Hello, a question about the default path of Solweig_2026a_calc_forprocessing.py (groundScheme = 0, outgoingLW = 0), which seems to differ from 2025a in a way that may or may not be intended.
In 2025a, gvf_2018a receives the raw ground temperature deviation Tg, and TsWaveDelay_2015a is applied afterwards to the returned gvfLup* grids. Inside the GVF (sunonsurface_2018a) the emission is computed as:
Lup = SBC * emis_grid * (Tg * shadow + Ta + 273.15) ** 4 - ...
so it expects Tg to be a deviation from air temperature, with the shadow weighting applied internally.
In 2026a the wave delay is applied before the GVF, to the absolute temperature:
TgTemp = Tgdiff * shadow + Ta
_, timeadd, Tg = TsWaveDelay_2015a(TgTemp, firstdaytime, timeadd, timestepdec, Tg)
and this Tg (absolute, already shadow-weighted) is then passed to the unchanged gvf_2018a. The GVF then appears to add Ta a second time and apply the shadow factor a second time. At Ta = 20 °C that raises the emitting temperature by roughly 20 K, on the order of +100 W/m² in Lup for sunlit ground, relative to 2025a. Since this is the default branch, it affects runs that have not enabled the new ground scheme.
Is this restructuring intended (perhaps as groundwork for the scheme), or should the default path stay equivalent to 2025a? If the latter, restoring the 2025a ordering (pass Tgdiff to gvf_2018a, delay the returned gvfLup* grids) looks sufficient, and a PR is no problem. Apologies in advance if a compensating change elsewhere has been missed here.
Hello, a question about the default path of
Solweig_2026a_calc_forprocessing.py(groundScheme = 0, outgoingLW = 0), which seems to differ from 2025a in a way that may or may not be intended.In 2025a,
gvf_2018areceives the raw ground temperature deviationTg, andTsWaveDelay_2015ais applied afterwards to the returnedgvfLup*grids. Inside the GVF (sunonsurface_2018a) the emission is computed as:so it expects
Tgto be a deviation from air temperature, with the shadow weighting applied internally.In 2026a the wave delay is applied before the GVF, to the absolute temperature:
and this
Tg(absolute, already shadow-weighted) is then passed to the unchangedgvf_2018a. The GVF then appears to add Ta a second time and apply the shadow factor a second time. At Ta = 20 °C that raises the emitting temperature by roughly 20 K, on the order of +100 W/m² in Lup for sunlit ground, relative to 2025a. Since this is the default branch, it affects runs that have not enabled the new ground scheme.Is this restructuring intended (perhaps as groundwork for the scheme), or should the default path stay equivalent to 2025a? If the latter, restoring the 2025a ordering (pass
Tgdifftogvf_2018a, delay the returnedgvfLup*grids) looks sufficient, and a PR is no problem. Apologies in advance if a compensating change elsewhere has been missed here.