FORCE_SW_MODE Issue
I compared the meteo input given as input, the processed meteo outputted by snowpack (i.e. the meteoio processed values) and the meteo outputed by snowpack (the met file that should be the values actually used for the computation).
I did not find any difference between the input meteo and the mio processed meteo (except what we expect for the used filters).
Regarding the outputted ISWR in the met file, I found a significant difference with respect to the input meteo data.
You have two plots in attachment, one for Bern with my data, one for Grossalp with Nander's data. The only difference is Nander's data contain TSS, which Bern data does not.
One the first page of these pdf, you can see the difference in ISWR over few years simulation period and the mean of this difference. We observe that in Grossalp the is no problem during winter.
In the next plots you see the two times series for some particular weeks and months of the year 2011, we see that :
- The met ISWR suddenly jumps from measurements to a nice sinus curve, explaining the added ISWR
- The fact that the is no issue in Grossalp during winter suggests that when there is snow there is no issue. However, in the week/month plot I also plotted the snow height (scaled to fit) and we see that this is not always the case. The problem might occur when there is snow (see Bern 20-21 jan) and inversely. Maybe it is related to the jumps in boundary conditions.
I made also several tests with/without canopy, the effect is still here. The bucket/richards choice makes also, as expected no, difference.
I made tests with alpine 3D output, and it looks like there is no issue in A3D.
Michi had a look to the source code, and one guess is that it maybe comes from the SurfacesFluxes Sdata object used in AsciiIo::writeTimeSeries method, so it is perhaps only an output issue (but in this case why we do not have this issue in A3D? doesn't' it use snowpack io module to write met files?)
Finally, note that I did not get any warning during the runs and I used LAPACK on MacOS.
Attachments: