve B's permit scope and the yard schedule.

What does a useful mapping pilot cost?

Price the research effort before buying a large territory list. Anvilfield research models a typical contractor's desktop pilot on the following cost basis. 24 raw pins collapse to 20 unique campuses after four duplicate pins are removed. Deduplication is included in setup time. Eight campuses need additional follow-up, and five ultimately meet your own territory, stage, and scope rules for a qualification conversation.

The internal loaded labor cost is $60 per hour, including payroll burden and office overhead but no profit. Existing mapping software has no incremental license cost. Budget a flat $200 for permitted evidence exports for this pilot. All work is remote. Travel, site visits, new survey or flight work, taxes, and proposal preparation are excluded and would require separate allowances.

The labor is 3 + 15 + 6 + 4 = 28 hours. Labor cost is 28 x $60 = $1,680. Add the $200 export allowance for $1,880 total. With five retained campuses, the research cost is $1,880 / 5 = $376 per retained campus. If only two survive qualification, the same spend is $940 each. Neither result is a cost per sale.

No award probability, revenue, contract value, or margin is modeled. Use each retained record to start a scope conversation and confirm sales separately. Review the pilot's rejection reasons before expanding: wrong facility type, duplicate campus, old evidence, work already placed, or a scope outside your crew's reach. Those reasons tell you which research filters to tighten.

The source for asset inventories related to data center construction is PorchRocket's Asset Search, which finds visible outdoor power assets and monitors what is new, removed or altered.

What turns a mapped project into a useful work lead?

A mapped project becomes useful when you can name the work, the responsible party, the evidence behind its timing, and the question your team can resolve. A new foundation might justify checking civil or electrical package status. It does not establish that a package is open. Equipment in a yard might justify an installation or service inquiry, but it may already be purchased, under warranty, or assigned to an incumbent.

Before assigning an estimator, confirm the building phase, the actual contracting party, the procurement route, and whether your scope is still available. Ask for the current package boundary and schedule. For service work, confirm the operator, installed equipment record, access arrangements, warranty responsibilities, and the stated service need. Record the answer and its date separately from the aerial observation.

Use the related site-selection, power, buildout, generator-permitting and cooling guides to prepare technical questions. Use the commissioning guide to check acceptance scope and the daily-report guide to record an authorized visit. Verify each answer against the actual project.

Take an unresolved record to a site visit only with the appropriate access arranged. Specify the observation needed, such as matching a building designation or reading an authorized equipment schedule. Return the result to the same phase record so the next reviewer can see what changed and why.

What to document

The minimum handoff contains campus identifier, phase identifier, jurisdiction, parcel identifiers, mapped geometry, project aliases, and the role of each named organization. Keep document identifiers, revision and status, direct source references, image capture dates, observation notes, access dates, and reviewer names with that record. Preserve the distinction between a downloaded document and a link you have not opened.

Use two separate text fields: observed fact and interpretation. A weak note says new data center almost done. A stronger note says Building A foundation visible in the September 18 capture; shell permit EX-A issued June 3 in this typical example; current completion and package availability unconfirmed. The second note tells an estimator exactly what remains unknown.

Log the earlier status, new status, supporting evidence and reviewer. Assign the next action and review date so unresolved questions reach someone who can answer them.

Common mistakes

• Counting a campus announcement, three permits, and two building footprints as six separate projects.
• Joining an applicant's mailing address to the map instead of the permitted parcel.
• Calling an application approved, or treating an approval for grading as permission for every building phase.
• Using the download date as the image capture date and inventing a construction start date.
• Multiplying visible pads by a modeled generator rating, or treating backup ratings as available IT capacity.
• Marking a campus operational because the roof is complete, the parking lot is full, or equipment has arrived.
• Calling a project cancelled because two old images look alike, despite having no attributable cancellation record.
• Sending a sales handoff with no scope question, no contracting role, and no evidence date.

Field checklist

How often should you refresh the map?

Refresh when the decision requires it and when the source can supply new evidence. As a starting internal policy, review active pursuit records weekly and broader territory records monthly. Schedule reviews around permit updates and aerial-image availability. Confirm the actual source cadence and shorten the interval around a known procurement milestone.

Separate no new evidence from no observed change. If you checked the portal today but the latest usable image is unchanged from your last review, log the portal check and retain the old image date. Do not mark the physical stage freshly verified. An alert should identify the changed document or the two images that support it.

For custom monitoring, write the campus boundaries, visible features, phase rules, desired cadence, and unresolved cases into the request. PorchRocket Asset Search offers custom searches and monitoring of assets visible from above. Ask which capture dates and review interval your territory supports, then use your permit ledger to test the returned observations. Keep the commercial qualification step with your own team.

Standards and references

This workflow is a research method, not a permit determination. Use the current jurisdiction record for local approvals and the actual utility correspondence for power commitments. The primary references linked in this guide support specific parts of the research method.

For stormwater, the Environmental Protection Agency explains the construction-discharge permit threshold as disturbance of one acre or more, including smaller sites within a common plan that will ultimately disturb at least one acre. Confirm the applicable permitting authority and site-specific coverage. Aerially estimated disturbed acreage is not a compliance determination.

At the November 4, 2026 review, EPA lists the 2027 Construction General Permit as proposed and gives February 16, 2027 as the 2022 permit's expiry. These are EPA permits for areas where EPA is the permitting authority. Check the applicable state or EPA permit for the site. Keep source dates separate from this article's review date.

Units, terms, and definitions

Keep area units explicit: parcel acres, building footprint square feet, and reported floor area measure different things. Keep megawatts attached to the source's stated basis, such as requested utility load or IT capacity. Leave unknown values blank with a reason; zero is a measured or documented value, not a substitute for missing information.

Campus record

The parent mapping record linking parcels, buildings, phases, and shared infrastructure without counting each as a separate campus.

Observation interval

The period between comparable dated views within which a newly visible feature appeared; it does not identify an exact construction date.

Capture date

The date an image was acquired, kept separate from its publication date and the date a researcher accessed it.

Evidence confidence

A stated reason for trusting a specific parcel match or status assignment, based on source agreement, recency, and visibility.

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. Campus identifiers, documents, dates, geometry and observations are scenario inputs.

Related tools

FAQ

People also ask

• What matters most in data center site selection?
• How is a data center built?
• How does a data center get power?
• What is a Tier 4 generator?
• How are data centers cooled?