Interactive Visualization of the CESM-LENS2 Climate Dataset – Lessons Learned and Recommendations for Visualizing Gridded Datasets using Open Science Tools

As climate models are growing to be ever more complex, so are their resulting data outputs. It can be challenging to not only analyze these datasets to find key patterns and trends, but also to communicate the results clearly and intuitively to the general public through data visualizations. Adding interactivity to data visualizations can bridge this gap by allowing the users to analyze different visualizations of the same dataset by turning a few knobs, which can empower the users to understand these datasets more intuitively. In this study, we compared two methods of creating interactive dashboards for the Community Earth System Model (CESM) - Large Ensemble Community Project (LENS2) climate dataset to assess their suitability from the perspective of Earth System scientists. The CESM-LENS2 dataset is a global-scale ensemble simulation of various Earth system processes at a spatial resolution of 1°x1° for the period of 1850-2100 under CMIP6 historical and SSP370 future radiative forcing scenarios. One of the dashboards is built using open-source Python packages in the PyData and HoloViz ecosystems, which enables users to rapidly create interactive visualizations using tools that are already familiar to climate scientists. The other dashboard uses a custom client-server model using FastAPI and D3.js, which allows users to build intricate, performant and professional dashboards. This study sheds light on two techniques for creating interactive dashboards from the perspective of scientists, which can also be generalized for any spatio-temporal gridded dataset. We also provide recommendations on their relative suitability of use in the various stages of scientific research, from exploration to communication.