I'd like to identify mountain-peaks from an SRTM TIFF (created/composed from the original HGT ones).
How I'd wish it would work: Check for each pixel whether it is the highest value within a given radius (and I'd like to play with this radius). I know... it will generate some noise, but I will deal with it by having a larger "moving-frame" (i.e. bigger radius) up to 1-2 km.
This entry: how-to-calculate-peaks-in-a-dem asks basically the same and especially the (second) comment below the question from user/whuber phrases the problem perfectly - but I'd need a description (somewhat step-by-step) how to execute it.
The first (and chosen) answer would have been a nice start, but executing r.param.scale only throws an error ("Datenquelle ungültig" - "data-source invalid"), even though the same layer/file perfectly works with other GRASS tools.
Turns out r.param.scale works perfectly, but you need to work in a projected coordinate system. So simply reprojecting to such a projected coordinate system solved my troubles.
within the r.param.scale one has to choose for the 'Morphometric parameter' and this will categorize all pixels from 1 to 6, being 1 = planes, 2 = pits, 3 = channels, 4 = passes, 5 = ridges, 6 = peaks.
As expected the results vary vastly depending on the chosen parameters.
I am aware of this entry too: how-to-extract-mountain-peak-points-from-a-given-digital-elevation-model But there is not much practical advice given.
It seems this is a somewhat popular question brought up sporadically by amateurs like myself...