Interactive-Explanatory Geospatial Data Visualization on the Web: A Review of Various Open Source Tools

Interactive web visualizations help facilitate the rapid communication of datasets through their relative ease of accessibility, their ability to engage viewers in data exploration, and their readiness of dissemination. However, climate and weather datasets often share characteristics that present unique challenges to dynamic and performant visualization in modern web browsers. These include substantial file sizes, multidimensional variables with broad geospatial and temporal components, and formats not natively supported in JavaScript, the core language of the web. Further challenges include the considerable storage and computational demands associated with calculating and rendering continuous gridded representations of data from points—such as in web maps—and supporting interactivity by efficiently responding to user queries that index data across different dimensions. Web-based visualization of geospatial data is therefore fragmented by a vast number of diverse web development tools. Major open source libraries include the HoloViz ecosystem for declarative Python visualizations, Leaflet for maps, D3.js for customization power, Plotly for chart generation, and Regl for harnessing WebGL’s advanced graphic rendering capabilities. Additionally, implementation of the popular front-end library React alongside modern data formats such as Zarr and MessagePack enable potential cloud-based approaches supported by highly extensible and reusable code components.

In this study, we explore the application of some of the above tools to develop explanatory interactive visualizations of two datasets—the Community Earth System Model - Large Ensemble Community Project (CESM-LENS2) and machine learning models associated with Warn-on-Forecast (WoFS) tornado prediction. We seek to share insights gained through rapid prototyping on approaches to address computational challenges in reading/rendering data, designing interactive user experiences conducive to the effective scientific communication of community datasets, and ultimately formulating open source web-based visualization workflows that support transparency and accessibility.