[GRASS-user] Upscaling an elevation model

[Nikos Alexandris]

* Helmut Kudrnovsky <hellik at web.de> [2019-02-02 08:17:03 -0700]:

>NikosAlexandris wrote
>> Dear "elevation" experts,
>>
>> I am looking for ways to upscale a 5m resolution elevation model.  The
>> upscaled models will be subjected to a series of trials using
>> `r.watershed`.  Derivative products will be inputs for soil erosion
>> modelling.
>>
>> In our wiki page about Interpolation [0], `r.resamp.stats` [1] is
>> presented as the only tool to resample a raster map to a coarser grid.
>> However, `r.resamp.rst` [2] also performs resampling to a lower
>> resolution.
>>
>> [0] https://grasswiki.osgeo.org/wiki/Interpolation
>> [1] https://grass.osgeo.org/grass76/manuals/r.resamp.stats.html
>> [2] https://grass.osgeo.org/grass76/manuals/r.resamp.rst.html
>>
>> While the target is to somewhat maintain "geomorphometric consistency"
>> across the different (coarser) scales, which tool is recommended?
>> For `r.resamp.stats`, would you recommend a resampling method other than
>> 'average'?
>>
>> Thank you, Nikos
>> _______________________________________________
>> grass-user mailing list
>
>> grass-user at .osgeo
>
>> https://lists.osgeo.org/mailman/listinfo/grass-user
>
>you can see
>
>https://grass.osgeo.org/grass74/manuals/addons/r.valley.bottom.html
>
>"[....]The MRVBF index assesses the flatness and lowness of terrain over
>multiple scales and DEM resolutions[...]"
>
>how it's there.

Ahoy Helmut,

resampling takes place through the helper function `refine()` [0]. Then,
if I understand it right, you use 'average' (for `r.resamp.stats`) only when
upscaling [1].

[0] https://trac.osgeo.org/grass/browser/grass-addons/grass7/raster/r.valley.bottom/r.valley.bottom.py#L391
[1] https://trac.osgeo.org/grass/browser/grass-addons/grass7/raster/r.valley.bottom/r.valley.bottom.py#L595.

You use 'bilinear' when downscaling (the slope and the "percentile" that
is), right? [2, 3]. The "percentile" (number of lower elevation points /
total number of points in surrounding region), is part of the algorithm.

[2] https://trac.osgeo.org/grass/browser/grass-addons/grass7/raster/r.valley.bottom/r.valley.bottom.py#L591
[3] https://trac.osgeo.org/grass/browser/grass-addons/grass7/raster/r.valley.bottom/r.valley.bottom.py#L601

Can we use this tool to say "look, our resampled elevation maps are
similar"?  But then, I'd need to run this for each resampled map and
compare the outputs (say MRVBF or/and MRRTF)?

Or is it simpler what you suggest?

Nikos