[flake] *** ERROR *** : DEPTH_LK >= 0.0 is not consistent with FR_LAKE < 0.5 ***

[jcanton]

There is a bug in flake in extpar 5.10 where the consistency check for DEPTH_LK and FR_LAKE somehow passes but then int2lm crashes with the error in the title. This worked fine with extpar 5.9
The error only "appears" if the lake model is turned on in int2lm (with llake=.true.).
I have encountered it with
raw_data_flake_filename=GLDB_lakedepth.nc

Compile-Date ......: 2023-02-21 18:12:58
Code is modified ..: modified
Library name ......: extpar-v5.10.1
Compiled on .......: daint
Revision number ...: 6cd9e5d
Checkout-Date .....: 2023-02-21 16:14:37
Binary name .......: ./extpar_consistency_check.exe
Compiled by .......: jcanton
Compiler version ..: GCC 9.3.0
Current start time : 2023-02-23 11:04

you can see in the two figures comparing 5.9 (left) to 5.10 (right) that 5.10 has a very different DEPTH_LK map despite having the same FR_LAKE

[jcanton]

more info:
it doesn't have to do with 5.9 vs 5.10 (I'm an idiot) but with tile_mode=1 which by accident I had in the extpar_consistency_check_io namelist I was using with 5.10...

I guess the answer is just that one should not use tile_mode=1 with a COSMO grid, but it might be worth adding a check and throwing an error if this happens?
or we decide I'm just an idiot and I should close this? 😬

[jcanton]

might actually be worth to look at this: the if tile_mode == 1 check is inside of a select case igrid_cosmo

1687 CASE(igrid_cosmo) ! COSMO grid
.....
1759 ! check consistency for "lake depth"
1760 IF (tile_mode == 1) THEN ! subgrid lakes for ICON