sweep
Sweep radials out from site and return received signal strength at every sample point.
Terrain profile resolution and receiver spacing are deliberately decoupled. P.1812 evaluates diffraction across the whole profile between transmitter and receiver, so the profile has to be sampled near the terrain's own resolution (Mapterhorn at z11 is ~75 m/px). Sampling it at the receiver spacing instead (900 m over a 30 km radius) makes the model see a jagged, aliased profile and invent diffraction loss that is not there: visible as spurious rings and spokes in the plot, and a badly depressed reachable fraction.
The expensive part is the number of P1812.predict calls, not the profile length, so a dense profile with sparse receivers is both more accurate and no more costly.
Parameters
the transmitter
how many bearings to sweep
how many receiver positions to evaluate along each radial
spacing of the terrain profile itself