Bias Correction of CESM Output in a Modern Python Framework

Global climate models (GCMs) contain regional scale biases, primarily due to insufficient spatial resolution and limited representation of some physical processes. It is common, therefore, to correct these biases in GCM output prior to using it as input to regional scale models.

This presentation will highlight recent work to update existing software to create a new version of a global bias-corrected climate dataset that is built from the NCAR Community Earth System Model (CESM) output. The existing legacy software, written in the NCAR Command Language (NCL) and Fortran, uses the bias correction technique described in Bruyère et al. (2015) and was applied to CESM output from simulations that participated in phase 5 of the Coupled Model Intercomparison Experiment (CMIP5). The updated software described in this presentation applies the same bias correction method and is written in a modern and open Python framework. One of the initial objectives is to investigate Python implementations that improve the method’s accessibility and scalability while maintaining code efficiency and accuracy. Using Python frameworks such as Xarray, cf-xarray, NumPy, Dask, xESMF, and GeoCAT, this was achieved within reason in early implementations of the code.

Although the immediate goal is to bias correct the CESM output that participated in CMIP6, our ultimate goal is to generalize the code to be used with any of the CF-formatted GCMs prevalent in the research community. It therefore will serve as an open, accessible, and valuable Python tool for the research community at large.