measurement
The Capture Date Decides Whether the Measurement Is Real
Aerial imagery is a series of snapshots, not a live view. Which snapshot a measurement came from is often the only thing that matters about it.
Every aerial measurement is a measurement of a photograph, and every photograph was taken on a day. Software presents the result as current. It is current as of that day and no later.
For most work that distinction is academic. For three situations it decides whether the number means anything, and all three are common in exterior trades.
After an event
A hailstorm passes on 14 June. A measurement run on 20 June may well be derived from imagery captured the previous spring.
That is not a defect. It is often the most useful imagery available, because it documents the roof as it was before the loss. But it is a measurement of the prior structure, and using it as a measurement of what is there now is a category error.
The useful discipline is to hold the two dates side by side: the date of loss, and the capture date of the image the measurement came from. The relationship between them determines what the measurement is evidence of. Before the loss, it is evidence of prior condition and prior geometry. After the loss, it is evidence of the current structure. Those are different claims and they should not be made with the same document.
Event dates themselves are published and queryable. The National Weather Service's API documentation covers the warning and observation products, which is where a defensible date of loss comes from rather than from a customer's recollection.
After a change to the building
An addition, a dormer, a new garage, a converted porch. Any of these makes imagery from before the change confidently wrong, by exactly the area of the change, with no indication in the output that anything is missing.
This is the failure mode that produces an under-order rather than a disagreement, because the measurement is internally consistent. It measures a real roof, accurately. Just not the one on the property today.
The only cheap defence is a glance: open the image, look at the building, and ask whether it matches what the customer described. Ten seconds, and it catches the case where the imagery predates a visible extension.
On new construction
A house finished this year may not appear in any available imagery. Depending on the provider's handling, the result is either a refusal, which is fine, or a measurement of whatever was on the lot, which is not.
A measurement returned for a property with no building in the image is the most dangerous output in the category, because it looks like every other result.
How often imagery refreshes
Refresh rates vary by provider and, more importantly, by where the property is. Dense urban areas are re-flown far more often than rural ones, which means the reliability of remote measurement is correlated with population density. That correlation is rarely stated and it is worth knowing when quoting a rural job from the office.
Public programmes publish currency and coverage. The USGS National Map is the reference for what exists federally and how recent it is, and commercial providers publish their own refresh commitments. Both are worth reading once, for your actual service area, rather than assuming a national average applies to it.
What to require of a tool
Three things, and the first is not negotiable.
The capture date, visible on the output. Not available on request, not in the metadata. On the report, next to the number.
A refusal when there is nothing to measure. A tool that returns a number for a vacant lot has chosen to produce an answer over producing a correct one.
A way to request fresher imagery, or to supply your own. When the available capture is too old, the escalation path should exist in the product rather than requiring a different workflow entirely.
What to do when the date is wrong for the job
Three options, in order of cost.
Use it anyway and say so. For a budgetary number on an unchanged building, year-old imagery is fine, and stating the capture date on the quote keeps everybody honest.
Verify the specific thing in doubt. If the concern is one addition, one photograph from the ground settles it. The rest of the measurement is still good.
Capture it yourself. Costly in time and subject to its own rules, and the right answer when the property has changed materially or the imagery predates a loss you are documenting.
What is not an option is quoting from a number whose capture date you did not look at. That is not a measurement error. It is a measurement of a different building, made with confidence.