w5e5_downscale / bias-correction / regridding for ISIMIP3 HighRes Experiments

Author: Johanna Malle (mailto:johanna.malle@slf.ch)

Main outputs

• downscaled/bias corrected daily mean Near-Surface Wind Speed @ target domain (stored as .nc-file)
• downscaled/bias-corrected daily mean Near-Surface Relative Humidity @ target domain (stored as .nc-file)
• downscaled/bias-corrected daily mean Surface Downwelling Longwave Radiation @ target domain (stored as .nc-file)
• bias-corrected daily mean Surface Air Pressure @ target domain (stored as .nc-file)

Requirements

Before running the script you need to download the required input data files:

Near-Surface Wind Speed from Global Wind Atlas

wget https://globalwindatlas.info/api/gis/global/wind-speed/10/

CHELSA-BIOCLIM+ monthly mean Near-Surface Relative Humidity at 30"

https://doi.org/10.16904/envidat.332

https://envicloud.wsl.ch/#/?prefix=chelsa%2Fchelsa_V2%2FGLOBAL%2Fmonthly%2F

W5E5 daily mean Near-Surface Wind Speed at 0.5deg

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/sfcWind_W5E5v2.0_19790101-19801231.nc

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/sfcWind_W5E5v2.0_19810101-19901231.nc

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/sfcWind_W5E5v2.0_19910101-20001231.nc

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/sfcWind_W5E5v2.0_20010101-20101231.nc

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/sfcWind_W5E5v2.0_20110101-20191231.nc

W5E5 daily mean Near-Surface Relative Humidity at 0.5deg

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/hurs_W5E5v2.0_19790101-19801231.nc

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/hurs_W5E5v2.0_19810101-19901231.nc

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/hurs_W5E5v2.0_19910101-20001231.nc

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/hurs_W5E5v2.0_20010101-20101231.nc

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/hurs_W5E5v2.0_20110101-20191231.nc

W5E5 daily mean Surface Air Pressure at 0.5deg

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/psl_W5E5v2.0_19790101-19801231.nc

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/psl_W5E5v2.0_19810101-19901231.nc

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/psl_W5E5v2.0_19910101-20001231.nc

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/psl_W5E5v2.0_20010101-20101231.nc

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/psl_W5E5v2.0_20110101-20191231.nc

W5E5 daily mean Surface Downwelling Longwave Radiation at 0.5deg

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/rlds_W5E5v2.0_19790101-19801231.nc

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/rlds_W5E5v2.0_19810101-19901231.nc

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/rlds_W5E5v2.0_19910101-20001231.nc

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/rlds_W5E5v2.0_20010101-20101231.nc

wget https://files.isimip.org/ISIMIP3a/SecondaryInputData/climate/atmosphere/obsclim/global/daily/historical/W5E5v2.0/rlds_W5E5v2.0_20110101-20191231.nc

Installation and environment set-up

Python 3.7 is required.

Since we use esmf for regridding this setup only works on linux OS. Tested on: Ubuntu 18.04.

Environment set-up

With Conda package manager (recommended)

Create the environment and install all required packages

conda env create

Activate the environment

Windows: activate W5E5_regrid_env

Linux: source activate W5E5_regrid_env

Getting Started

Update the file application_example with your file paths etc. and then just run: ./application_example.sh

Note that the longwave calculation relies on relative humidity... hence this needs to be run first!

Main processing steps

Daily mean Near-Surface Wind Speed

• To include orographic effects into daily mean near-surface wind speed (sfcwind) we follow the approach of Brun et al. 2022, and use an aggregation of the Global Wind Atlas 3.0 data (https://globalwindatlas.info) in combination with daily 0.5° sfcwind from W5E5.

• We first regrid both the Global Wind Atlas data and the W5E5 sfcwind data to the target grid/extent using bilinear interpolation. For all regridding xesmf is used (https://xesmf.readthedocs.io/en/latest/).

• The Global Wind Atlas data product represents the period 2008-2017, we therefore average daily regridded W5E5 sfcwind data over this time period. We assume surface wind follows a Weibull distribution and log-transform both datasets before computing the difference layer. We add this difference layer to each log-transformed daily W5E5 raster, and back-transform the sum to obtain the final daily mean near-surface wind speed raster.

$$ sfcwind_{dly} = exp^{(log(sfcwind_{dly}^{W5E5} + c)+ Δsfcwind_{cli})} - c $$

$$ Δsfcwind_{cli} = log(sfcwind_{cli}^{GWA} + c) - log(sfcwind_{cli}^{W5E5} + c) $$

Daily mean Near-Surface Relative Humidity

• The provided downscaling algorithm combines monthly 30’’ CHELSA-BIOCLIM+ data (https://doi.org/10.16904/envidat.332 , Brun et al. 2022) with daily W5E5 data
• In a first step we regrid daily 0.5° W5E5 hurs to the target grid (30”, 1.5’ or 5’) and extent using bilinear interpolation. We assume relative humidity follows a beta-distribution and logit-transform both regridded monthly-averaged W5E5 and monthly CHELSA-BIOCLIM+ relative humidity data. The difference layer is then added to daily regridded and logit-transformed W5E5 hurs of the respective month, and the final raster is obtained by back-transforming the sum.

$$ hurs_{dly} = {1 \over (1+exp^{-h})} $$

$$ h = log({hurs_{dly}^{W5E5} \over {1 - hurs_{dly}^{W5E5}}}) + Δhurs_{mon}$$

$$ Δhurs_{mon} = log({hurs_{mon}^{CHELSA} \over {1 - hurs_{mon}^{CHELSA}}}) - log({{hurs_{mon}^{W5E5}} \over {1- hurs_{mon}^{W5E5}}}) $$

Daily mean Surface Air Pressure

• uses W5E5 daily mean sea-level pressure and a DEM to calculate surface air pressure via the barometric formula:

$$ps_{dly} = {psl_{dly}^{W5E5}exp^{-(g * orog * M) / (T_{0} * R)}} $$

with $ps_{dly}$ being the regridded 0.5° W5E5 daily mean sea-level pressure (bilinear interpolation), $g$ the gravitational acceleration constant (9.80665 m/s2), $orog$ the height at which air pressure is calculated (CHELSA-W5E5 orog, m), $M$ the molar mass of dry air (0.02896968 kg/mol), $R$ the universal gas constant (8.314462618 J/(mol K)) and $T_{0}$ the sea level standard temperature (288.16 K).

Surface Downwelling Longwave Radiation

• follows Fiddes and Gruber (2014, https://doi.org/10.5194/gmd-7-387-2014) and Konzelmann et al. 1994 (https://doi.org/10.1016/0921-8181(94)90013-2)
• assume cloud emissivity is the same at coarse and fine scale, but account for reduction in clear-sky emissivity with elevation
• uses fine-scale temperature (CHELSA) and RH data (hence the RH algorithm above must be run first)