possible bug in ground layer initialization
in /src/model/grille.f
(1) the first loop which sets layers conductivity/capacity sets also NMAX, which I assume is intended to be the lowest layer to be considered, but actually it is set to the highest layer with thickness>0.4m
it seems the the code assumes LAYER array has layer depths, but actually it has layer thicknesses
(2) the same problem occurs later in the loops where CS/KS are interpolated to Ko/C - there is check:
IF ( dpTemperatureDepth(j) .ge. ZONES(INTER) ) THEN
which compares depth to layer thickness instead of layer depth .. at least in some layer configurations this results in KS(INTER+1) and CS(INTER+1) being read from uninitialized array elements.
I've attached the station configuration which exposes the problem.site.xml
I may be completely wrong here, though ..