## Model initialization from ERA-Interim data: what happens with soil temperature?

Dear colleges,

I use COSMO-CLM model for seasonal simulations, driven by ERA-Interim data (so yinput_model='CM’). I was looking how model simulates soil temperatures (in comparison with observation data at different depths) and found very strange problem: initial soil temperature profiles looks completely different from soil temperature profiles in reanalysis data (both from T_SO in caf-files, both from the data, taken directly from ERA-Interim website). It seems that soil in model is initialized without usage of soil temperature from reanalysis.

When I’ve increased idbg_level, I get this messages:

variable 33 T_SO not needed for boundary fields 70
variable 34 W_SO not needed for boundary fields 70

So, am I right that soil temperature from reanalysis is not used for model initialization? If I am right, what should I do to use it? Maybe have I’ve forget some switches?

Mikhail

### Replies (3)

#### RE: Model initialization from ERA-Interim data: what happens with soil temperature? - Added by Burkhardt Rockelabout 1 year ago

You are right. Presently only the initial soil water is taken from the coarse grid model in climate mode (yinput_model='CM’ and itype_w_so_rel > 0) . The initial soil temperature is calculated as an exponential function between the T_S from the coarse grid model and the climatological T_CL from the external data set. The T_SO in the reanalyses data is presently not used.

#### RE: Model initialization from ERA-Interim data: what happens with soil temperature? - Added by Mikhail Varentsovabout 1 year ago

Dear Burkhardt, thank you for the answer!

1) In this case, are there any possibility to use T_SO from reanalysis? Or it will be simpler to correct corresponding LAF-file before running CCLM?
2) If soil water content is taken from reanalysis, why I have message that "variable 34 W_SO not needed for boundary fields"?

#### RE: Model initialization from ERA-Interim data: what happens with soil temperature? - Added by Burkhardt Rockelabout 1 year ago

1) The best way would be to introduce an interpolation in the subroutine init_multi_layer_cm which is located in the module file src_2d_fields.f90. The line

            t_so_lm(i,j,kso) = t_cl_lm(i,j) + (t_s_lm(i,j) - t_cl_lm(i,j)) * EXP(-zmls(kso)/2.8)


calculates T_SO by the exponential function I mentioned in my previous reply. At this place one may put the interpolation of T_SO of the reanalysis data. If you decide to do this it would be greatly appreciated. In case you succeed, please provide your changes to the CLM-Community. Please take into account that the lowest model level has the fixed value T_CL.

2) T_SO and W_SO are only used from reanalysis for initialization, during the further simulation they develop freely.

(1-3/3)