A contractor offers a terrain surface “ready for drainage analysis.” Before accepting it, ask for the surface before sink filling, the surface afterwards, and a way to locate the cells that changed. A depression can disappear during processing without a shovel touching the site. A drainage-ready model and a record of surveyed ground answer different questions. On a UG21 survey purchase, name both products and designate which one is evidence of ground elevation.

The low spot next to the service road

Consider a facilities manager investigating water near a loading yard. An old plan shows a low strip alongside a service road. A new map appears to show a smooth path from the yard into a ditch. The contractor explains that the surface was conditioned before the drainage arrows were drawn. Does the map establish that runoff can leave the yard? Not on its own. It shows what the analysis concluded on the surface it received. The buyer still needs to know whether the route existed in the original terrain product.

That distinction matters well before anyone models a storm. The maintenance crew needs to know where to look for standing water. The designer may need a connected surface to explore a potential flow path. The asset owner may need a defensible baseline for comparing future changes. Giving all three the same file, labelled simply “final terrain,” invites each person to read a different promise into the same numbers.

Ask each recipient to finish one sentence: “I will use this file to decide whether to ...” A crew deciding where to walk the site may need the unresolved depressions left visible. An analyst testing a drainage scenario may deliberately want some pits removed. Those are compatible requirements if the contractor is allowed to deliver separate layers. They become incompatible when every recipient is promised that one edited surface is the original record.

What the fill limit actually selects

In Esri Fill (Spatial Analyst), Esri describes a threshold that compares a sink's depth with the elevation of its pour point. Think of the pour point here as the level at which the depression can spill. For this tool, the limit helps choose which depressions to fill. It does not mean that the software adds a fixed thickness everywhere it encounters a pit. Other tools and settings need their own descriptions; do not extend this interpretation to an unnamed workflow.

Use a deliberately simple test in the bid interview. Suppose the low cell is at 98 metres and the spill level is 101 metres in the same vertical reference. The difference is three metres. With a two-metre limit the depression would not qualify under the described rule. With a four-metre limit it would qualify and be filled to its pour point. The example is about an algorithm, not the accuracy of a survey or a recommended project setting. A bidder should be able to explain the difference without changing the question to flight performance.

What if the analyst says the site has a three-metre fill allowance? Ask whether they mean a selection threshold, a maximum alteration to any cell, or a permitted engineering change to the site design. Those statements sound similar over the phone. In a tender they are separate requirements, and each needs a different acceptance test. Record the vertical units with the value, not just in somebody's meeting notes.

Specify the two surfaces and their relationship

Call the first deliverable the terrain input to hydrologic conditioning. Do not call it raw sensor data unless it really is. It may already reflect classification decisions, interpolation or editing. Request a description of those steps and the extent within which the elevations can be interpreted. Then give the conditioned surface a different file name, revision and description. The important question is whether the second can be traced back to the exact first file the buyer received.

Request a change layer as well. For this purchase, define it explicitly as conditioned elevation minus input elevation, on corresponding cells. Give unobserved areas an explicit missing-data treatment. An unobserved cell is not an unchanged one, and treating both as zero makes a difference map look more conclusive than it is. Ask the contractor to state whether the two grids share the same cell alignment and units before any subtraction.

For a manageable field check, select a few locations by purpose rather than by appearance. Include a shallow feature that the contractor expects to fill, a deeper hollow expected to remain, and a location outside both. Keep the sample identifiers tied to actual coordinates. A screenshot over a basemap is helpful for discussion but rarely lets a second analyst identify exactly which elevation pair the first reviewer compared.

Read a change map without diagnosing the site from the desk

Suppose the largest patch of changed cells sits behind an embankment. One possibility is a real storage area. Another is a drainage connection that the surface does not represent. A third is an artifact of the input data. The differences establish where the operation altered the model; they do not select among those explanations. Put the patch on an exceptions list and decide who can supply independent evidence before it is used in a design conclusion.

Useful evidence can be modest. The engineer may already hold a culvert inventory. The site supervisor may be able to identify an outlet that is obscured in the imagery. A survey specialist may recommend a separate check of the controlling elevation. Record what each item supports. An inventory entry can justify investigating a connection; it does not by itself supply a verified elevation or prove that a drain was functioning on the survey date.

For repeat surveys, carry forward the decision about conditioning, not just the number entered in a dialog. If this year's contractor uses a different boundary, a different input or a different treatment of a culvert, subtracting the two final products can mix physical change with changed assumptions. Keep an unconditioned comparison available. Where the inputs are not comparable, label the uncertainty instead of treating a tidy change polygon as ground movement.

A practical acceptance meeting

Have the reviewer open the input surface first, without the contractor's flow arrows. At the test location, note the low point, candidate spill level and the source of each number. Turn on the conditioned layer and examine the same cells. Finally switch on the difference layer. This order prevents a convincing drainage graphic from becoming the premise of the test. The person signing the deliverable should be able to reproduce at least one conclusion on their own workstation.

Use separate outcomes for delivery and interpretation. A correct difference raster can pass the file test even when a particular depression still requires a site visit. Conversely, a sensible explanation from the project engineer does not repair a difference layer computed on shifted grids. Write the unresolved question, its owner, and the use that must wait. That keeps a technical qualification attached to the decision it affects rather than buried in the minutes.

A useful contract clause is short: supply the pre-conditioning terrain, the derived surface, and a spatial record of changes; identify the software, settings, vertical units and exceptions; demonstrate one feature kept and one changed. Let the bidder propose a test site appropriate to the job. Avoid requiring a preserved or removed feature where the actual terrain offers no valid example. A small synthetic demonstration can show the rule, but it cannot validate the site survey.

What to ask about the aircraft

The UG21 fixed-wing VTOL aircraft is an aircraft to consider for mapping work, not a designation for a particular processing service. Ask the supplier to separate aircraft configuration, data acquisition, terrain production and drainage interpretation in the quote. A successful acquisition does not show that a specific surface meets the buyer's downstream requirements. Equally, a processing problem need not mean that the aircraft or the capture has to be replaced.

When comparing the VTOL and fixed-wing aircraft collection, give bidders the same deliverable checklist. Ask who retains the input, who processes the derivative, who will explain changed cells and who can answer a question after the delivery portal closes. A lower-priced offer may simply omit one of those responsibilities. Clarify the omission before comparing totals. For a coverage boundary on a linear site, define the required bank of a one-sided corridor strip; for a UK site close to an aerodrome, separate UK aerodrome permission from land access.

Put one depression in the inquiry

Choose one low area that matters to the project and describe the decision it drives. Is the customer trying to inspect ponding, locate a possible outlet or compare a before-and-after grade? State which decisions are out of scope. Provide a sketch, the required coordinate reference, intended recipients and the date when the result is needed. Keep the example small enough that the vendor can show exactly what it would return.

At closeout, store the change layer beside its two named inputs, not in a general folder of project illustrations. A reviewer looking at one coloured patch months later needs the corresponding elevations, not just the legend. Record explicitly if another team later replaces either input; the original comparison then belongs to the earlier revision and should not be presented as an assessment of the new one.

Use the UNITED UAV configuration inquiry to request configuration details and a sample handover with distinct file identities. Ask for a brief written answer to this final question: if a depression disappears from the drainage layer, how will the owner recover its original elevation and discover why it changed? A supplier who can point to the input, rule and difference record has described something the buyer can actually check.