search

metno / emep-ctm

Open Source EMEP/MSC-W model
GNU General Public License v3.0
30 stars 20 forks source link

Land cover problem rv4.3x #84

Closed mvieno closed 3 years ago

mvieno commented 3 years ago

Hi, I am using a specifically created land cover for the UK. When I use this land-cover I get to lines (see C5H8 emissions as an example as directly linked with land cover) at 52N and 56N. I have this issue with rv4.34 and rv4.36.

This problem occurs when I use the new land cover in combination with the global land cover as follow:

from the cnnfig.nml: LandCoverInputs%MapFile = 'DataDir/EU_LCM_2020_08_06.nc', DataDir/glc2000xCLMf18.nc',

Emis_mgm2_BioNatC5H8_UK_fullrun

however, as a workaround if I use one land cover for both calls in the config.nml the issue goes away: LandCoverInputs%MapFile = 'DataDir/EU_LCM_2020_08_06.nc', 'DataDir/EU_LCM_2020_08_06.nc',

Emis_mgm2_BioNatC5H8_UK_fullrun_fix

gitpeterwind commented 3 years ago

Do you know if the "lines" correspond to the glc landuse?

mvieno commented 3 years ago

@gitpeterwind I need to double check this. I need to make a map of the actual land-cover. rather than isoprene emissions. I

gitpeterwind commented 3 years ago

If you only use the GLC landcover, then you could see whether the line is has the same value for isoprene. (then we know it is the mechanism that mixes the landcovers which fails)

mvieno commented 3 years ago

@gitpeterwind, ok I am running one day with the global land in both slots.

LandCoverInputs%MapFile = 'DataDir/glc2000xCLMf18.nc', DataDir/glc2000xCLMf18.nc',

mvieno commented 3 years ago

@gitpeterwind here are the maps.

Emis_mgm2_BioNatC5H8_UK_fullrun

mvieno commented 3 years ago

@gitpeterwind in other EMEP-WRF UK applications (UK 1x1 km2) the lines are also present in the vertical direction.

gitpeterwind commented 3 years ago

Thanks! It seems the line correspond to the GLC values. Then it is most probably the criterion for switching which is too tight. In Landuse_mod.f90 there is a test if(landuse_tot(i,j)< 0.99999 ) then . Try to replace the number with 0.99 instead. It might be that your landuse file has larger numerical inacuracies (sum of fractions is not exactly one), or that the interpolation increases those too much for the very fine scale you are using.

mvieno commented 3 years ago

@gitpeterwind ok I'll check out the land cover, and test your suggestion. Thanks

mvieno commented 3 years ago

Hi. I changed the test from 0.99999 to 0.9999 and it seems to work! The land cover we created should be more precise than this so I do not fully understand it. I checked and it does add up to 0.9999999 but it has some values where the sum does not add up to 1.

Emis_mgm2_BioNatC5H8_UK_fullrun

many thanks!

gitpeterwind commented 3 years ago

Good! The lost precision might be due to interpolation too, although the model calculates with about 16 digits, so I do not fully understand this either. We may loosen this limit a bit and perhaps rather renormalize the values in the next version.