Market Intelligence
Satellite vs aerial imagery for asset finding
Choose imagery by asset size, usable detail and capture date. Compare Anvilfield estimates, practical review gates and the work needed to turn pictures into building records.
Direct answer
Choose imagery that resolves the smallest asset and meets the required capture date. Aerial imagery is useful for detailed roof and equipment inventories; satellite imagery is useful for broad coverage and repeat observations, including detailed asset work when its sampling is fine enough. Check native pixel size, visible coverage, viewing angle and usable dates before buying either. The platform name alone does not determine whether you can count rooftop units or spot a new tank.
On this page
Which image should you choose for asset finding?What is the practical difference between satellite and aerial imagery?Anvilfield estimates of asset detail and observation timingHow many pixels does an asset need?Does a sharper color image prove more detail?How fresh must imagery be for finding and monitoring assets?Field benchmarks for comparing imageryBest practices for choosing imagery around the commercial taskWorked example: a regional HVAC distributor chooses coverageImagery acceptance checklistHow do leaders turn imagery into useful building records?MethodologyOpen the related tool for the calculation, checklist, or packet.
Key takeaways
- Compare native sampling and capture dates before choosing the platform.
- Keep geometric sampling, classification confidence and condition assessment separate.
- Use dated evidence and explicit unresolved statuses when building a prospect file.
- Require comparable before and after views for a change finding.
Which image should you choose for asset finding?
Start with the smallest feature that changes your commercial decision. Locating a large paved yard takes a different view from separating adjacent rooftop cabinets. Identifying a roof boundary takes a different view from reading the arrangement of fans and pipes. Write the asset, counting unit and required observation date into the brief before choosing a platform.
Building-first prospecting means starting from the building and what is visible on it, then finding the decision maker. Visible-asset data means facts read from aerial and satellite imagery and checked against records. Both methods depend on an image that supports the particular observation entered in the prospect file. A crisp-looking map is only the beginning of that evidence.
Ask the provider to show representative difficult properties from the actual territory: shaded roofs, dense equipment groups, buildings near image seams and sites with older coverage. Compare the source sampling, capture date and visible area for each property. A demonstration over an open industrial yard says little about a rooftop hidden by taller neighbors.
Choose the deliverable at the same time. A building list needs reliable building identity and a reason for inclusion. An equipment inventory needs distinct asset identifiers and counting rules. A monitoring report needs comparable dated views and a clear description of the change. Buying the same picture for all of those tasks creates avoidable review work.
Building-first prospectingVisible-asset dataHow aerial imagery finds rooftop units, tanks and lotsWhat is the practical difference between satellite and aerial imagery?
Satellite describes the collection platform in orbit. Aerial imagery comes from an aircraft, including a crewed aircraft or drone. The product delivered to the analyst matters just as much: a vertical orthophoto, an oblique view, a native spectral band and a sharpened color composite answer different questions. Record the product type alongside the platform.
Ground sample distance, or GSD, is the nominal ground spacing represented by image samples. Smaller spacing puts more samples across the same object. An orthophoto is corrected for mapping use; that correction does not reveal equipment hidden behind a parapet. An oblique view can show a side-mounted feature while changing the apparent arrangement of nearby objects. The CDC specifically identifies oblique imagery as useful for side-mounted cooling-tower fans. [6]
Use broad scenes to establish regional context and sufficiently detailed views to investigate individual assets. An aerial archive can supply a consistent historical baseline. A new satellite acquisition can supply a later observation where suitable collection is available. An authorized drone survey can answer a remaining site question after the territory work has identified a worthwhile visit.
[1] USGS system characterization report, satellite sampling (2021, version 1.1 (October))[2] USDA NAIP technical presentation (2018)[3] New York City aerial imagery metadata (2024 capture)[4] Copernicus SentiWiki mission specifications (Live technical reference accessed 2026-11-05)[5] NASA Landsat 8 specifications (Mission launched 2013; reference accessed 2026-11-05)[6] CDC Strategies for Identifying Cooling Towers (2026)Drone roof and facade inspectionAnvilfield estimates of asset detail and observation timing
Anvilfield research estimates the asset detail and observation timing below. Keep the units and the range with each quoted result; Methodology defines the size, policy and timing cases.
Anvilfield research estimates 0.1 pixels across a 3 m asset in 30 m satellite imagery, with 0.05 to 0.2 pixels across the selected asset-size range.
Anvilfield research estimates 0.3 pixels across a 3 m asset in 10 m satellite imagery, with 0.15 to 0.6 pixels across the selected asset-size range.
Anvilfield research estimates 2.4 pixels across a 3 m asset in native 1.24 m satellite color bands, with 1.2 to 4.8 pixels across the selected asset-size range.
Anvilfield research estimates 5 pixels across a 3 m asset in 0.6 m aerial imagery, with 2.5 to 10 pixels across the selected asset-size range.
Anvilfield research estimates 9.7 pixels across a 3 m asset in 0.31 m satellite panchromatic imagery, with 4.8 to 19.4 pixels across the selected asset-size range.
Anvilfield research estimates 10 pixels across a 3 m asset in 0.3 m aerial imagery, with 5 to 20 pixels across the selected asset-size range.
Anvilfield research estimates 19.7 pixels across a 3 m asset in 0.1524 m aerial imagery, with 9.8 to 39.4 pixels across the selected asset-size range.
| Anvilfield estimate | Low | Central | High |
|---|---|---|---|
| Pixels across asset: 30 m satellite | 0.05 | 0.1 | 0.2 |
| Pixels across asset: 10 m satellite | 0.15 | 0.3 | 0.6 |
| Pixels across asset: 1.24 m satellite color | 1.2 | 2.4 | 4.8 |
| Pixels across asset: 0.6 m aerial | 2.5 | 5 | 10 |
| Pixels across asset: 0.31 m satellite pan | 4.8 | 9.7 | 19.4 |
| Pixels across asset: 0.3 m aerial | 5 | 10 | 20 |
| Pixels across asset: 0.1524 m aerial | 9.8 | 19.7 | 39.4 |
| Short side at review gate: 0.31 m satellite (m) | 1.24 | 1.86 | 3.1 |
| Short side at review gate: 0.6 m aerial (m) | 2.4 | 3.6 | 6 |
| Short side at review gate: 0.1524 m aerial (m) | 0.61 | 0.91 | 1.52 |
| Mean usable-view wait at 5-day cadence (days) | 2.5 | 7.5 | 17.5 |
| Wait to next capture on 2-year cycle (months) | 0 | 12 | 24 |
| Age of NYC 2024 views on Nov. 5, 2026 (days) | 956 | 961 | 966 |
How to cite this estimate: Anvilfield, Satellite vs aerial imagery for asset finding, 2026, https://anvilfield.com/field-guides/market-intelligence/satellite-vs-aerial-imagery-for-asset-finding/
How many pixels does an asset need?
Use the short side of the feature you need to distinguish. A long, narrow object can occupy a large patch of the image while offering little evidence across its width. Shadows, blur, roof clutter, view angle and contrast further affect interpretation. Ask a reviewer to explain which visible feature supports the asset label.
Anvilfield uses 6 pixels across the short side as a review gate for this comparison. Passing the gate sends an object to classification review. The gate is an operating policy that must be checked against labeled examples for the asset class. Location, equipment family, component count and operating condition remain separate claims.
Anvilfield research estimates a central short-side review threshold of 1.86 m for 0.31 m satellite imagery, with a policy range of 1.24 to 3.1 m.
Anvilfield research estimates a central short-side review threshold of 3.6 m for 0.6 m aerial imagery, with a policy range of 2.4 to 6 m.
Anvilfield research estimates a central short-side review threshold of 0.91 m for 0.1524 m aerial imagery, with a policy range of 0.61 to 1.52 m.
Below the gate, route the site to a finer view or supporting records. Above it, inspect the distinguishing features and compare accepted and rejected examples. Keep the original evidence crop so a second reviewer can check the decision. A fan count, cabinet count and complete equipment-assembly count need separate definitions.
Distinguish cooling towers, chillers and dry coolersRooftop equipment aerial-survey limitsReviewing machine-detected assets: a quality checklistDoes a sharper color image prove more detail?
Check native sampling for each band. Panchromatic imagery records a broad band of brightness. Multispectral imagery records separate wavelength bands. A sharpened color product combines information from those sources, so its displayed pixel size does not describe the native sampling of every color measurement. The USGS reference reports different sampling for the panchromatic and color bands. [1]
That distinction matters when the sales rule depends on surface appearance. A sharp roof edge helps locate a boundary; a color pattern needs its own evidence before it supports a material label. Enlarging an image changes the display. It adds no new source observations. Ask for the native specification and processing description, particularly when a product advertises an enhanced resolution.
Keep image detail separate from location accuracy. An asset can look distinct while the image and parcel layer disagree at a boundary. Compare stable surrounding features before attaching the observation to a building. Preserve the asset point, building identifier and parcel identifier separately so a later ownership correction does not erase the original finding.
Take the remaining technical question into the field. A roof color needs material and installation records. A solar array outline needs mounting and electrical review. A tank outline needs asset identity and operating records. Those handoffs turn useful imagery into a service scope without adding unobserved properties to the prospect file.
[1] USGS system characterization report, satellite sampling (2021, version 1.1 (October))Commercial roof inspection and maintenance recordsRooftop solar mounting and rackingAboveground storage tank inspectionHow fresh must imagery be for finding and monitoring assets?
Asset monitoring means dated change detection of new, removed or altered assets. Start by naming the change that deserves action: a new enclosure, a removed tank or an expanded paved yard. The evidence needs enough detail to distinguish that change and an observation interval useful to the receiving team.
Anvilfield research estimates a mean wait of 7.5 days for a usable observation of a permanent change at the coarse reference mission's nominal 5-day cadence, with a range of 2.5 to 17.5 days across usable-coverage cases.
Anvilfield research estimates a mean capture wait of 12 months on a regular 2-year aerial cycle, with a timing range from 0 to 24 months for permanent visible changes.
Anvilfield calculates that NYC's March 2024 imagery is 961 days old at the capture-window midpoint on November 5, 2026, with a range of 956 to 966 days across the published dates.
The city example demonstrates why a detailed archive remains an archive. Use its capture dates as the historical baseline, then seek a later suitable observation when the decision depends on current equipment. Keep capture, analyst review and field verification dates separate. A recent download or a refreshed web map does not reset the underlying capture date.
Set the monitoring requirement in usable observations. Cloud, shadows, obstruction and unsuitable viewing geometry interrupt optical evidence. Delivery and review add further elapsed time. Retain the before and after capture dates with every change finding; they establish an observation interval rather than an exact installation day.
[3] New York City aerial imagery metadata (2024 capture)[4] Copernicus SentiWiki mission specifications (Live technical reference accessed 2026-11-05)Asset monitoringChange detection: spotting new installations earlyMonitoring cadences: one-time, quarterly and daily alertsField benchmarks for comparing imagery
Apply these Anvilfield operating benchmarks in the acceptance brief and retain the review result with the delivery.
| Rule of thumb | Benchmark | Practical decision |
|---|---|---|
| Measure the short side. | Pixels across = short-side metres / GSD metres. | Check the feature that distinguishes the asset. |
| Use a stated sampling gate. | 6 pixels across the short side is this article's review gate. | Request better evidence below the gate; review classification above it. |
| Translate the gate into a purchasing specification. | A 3 m asset needs GSD of 0.5 m or finer at the 6-pixel gate. | Compare native detail before comparing platform names. |
| Keep display enhancement separate. | Resampling adds 0 new source observations. | Save native band sampling with the delivered file. |
| Preserve the evidence status. | Use 3 outcomes: supported, rejected and unresolved. | Keep obscured areas in the unresolved queue. |
| Date each evidence stage. | Keep 3 dates: capture, review and field verification. | Measure freshness from the capture supporting the claim. |
| Name the thing being counted. | Assign 1 identifier per distinct asset assembly. | Track components, buildings and parcels separately. |
| Require a before and after. | Use 2 comparable dated observations for a change finding. | Report the change interval and the evidence supporting it. |
Best practices for choosing imagery around the commercial task
Fit-first lists are lists filtered to the buildings that fit a company's work. Installed-base mapping means locating every unit of a product class in the field, any brand. A standard building filter and an equipment inventory therefore need different acceptance rules. Set the physical-fit decision first, then specify how much imagery evidence that decision requires.
Manufacturers, distributors and service companies can measure physical fit building by building with PorchRocket's Ground Control, whose standard commercial filters include roof age band, material, size, business type, rooftop HVAC, parking areas, storage tanks and loading docks.
For a building selection task, ask whether the relevant feature is present and whether the image date supports the campaign. For an inventory task, define separable assemblies, obscured areas and duplicate views. For a pavement campaign, agree on the service boundary. For a generator campaign, distinguish a visible enclosure from an equipment identity confirmed by records.
For a custom asset class, PorchRocket's Asset Search finds and monitors visible assets across the lower 48 and returns the owner of record with each find.
Write the exception rules into the purchase brief. A campus needs a building-level output even when the parcel is shared. A partly hidden equipment yard needs an unresolved status. A rejected candidate needs its rejection reason. Those instructions make a pilot useful because the team can compare what was requested with what the imagery actually supported.
Compare total effort through the agreed release stage. Include image rights, analyst review, corrections, ownership research and refresh work in the buying decision. Public access to an image answers an access question; its permitted reuse and suitability for the intended deliverable still deserve an explicit check. Keep the commercial decision tied to usable records.
Fit-first listsInstalled-base mappingThe economics of buying vs building asset data in-houseParking lot layout and accessibility fieldworkStandby generator maintenance recordsWorked example: a regional HVAC distributor chooses coverage
A regional heating, ventilation and air conditioning (HVAC) distributor screens 1,000 commercial buildings for a rooftop-equipment campaign. Its smallest target assembly has a 3 m short side. The analyst applies the 6-pixel gate before classifying equipment. The branch needs evidence suitable for account research and an eventual equipment survey.
The coverage file contains 600 buildings with 0.1524 m aerial views and 400 buildings with 0.6 m aerial views. At the target width, the finer views supply 19.7 pixels across the asset; the coarser views supply 5. The first group passes the sampling gate. The second group enters the request-for-better-evidence queue.
The distributor requests suitable 0.31 m satellite views for those 400 buildings. They provide 9.7 pixels across the same target, so the full 1,000-building file passes the geometric gate after the finer views arrive. Each property still needs a usable capture date, visible target area and classification review. The gate releases a review task, not a claim that every building has the equipment.
Anvilfield research estimates 83.3 analyst hours for this typical coverage-and-review case: 50 hours for the initial pass and 33.3 hours for the finer-view follow-up. The 40% escalation share directs the additional effort to the buildings whose initial evidence fails the sampling rule.
The branch keeps older detailed views as historical evidence and uses later suitable views for current observations. A building with unclear equipment identity stays unresolved. A supported cabinet arrangement advances to ownership and service-responsibility research. The field team then confirms the equipment schedule, model information and actual service scope.
That sequence lets the purchasing team compare a mixed-source option with a uniform collection. It also exposes where the workload sits: broad initial screening, targeted evidence improvement and asset classification. A lower image price loses its advantage when the delivered detail creates a larger unresolved queue.
Rooftop unit installation and startup evidenceFind commercial HVAC accounts using rooftop-unit densityImagery acceptance checklist
Review the record against the commercial claim before handing it to a rep or branch. A visible object, an asset class, a count and a condition assessment require different evidence. Keep the observation and interpretation in separate fields so a reviewer can correct the label without destroying the underlying finding.
The CDC identifies small, hidden, underground and shadowed cooling towers as difficult to find from imagery. Its guidance combines image review with records and ground investigation. Use that visibility lesson when building a commercial research file: an obscured area requires another evidence path. A blank observation field does not establish an empty site. [6]
Keep the acceptance checklist with the scope and the resulting file. When a field visit overturns a classification, preserve the reason and use it to improve the next batch of reviewed examples. Track the unsupported claim that caused the correction, rather than treating every correction as a general failure of the platform.
[6] CDC Strategies for Identifying Cooling Towers (2026)Stormwater pond maintenance scopeAboveground tank inspection prospect qualificationWant this checklist to run itself on every job — with photo proof and a signed record crews can hand the customer? That's FieldOS.
How do leaders turn imagery into useful building records?
Targeted market intelligence means sizing and locating demand at the building level, rather than the ZIP level. The useful output connects a physical observation, a building, a fit reason and a next action. Teams get more value from imagery when those relationships survive a change of analyst, sales rep or territory assignment.
Keep a stable building identifier and link the dated asset observations to it. Retain rejected and unresolved observations alongside supported ones. An analyst should be able to explain why a building entered a list, what remains unknown and which evidence would settle the question. That structure also keeps a new capture from silently overwriting the history.
Teams planning repeated coverage can scope PorchRocket's Asset Search around custom finding and monitoring, with pricing set per scope.
Define the acceptable observation interval, the change worth reporting and the handling of unsupported cases. Give the receiving rep a task: verify the operator, request the equipment schedule or arrange an authorized survey. A property owner of record is an ownership finding; confirm who controls the specific maintenance or replacement budget.
Use field feedback to improve the next selection. Preserve why an apparent tower proved to be another equipment class, why a lot belonged to a neighboring parcel or why an older image missed a replacement. The result is a more useful evidence standard for the next purchase and a clearer handoff between research and sales.
Targeted market intelligenceMethodology
Methodology, November 5, 2026. Public inputs are the nominal sampling and dates in the linked primary sources. The USGS report is the October 2021 version; USDA supplies historical 2018 aerial reference scales; NYC supplies its 2024 collection specification, capture interval and stated repeat cycle. NASA and Copernicus supply mission specifications opened for this edition. These references establish comparison scales rather than territory-wide availability.
Anvilfield modeling model settings: exposed asset short sides of 1.5, 3 and 6 m; orthogonal sampling; and review gates of 4, 6 and 10 pixels, with 6 as the central operating policy. Pixel span equals width divided by GSD. Gate width equals gate pixels multiplied by GSD. Convert the city specification using 0.0254 m per inch, giving 0.1524 m. Pixel ranges represent selected asset sizes; gate ranges represent policy sensitivity. Neither range is a measured accuracy interval.
The nominal 5-day timing model uses permanent detectable changes uniformly timed between looks and independent usable-look probabilities of 100%, 50% and 25%. Mean wait equals half the interval plus interval multiplied by (1 minus usable probability) divided by usable probability. It excludes review and delivery time. Correlated cloudy periods require a local coverage model. This cadence belongs to the coarse reference mission; the fine satellite rows use only its separate sensor specifications.
The aerial timing row uses the documented 2-year cycle, 12 months per year, uniform permanent changes and usable imagery at every scheduled capture, available immediately. Its endpoints are minimum and maximum waits; the central value is the mean. Archive age subtracts March 14 to March 24, 2024 from November 5, 2026. The central age uses the capture-window midpoint without area weighting. The calculation makes no claim about whether a newer city collection is available.
The worked-case model settings are 1,000 buildings, a 3 m target, 600 fine-aerial and 400 coarse-aerial sites, and suitable finer satellite views for the latter group. Initial review takes 3 minutes per building; follow-up takes 5 minutes per escalated building. Sum minutes and divide by 60 for hours. Image acquisition, ownership research, travel, field surveys and outreach sit outside this workload calculation.
For a standard-list purchasing comparison, PorchRocket's Ground Control publishes commercial pricing of $10 to $20 per building and delivery within 48 hours after payment.
The AF-M092 script retains unrounded results and source snapshots. These geometric ranges describe sampling and operating choices; recognition accuracy requires a labeled evaluation of the actual asset and product.
[1] USGS system characterization report, satellite sampling (2021, version 1.1 (October))[2] USDA NAIP technical presentation (2018)[3] New York City aerial imagery metadata (2024 capture)[4] Copernicus SentiWiki mission specifications (Live technical reference accessed 2026-11-05)[5] NASA Landsat 8 specifications (Mission launched 2013; reference accessed 2026-11-05)[6] CDC Strategies for Identifying Cooling Towers (2026)| Public input | Year | Value used or role |
|---|---|---|
| [1] USGS sensor characterization | 2021, October version | 0.31 m panchromatic and 1.24 m native color sampling |
| [2] USDA NAIP presentation | 2018 | 0.6 m and 0.3 m aerial reference scales |
| [3] NYC aerial metadata | 2024 collection | 6-inch resolution; March 14 to 24 capture; stated 2-year cycle |
| [4] Copernicus mission documentation | 2026 access | 10 m visible sampling; nominal 5-day revisit |
| [5] NASA mission documentation | 2013 mission; 2026 access | 30 m reflective sampling |
| [6] CDC identification guidance | 2026 update | Visibility limits and need for records and ground investigation |
| [7] Ground Control commercial page | 2026 access | Standard filters; $10 to $20 per building; within 48 hours after payment |
| [8] Asset Search page | 2026 access | Custom visible-asset finding and monitoring; owner of record; scoped pricing |
FAQ
Is aerial imagery always more detailed than satellite imagery?
Compare the product specifications. Anvilfield research estimates that a 3 m asset spans 9.7 pixels in 0.31 m satellite imagery and 5 pixels in 0.6 m aerial imagery. The platform label alone does not rank those products. Check native detail, capture date, viewing geometry and visibility together.
Can free regional satellite imagery count individual rooftop units?
Anvilfield research estimates that a 3 m asset spans 0.3 of a pixel at 10 m sampling and 0.1 of a pixel at 30 m sampling. Those scales support regional context. An individual cabinet inventory needs finer evidence and a defined classification review.
What resolution should I specify for a small asset?
Specify the distinguishing short-side dimension and the sampling gate together. A 3 m feature needs 0.5 m GSD or finer at this article's 6-pixel gate. Check actual labeled examples before accepting an equipment classification or a count.
Does pan-sharpening make all color measurements equally detailed?
A sharpened composite combines finer brightness information with separate spectral measurements. Preserve the native sampling of the contributing bands. A sharp boundary supports a different claim from an inferred surface material, equipment brand or condition.
Does a short revisit interval guarantee a current usable image?
A revisit specification describes collection opportunities. Cloud, shadow, viewing geometry and obstruction affect whether a particular observation answers the asset question. Specify usable coverage and include processing, review and delivery when setting the reporting interval.
How should older aerial imagery be used?
Keep it as dated historical evidence when its detail supports the feature of interest. Seek a later suitable view for a current observation. Preserve both dates when reporting a change, and keep the capture date separate from the date an analyst downloaded or reviewed the image.
Can imagery identify the person who buys maintenance?
Imagery establishes a physical observation. Property records support ownership research. An operator, tenant, facility manager or portfolio office can control the relevant work, so confirm the buying role for that scope before treating the record as a sales opportunity.
When should the team request a site survey?
Request an authorized survey when the decision depends on concealed features, equipment identification, condition, access or dimensions that the available views do not resolve. Send the field team the dated evidence and the specific unresolved question so the visit closes the information gap.
People also ask
Sources checked
- [1] USGS system characterization report, satellite sampling 2021, version 1.1 (October). Opened and downloaded with curl for the 2026-11-05 edition; HTTP 200. Saved snapshot and checksum in the AF-M092 source manifest.
- [2] USDA NAIP technical presentation 2018. Opened and downloaded with curl for the 2026-11-05 edition; HTTP 200. Saved snapshot and checksum in the AF-M092 source manifest.
- [3] New York City aerial imagery metadata 2024 capture. Opened and downloaded with curl for the 2026-11-05 edition; HTTP 200. Saved snapshot and checksum in the AF-M092 source manifest.
- [4] Copernicus SentiWiki mission specifications Live technical reference accessed 2026-11-05. Opened and downloaded with curl for the 2026-11-05 edition; HTTP 200. Saved snapshot and checksum in the AF-M092 source manifest.
- [5] NASA Landsat 8 specifications Mission launched 2013; reference accessed 2026-11-05. Opened and downloaded with curl for the 2026-11-05 edition; HTTP 200. Saved snapshot and checksum in the AF-M092 source manifest.
- [6] CDC Strategies for Identifying Cooling Towers 2026. Opened and downloaded with curl for the 2026-11-05 edition; HTTP 200. Saved snapshot and checksum in the AF-M092 source manifest.
- [7] PorchRocket Ground Control commercial lists Live product reference accessed 2026-11-05. Opened and downloaded with curl for the 2026-11-05 edition; HTTP 200. Saved snapshot and checksum in the AF-M092 source manifest.
- [8] PorchRocket Asset Search Live product reference accessed 2026-11-05. Opened and downloaded with curl for the 2026-11-05 edition; HTTP 200. Saved snapshot and checksum in the AF-M092 source manifest.