Repository for evaluating CCAM runs.
ausclimateservice / ccam-evaluation Goto Github PK
View Code? Open in Web Editor NEWRepository for evaluating CCAM runs.
Repository for evaluating CCAM runs.
NCL script to plot climate extreme indices using output from the Climpact script (See Issue #1 )
plot_climpact_automate.ncl.txt
Uses a Shell script to input command line arguments into the NCL script. For example:
Reads Observed data (AGCD) for re-gridding to the Observed data set grid.
Lines 21/22 need to be updated to point to Climpact output directory.
As per #11 , this task will be to produce the full suite of Climpact indices in Python for intercomparison.
Script to calculate and output climate extreme indices in NetCDF.
Uses R script from Climpact (https://github.com/ARCCSS-extremes/climpact/). Takes ~8 hours to run on CCAM 1980-2019 25km data when calculating all indices.
Requires the following modules on Gadi:
Input variables:
I'll start drafting a master document to establish a development roadmap for the CCAM evaluation tools.
I refer to the regridding standards documented as "Observations and GCMs should be regridded to RCM/CCAM grid using conservative area weighted remapping" on https://github.com/AusClimateService/ccam-evaluation/blob/main/evaluation_standards.md.
We agree that conservative method is suitable for upscaling.
For downscaling, it is less clear, whether NN, bilinear or conservative. The use cases are evaluation of mean states, climate indicators, and added-value analyses. I assume using the implementations given by xesmf.Regridder, https://xesmf.readthedocs.io/en/latest/notebooks/Compare_algorithms.html
Do we also want to worry about any height correction (to temperature for instance) using lapse rates? Which can be messy.
I raise this issue for discussion and reaching agreement.
As per discussions with @tha051 today, we're going to run a parallel evaluation of Climpact and Python for computing indices.
The first port of call is to run the Climpact suite against test data (CCAM?) to produces a directory of netcdf files containing the indices. We will then do the same analysis with Python and compare results.
@ngben, as you are already running Climpact are you able to assist with the R side of this evaluation? I will take care of the Python side of things as well as the inter comparison itself.
NCL script to calculate Added Value, bias (CCAM/RCM minus Obs (AGCD) climatology), and RMSE.
Will need to modify script to calculate realised added value (using future simulations).
Works on precip, tasmin, tasmax (other variables can be added but would need to use a different Obs data set)
Calculates average, variance, 1st and 99th percentile AV. CCAM and GDD are regridded to Obs grid.
Uses a Shell script to input command line arguments into the NCL script. For example:
Some variables are outputting data which may need to be processed before running evaluation scripts.
tasmax/tasmin: Fix time units to start from 1/01/1980 and end 31/12/2014 (As an example for NorESM2-MM output). Current output represents the previous 24 hours (e.g., first timestep is 00:00 2/01/1980 and last timestep is 1/01/2015).
rainfall: Output daily data. Current output is hourly.
The above issues can be fixed using CDO to process the data.
Consistent output directory formatting will also make it easier to automate the evaluation scripts.
We need a master list of the minimum set of metrics to be coded up and automated. This issue can be used to discuss and a "metrics.md" file will be written to track progress.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.