A Python package to perform climate downscaling at the hillslope scale
Project description
TopoPyScale
Python version of Toposcale packaged as a Pypi library. Toposcale is an original idea of Joel Fiddes to perform topography-based downscaling of climate data to the hillslope scale.
Documentation avalaible: https://topopyscale.readthedocs.io
References:
- Fiddes, J. and Gruber, S.: TopoSCALE v.1.0: downscaling gridded climate data in complex terrain, Geosci. Model Dev., 7, 387–405, https://doi.org/10.5194/gmd-7-387-2014, 2014.
- Fiddes, J. and Gruber, S.: TopoSUB: a tool for efficient large area numerical modelling in complex topography at sub-grid scales, Geosci. Model Dev., 5, 1245–1257, https://doi.org/10.5194/gmd-5-1245-2012, 2012.
Kristoffer Aalstad has a Matlab implementation: https://github.com/krisaalstad/TopoLAB
Contribution Workflow
Please follow these simple rules:
- a bug -> fix it!
- an idea or a bug you cannot fix? -> create a new issue if none doesn't already exist. If one exist, then add material to tit.
- wanna develop a new feature/idea? -> create a new branch. Do the development. Merge with main branch when accomplished.
- Create release version when significant improvements and bug fixes have been done. Coordinate with others
Create a new release: Follow procedure and conventions described in: https://www.youtube.com/watch?v=Ob9llA_QhQY
And check out our Slack: tscaleworkspace.slack.com
Contributors to the current version (2021) are:
- Simon Filhol
- Joel Fiddes
- Kristoffer Aalstad
Design
- Inputs
- Climate data from reanalysis (ERA5, etc)
- Climate data from future projections (CORDEX) (not avail.)
- DEM from local source, or fetch from public repository: SRTM, ArcticDEM, ASTER
- Run TopoScale
- compute derived values (from DEM)
- toposcale (k-mean clustering)
- interpolation (bilinear, inverse square dist.)
- Output
- Cryogrid format
- FSM format
- CROCUS format
- Snowmodel format
- basic netcfd
- For each method, have the choice to output either the abstract cluster points, or the gridded product after interpolation
- Validation toolset
- validation to local observation timeseries
- plotting
- Gap filling algorithm
- random forest temporal gap filling
Validation (4) and Gap filling (4) are future implementation.
Installation
conda create -n downscaling python=3.8 ipython
conda activate downscaling
# Recomended way to install dependencies:
conda install -c conda-forge xarray matplotlib scikit-learn pandas numpy netcdf4 h5netcdf rasterio pyproj dask
# OPTION 1 (Pypi release):
pip install TopoPyScale
# OPTION 2 (development):
cd github # navigate to where you want to clone TopoPyScale
git clone git@github.com:ArcticSnow/TopoPyScale.git
pip install -e TopoPyScale #install a development version
#----------------------------------------------------------
# OPTIONAL: if using jupyter lab
# add this new Python kernel to your jupyter lab PATH
python -m ipykernel install --user --name downscaling
# Tool for generating documentation from code docstring
pip install lazydocs
Then you need to setup your cdsapi
with the Copernicus API key system. Follow this tutorial after creating an account with Copernicus. On Linux, create a file nano ~/.cdsapirc
with inside:
url: https://cds.climate.copernicus.eu/api/v2
key: {uid}:{api-key}
Basic usage
- Setup your Python environment
- Create your project directory
- Configure the file
config.ini
to fit your problem (seeconfig.yml
for an example) - Run TopoPyScale
import pandas as pd
from TopoPyScale import topoclass as tc
from matplotlib import pyplot as plt
# ========= STEP 1 ==========
# Load Configuration
config_file = './config.yml'
mp = tc.Topoclass(config_file)
# Compute parameters of the DEM (slope, aspect, sky view factor)
mp.compute_dem_param()
# ========== STEP 2 ===========
# Extract DEM parameters for points of interest (centroids or physical points)
mp.extract_topo_param()
# ----- Option 1:
# Compute clustering of the input DEM and extract cluster centroids
#mp.extract_dem_cluster_param()
# plot clusters
#mp.toposub.plot_clusters_map()
# plot sky view factor
#mp.toposub.plot_clusters_map(var='svf', cmap=plt.cm.viridis)
# ------ Option 2:
# inidicate in the config file the .csv file containing a list of point coordinates (!!! must same coordinate system as DEM !!!)
#mp.extract_pts_param(method='linear',index_col=0)
# ========= STEP 3 ==========
# compute solar geometry and horizon angles
mp.compute_solar_geometry()
mp.compute_horizon()
# ========= STEP 4 ==========
# Perform the downscaling
mp.downscale_climate()
# ========= STEP 5 ==========
# explore the downscaled dataset. For instance the temperature difference between each point and the first one
(mp.downscaled_pts.t-mp.downscaled_pts.t.isel(point_id=0)).plot()
plt.show()
# ========= STEP 6 ==========
# Export output to desired format
mp.to_netcdf()
TopoClass will create a file structure in the project folder (see below). TopoPyScale assumes you have a DEM in GeoTiFF, and a set of climate data in netcdf (following ERA5 variable conventions).
TopoPyScale can easier segment the DEM using clustering (e.g. K-mean), or a list of predefined point coordinates in pts_list.csv
can be provided. Make sure all parameters in config.ini
are correct.
my_project/
├── inputs/
├── dem/
├── my_dem.tif
└── pts_list.csv (optional)
└── climate/
├── PLEV*.nc
└── SURF*.nc
├── outputs/
└── config.ini
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for topopyscale-0.2.1-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | bcc298aa9af7cb09bf2cc70cd050b1ecb761352ea0ef6ce312c4d2b059c478a0 |
|
MD5 | 80d5c6e0943e2df2df100b91b1af6dda |
|
BLAKE2b-256 | d6df9c574f83bd01b48c4415553f5df5f7029a40209f4aacb3451c6c1505e036 |