TopoSCALE-CCI

TopoSCALE downscaling for the ESA CCI global permafrost model

Setup

git clone https://github.com/joelfiddes/TopoSCALE-CCI.git
pip install -r requirements.txt

Environment setup on SAGA

https://stackoverflow.com/questions/41274007/anaconda-export-environment-file

Maintainer steps

Conda create

conda create -n tscale python=3.7

Export env

cd TopoSCALE-CCI
conda env export  > environment.yml

User steps

User runs env setup:

cd TopoSCALE-CCI
conda env create -f environment.yml

Conda activate

conda activate tscale

ERA5 retrieval

Retrieval of all CCI parameters as single parameter global annual files. Requires CDS API is set up on server (done for SAGA): https://cds.climate.copernicus.eu/api-how-to

Download parameters are configured (in practice just year) within script era5_request_CCI.py:

startYear = 2019
endYear = 2020
era5dir= "/cluster/projects/nn9606k/era5/"
timestep = "6"
plevels = ["1000", "700", "500", "200"]

and then run:

python era5_request_CCI.py

ERA5 input files

Files produced by retrieval with naming convention:

PLEV_ERA5PARNAME_YEAR.nc
SURF_ERA5PARNAME_YEAR.nc

PLEV_geopotential_1980.nc SURF_2m_dewpoint_temperature_1980.nc SURF_instantaneous_surface_sensible_heat_flux_1980.nc PLEV_relative_humidity_1980.nc SURF_2m_temperature_1980.nc SURF_surface_solar_radiation_downwards_1980.nc PLEV_temperature_1980.nc SURF_friction_velocity_1980.nc SURF_surface_thermal_radiation_downwards_1980.nc PLEV_u_component_of_wind_1980.nc SURF_geopotential_1980.nc SURF_TOA_incident_solar_radiation_1980.nc PLEV_v_component_of_wind_1980.nc SURF_instantaneous_moisture_flux_1980.nc SURF_Total_precipitation_1980.nc

ERA5 Input NetCDF format

Standard ERA5 NetCDF format

[fiddes@login-3.SAGA ~/nn9606k/era5]$ ncdump -h SURF_2m_temperature_1980.nc 
netcdf SURF_2m_temperature_1980 {
dimensions:
	longitude = 1440 ;
	latitude = 721 ;
	time = 1464 ;
variables:
	float longitude(longitude) ;
		longitude:units = "degrees_east" ;
		longitude:long_name = "longitude" ;
	float latitude(latitude) ;
		latitude:units = "degrees_north" ;
		latitude:long_name = "latitude" ;
	int time(time) ;
		time:units = "hours since 1900-01-01 00:00:00.0" ;
		time:long_name = "time" ;
		time:calendar = "gregorian" ;
	short t2m(time, latitude, longitude) ;
		t2m:scale_factor = 0.0019369995756104 ;
		t2m:add_offset = 260.62106366574 ;
		t2m:_FillValue = -32767s ;
		t2m:missing_value = -32767s ;
		t2m:units = "K" ;
		t2m:long_name = "2 metre temperature" ;

// global attributes:
		:Conventions = "CF-1.6" ;
		:history = "2019-11-15 02:52:33 GMT by grib_to_netcdf-2.14.0: /opt/ecmwf/eccodes/bin/grib_to_netcdf -o /cache/data0/adaptor.mars.internal-1573786171.4555001-10939-16-686e4b7d-fce8-4399-a243-f8135543b018.nc /cache/tmp/686e4b7d-fce8-4399-a243-f8135543b018-adaptor.mars.internal-1573786171.4562213-10939-8-tmp.grib" ;
}

TopoSCALE CCI

Adapted from TopoSCALE3D a three dimensional downscaling of ERA5 forcing files to a set of points defined in a coordinates text file. In CCI the target resolution is 1km so we simplify the radiation algorithms as slope, aspect and sky view factor are not available.

Run example:

python tscale_cci_run.py "/home/joel/sim/cci_perm_final/sebs_coords.txt"  "/home/joel/sim/cci_perm_final/era5" "/home/joel/sim/cci_perm_final/era5/out" 1980 1980
python tscale_cci_run.py  "/cluster/home/fiddes/sebs_points/coords.txt" "/cluster/home/fiddes/nn9606k/era5" "/cluster/home/fiddes/sebs_points/" 1980 2018
 (tscale) [fiddes@c5-57.SAGA ~/nn9606k/TopoSCALE-CCI]$ python tscale_cci_run.py  ../tscalccitest/coords_seb.txt ../era5 ../tscalccitest/ 1980 1981

Coordinates file format

TopoSCALE-CCI only requires a comma separated ascii file with 3 columns "latitude [decimal degrees, -90->90], longitude[decimal degrees, -180->180], elevation (m)" in that order (nb: no header) e.g.:

69.4049, -139.0497,2000

Output NetCDF format

The output is a set of single parameter 9 day NetCDF with the following naming convention:

PARAMETER_TIMEPERIOD_YEAR.nc

eg.

lwin_1_1980.nc prate_1_1980.nc shum_1_1980.nc swin_1_1980.nc t_1_1980.nc ws_1_1980.nc lwin_2_1980.nc prate_2_1980.nc shum_2_1980.nc swin_2_1980.nc t_2_1980.nc ws_2_1980.nc

NetCDF dimensions

long amd lat are named as dimensions but do not exist.
Each point is indexed as 1:Npoints corresponding to order in coordinates file
time is in hours since start of period
time step is 6h

    ncdump -c swin_1_1980.nc 
    netcdf swin_1_1980 {
    dimensions:
        lon = 27 ;
        lat = 27 ;
        time = 36 ;
    variables:
        float lon(lon) ;
        float lat(lat) ;
        int time(time) ;
            time:units = "hours since 1980-01-01 00:00:00" ;
        float swin(lon, time) ;

    // global attributes:
            :history = "Created by toposcale on " ;
    data:

     lon = _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, 
        _, _, _, _ ;

     lat = _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, _, 
        _, _, _, _ ;

     time = 0, 6, 12, 18, 24, 30, 36, 42, 48, 54, 60, 66, 72, 78, 84, 90, 96, 
        102, 108, 114, 120, 126, 132, 138, 144, 150, 156, 162, 168, 174, 180, 
        186, 192, 198, 204, 210 ;
    }

joelfiddes/TopoSCALE-CCI