FAQ

Longley–Rice, the Irregular Terrain Model (ITM), run over real elevation data. Terrain diffraction is computed along each path rather than approximated, landcover attenuation and tree-canopy loss are applied, and building footprints enter as solid knife edges.

The ITM time-variability term is applied, so you can ask for a confidence level — the map for "signal 50% of the time" and the map for "signal 99% of the time" are different maps.

USGS 3DEP. Two tiers matter in practice: 1 m LiDAR where it has been flown, and 1/3 arc-second — about 10 m — across CONUS everywhere else. The tiers are blended across their boundary rather than switched, so a resolution change does not show up as a seam in the coverage.

Which tier you are on is the single biggest control on how fine a render is worth paying for. See Resolution & cost.

It is a prediction. Longley–Rice is a statistical model with published uncertainty, and that uncertainty is a property of the model, not of this implementation; running it faster does not make it more certain. In addition, it inherits any errors in the underlying terrain and clutter data.

We do not publish an accuracy figure for this service, because we have not measured one we would be willing to defend. If your field measurements disagree with the map, the field measurements are right.

No. This performs no coordination, produces no study anyone is obliged to accept, and makes no claim about interference to or from any licensed facility. It is a planning tool. Anything with a regulatory consequence requires a process designed for that purpose, and this service is not one.

No. It models outdoor path loss over terrain and ground cover. Buildings are obstructions in the path, not volumes with an inside. There is no building-penetration term and no floor height.

Yes, from a built-in catalogue of named patterns. You pick one by name and give the antenna an orientation; you do not upload a pattern.

What is in the catalogue:

• Generic shapes — omnidirectional, cardioid (for suppressing radiation toward a protected direction), Yagi–Uda at a range of forward gains, and corner reflector.
• Named commercial LMR models, spanning low band, VHF, 220, UHF, 700, 800 and 900 MHz — exposed-dipole and folded-dipole arrays, collinear verticals, offset and directional types.

Both planes are applied. The azimuth (H-plane) pattern and the elevation (E-plane) pattern are both applied for the chosen antenna, so vertical beamwidth and null-fill affect the result rather than being ignored.

What is not supported: importing your own .PAT or .NEC pattern files. If your antenna is not in the catalogue, select the closest match. Sectorized panel antennas of the kind used in cellular and microwave point-to-point work are out of scope.

Credits are counted from the finished image and from two things you choose about how it is made:

credits = ceil( output megapixels × supersample² × transmitters )

All three are fields in your own request, so the cost is computable by you before anything runs, and unaffected by how the service divides the work internally. The reasoning, and why supersampling and transmitter count are in there, is on the pricing page.

NOT SET. The price per credit has not been decided, and this site deliberately contains no figure that could be mistaken for one. The pricing page lists every open parameter.

Prepayment means there is no bill after the fact: a job is priced and checked against your balance before it is accepted. You also do not pay for months in which you do not render.

The cost of this arrangement to you is that we hold your payment before the processing is performed. Credits do not expire, so the balance remains available until you use it. That trade-off is set out in full on the pricing page.

Range is not a factor in the formula as it stands, so at a fixed output size a 100 km map and a 10 km map are the same number of credits. Treat that as NOT FINAL rather than a promise — whether range gains a factor is open, and it is listed with the other undecided parameters on the pricing page.

What range does change is the ground sample: the same pixels spread over ten times the distance. That is a resolution decision, and it is the one most worth considering.

Yes — as the square of the setting. Supersample ×2 is four times the credits, ×3 is nine times.

The reason is that supersampling is passes: the scene is rendered that many times and the result is resolved into an output image of unchanged size. Because the image dimensions do not change, the extra work does not show up in a megapixel count, which is why the megapixel count alone is not a fair unit and why supersample² is in the formula.

When you want a smoother-looking map rather than more detail, supersampling is still the right setting to use in place of a finer grid. It is simply not free.

Because elapsed time is not a property of your request. The same job finishes at different speeds depending on current demand on the service, and a price that cannot be quoted in advance is not a price.

The largest we have measured is 16384 × 16384 — 268 megapixels, taking a mean of 152 seconds across 29 such renders. Typical work is between 512 and 8192 pixels per side.

A render of that size is only worthwhile if the terrain supports it. Check the terrain first.

Yes. The minimum purchase is $30.

Yes. Larger purchases receive a lower rate per credit. The volume tiers are still being finalised, and are listed as open on the pricing page.

Yes. Every new account receives $20 of free credit.

Credits are charged per render, and any part of a credit is rounded up to a whole credit. A 4096 × 4096 render at supersample ×1 is 16.8 megapixels and is charged 17 credits.

Credits do not expire. Unspent credit is not refundable at this time.

Yes. Texas sales tax applies to this service, and the amount is shown at purchase, before you pay.

The intended behaviour is that this cannot happen: a job is priced and checked before it is accepted, and either runs to completion or is declined up front with the shortfall shown. Renders are not stopped part-way to enforce a balance, because a partly finished render would be work you were charged for that never produced an image.

A render that fails for any other reason is not charged, and can be run again.

• A georeferenced map overlay for a web basemap.
• A tile pyramid, so the result can be panned and zoomed without re-rendering.
• KMZ, for Google Earth and any other application that reads it.

Yes, in two ways, which answer different questions. Best server takes the strongest signal at each pixel, which is the right method for a trunked or roaming system where a radio selects one site. Sum of power adds the signal from all sites, which is the right method for simulcast, or for determining the total RF energy reaching a location.

Choosing the wrong method produces a plausible map that answers a different question.

A field. You get predicted signal strength across the area and choose your own threshold afterwards, rather than accepting a predetermined definition of where "coverage" ends. A contour is a decision; this gives you the data the decision is made from.