A Cloud-Based Python Toolkit for Streamlining Polar Climate Model Assessments

Assessing climate model output and reanalyses against observations (e.g. satellite-derived data products) requires considerable data wrangling effort that often results in disjointed or repeated code development by individual researchers or groups. Here, we present a generalizable python toolkit designed to streamline model validation of polar sea surface and land ice variables from reanalyses (MERRA/MERRA-2/ERA5) and climate model (CESM) output against satellite observations. In order to generate a simple and computationally efficient cloud-based workflow, we translate raw data for a variety of different data products into cloud-optimized formats (e.g. zarr), and develop interactive and easily mutable jupyter notebooks that build on a set of common functions for reading, regridding, and analyzing the data. We present model assessment using a novel scorecard system developed through NASAs PolarMERRA project, a joint effort between NASA Cryospheric Sciences and Modeling and Prediction programs, to implement a simple graphical representation of model/reanalysis performance. As an initial demonstration, we provide assessments of surface conditions from several reanalyses against satellite-derived estimates of surface albedo (CLARA-SAL) and skin/surface temperature (AIRS) over sea ice and ice sheets. Our approach provides an example of how a cloud-based workflow can enable more efficient and transparent climate model assessments.