Methods to downscale climate timeseries from CORDEX RCM data.
This scheme specifically addresses the need for hillslope scale atmospheric forcing timeseries for modeling the local impact of regional climate change projections on the land surface in complex terrain. The method has a global scope and is able to generate the full suite of model forcing variables required for hydrological and land surface modeling at hourly timesteps.
It achieves this by utilising the previously published TopoSCALE scheme (Fiddes & Gruber, 2014) to generate a synthetic observation of current climate at hillslope scale, while accounting for a broad range of surface-atmosphere interactions. These synthetic observations are then used to debias (downscale) CORDEX climate variables using the quantile mapping method. A further temporal disaggregation step produces sub-daily fields. This approach has the advantages of other empirical-statistical methods, namely speed of use, while avoiding the need for ground data, which is often limited. It is therefore a suitable method for a wide range of remote regions where ground data is absent, incomplete, or not of sufficient length. The approach is evaluated using a network of high elevation stations across the Swiss Alps and a test application of modelling climate change impacts on Alpine snow cover is given.
Please see the following discussion-paper for further details:
Fiddes, J., Aalstad, K., and Lehning, M.: TopoCLIM: Rapid topography-based downscaling of regional climate model output in complex terrain v.1.0, Geosci. model Dev. Discuss. [preprint], https://doi.org/10.5194/gmd-2021-60, in review, 2021.
To begin you'll need a copy of the source code. Either fork the TopoCLIM repository to your own github username, or clone directly.
git clone https://github.com/joelfiddes/topoCLIM.git
cd topoCLIM
It's recommended (but not essential) that you use some sort of python environment manager, such as using the Anaconda distribution and creating an environment (in the code below called "tclim"), or using virtualenv instead. Here we will use Anaconda.
Set up Python 3 env (Python >= 3.6 supported) and install Python packages in requirements.txt:
conda create -n tclim python=3.9
conda activate tclim
(tclim) pip install -r requirements.txt
Install R and R package dependencies:
sudo apt-get r-base
Rscript -e 'install.packages(c("ncdf4","qmap"), repos="http://cran.us.r-project.org")'
Install ESGF client and dependencies for downloading CORDEX data:
conda install -c conda-forge esgf-pyclient
conda install -c conda-forge defusedxml
conda install -c conda-forge myproxyclient
Install Climate Data Operators (CDO) used in postprocessing CORDEX data:
conda install -c conda-forge cdo
A minimal example for a single point (WFJ) and two climate models is provided here including raw cordex and toposcale downscaled data so it can be run straight out of the box:
cd ./topoCLIM/tclim
python tclim_run.py ../examples/ ../examples/tscale ../examples/cordex/
TopoSCALE is a separate python based package available at: https://github.com/joelfiddes/tscaleV2.git
We include a test TopoSCALE timeseries here in "/examples" for running and testing TopoCLIM.
To access CORDEX data on the Earth System Grid Federation nodes credentials are required as described here: https://esgf-data.dkrz.de/user/add/?next=http://esgf-data.dkrz.de/user/add/
CORDEX data is downloaded using the helper script:
python esgf_get.py cordex_domain openid output_dir
And postprocessed:
python esgf_post.py cordex_domain output_dir mydomain output_res
Currently the code requires historical, RCP 2.6 and RCP 8.5 scenarios. esgf_get.py is currently configured to obtain datasets required by TopoCLIM. Future work will focus on making these scenarios configurable.
We use the ini config file format to contain parameters used in a simulation. This promotes reproducability as a config file can be stored with simulation results, permitting re-runs. They can also be committed to a version controlled repository. The following parameters are defined:
startDateHist (str): start common historical period of era5 TopoSCALE datset and CORDEX in format YYYY-MM-DD ("1980-01-01") endDateHist (str): end common historical period of era5 TopoSCALE datset and CORDEX in format YYYY-MM-DD ("2004-12-31") startDateRcp (str): start Date CORDEX RCP data in format YYYY-MM-DD ("2006-01-01") endDateRcp (str): end Date CORDEX RCP data in format YYYY-MM-DD ("2099-12-31") startDateQmap (str): start date quantile mapping period in format YYYY-MM-DD ("1980-01-01") endDateQmap (str): end date quantile mapping period in format YYYY-MM-DD ("2004-12-31")
Now you should be able to run your TopoCLIM jobs:
cd ./topoCLIM/tclim
python tclim_run.py workdir path2tscale path2cordex
Downscaled timeseries for each climate scenario and for each TopoSCALE point will be written to the workdir under a folder structure corresponding to each TopoSCALE point.