A New Full-stack Web-based Tool to Visualize Weather Data

In the physical sciences, datasets have traditionally been visualized through the creation of static images stored on servers. Scientific visualization libraries, such as Python’s Matplotlib, have been optimized to generate refined and highly-customizable plots and maps that aid in research. Using these static images to develop web-based visualizations, however, is not ideal, as large volumes of images require lots of storage space, many of these images will typically be infrequently viewed, and interfaces designed around these images are inherently less flexible.

Recent advances in browser-based visualization tools, such as JavaScript’s D3 (Data Driven Documents) library, have paved the way for front-end (client-side) development of interactive scientific visualization products. On the back-end (server-side), database software like PostgreSQL can enable user interfaces (UIs) to achieve a higher level of user-defined control and flexibility. Application program interface (API) routines link the front and back ends of a browser-based product, and can be written in such languages as Python or PHP.

A full-stack browser-based tool to visualize gridded reanalysis, reforecast, and model data has been developed at the University at Albany through collaboration between the Department of Atmospheric and Environmental Sciences (DAES) and the Albany Visualization and Informatics Laboratories (AVAIL; http://availabs.org). The D3 JavaScript Library and HTML5 canvases, along with the ReactJS UI library, comprise the front-end of the product. PostgreSQL and Python were selected to build the back-end and product APIs, respectively.

The attached image shows the base product at an early stage of development. In addition to the gridded data display, a dashboard of relevant statistical analyses will be made available to all users. Product source code will be hosted on GitHub to encourage contributions and adaptations from the atmospheric and climate science communities.