1 Installation
See setup instructions for detailed installation options, including via the python
package pdm
, conda
and docker
.
If the python
clim-recal
package is installed, some of the components can be used via the command line. For details run:
$ clim-recal --help
2 Data Preparation Pipeline
The python
clim-recal
package supports two main functions:
- Downloads the UK Met Office
observation
andprojection
data - Converts that data to aligned spatial time series grids for debiasing
We currently provide the the following converted and aligned files in our Download Datasets page:
- Maximum temparture:
tasmax
- Maximum temparture:
tasmin
- Rainfall:
rainfall
/pr
We provide 5 projection runs:
01
05
06
07
08
We demonstrate the advantage of focusing on 05
through 08
in Identifying Runs.
Below is how these files are produced.
2.1 Download UK Met Office data
We use two datasets maintained by the Met Office
This data can be quite large and take significant amounts of time to download.
To download this an account is needed. The registration process can be found here: https://services.ceda.ac.uk/cedasite/register/info/.
We hope to incorporate this module in the clim-recal
command line interface in future, but at present the ceda_ftp_download.py
module is used to download this data and can be run from the command line via python3
:
$ cd clim-recal/python/clim_recal
$ ./ceda_ftp_download.py --help
usage: ceda_ftp_download.py [-h] --input INPUT [--output OUTPUT] --username
USERNAME --psw PSW [--reverse] [--shuffle]
[--change_hierarchy]
options:
-h, --help show this help message and exit
--input INPUT Path where the CEDA data to download is located. This can be
a path with or without subdirectories. Set to
`/badc/ukcp18/data/land-cpm/uk/2.2km/rcp85/` to download all
the raw UKCP2.2 climate projection data used in clim-recal.
--output OUTPUT Path to save the downloaded data
--username USERNAME Username to connect to the CEDA servers
--psw PSW FTP password to authenticate to the CEDA servers
--reverse Run download in reverse (useful to run downloads in
parallel)
--shuffle Run download in shuffle mode (useful to run downloads in
parallel)
--change_hierarchy Change the output sub-directories' hierarchy to fit the
Turing Azure fileshare hierarchy (only applicable to UKCP
climate projection data, i.e. when --input is set to
2.2 Aligning HADs and CPM UK data
These dataset are provided in the following forms
Dataset | Resolution | Coordinates | Calendar |
---|---|---|---|
UKHAD | 1 km | British National Grid1 | Standard |
RCP8.5 | 2.2 km | Rotated Pole2 | 360 day |
To align them we:
- Interpolate RCP8.5 to standard Gregorian calendar via the
nearest
method3. - Reproject RCP8.5 to British National Grid Coordinate structure.
- Resample UKHAD to a 2.2 km grid.
- Reproject UKHAD bounds to align with RCP8.5
Links to these files are in the Download Datasets section.
2.3 Running debiasing methods
Once the data is aligned, it is ready for you to run your choice of debiasing methods on it.
Footnotes
British National Grid spec: https://epsg.io/27700↩︎
Rotated pole of latitude
37.5
and longitude177.5
. See UKCP Guidance: Data availability, access and formats appendix B.↩︎See the CPM Projection section for more details, including a comparison with a linear interpolation.↩︎