Paleoclimate Visualizer
AWS Cloudflare FastAPI Python Svelte TypeScript TailwindCSS
About
This is the project I developed while working as an Undergraduate Research assistant for the University of Washington's School of Oceanography during the Summer of 2024. The goal was to create a web application for visualizing paleocliamte data reconstructions. I worked closely with my advisor, Gemma O'Connor, to create an app capable of telling the full, accurate story of the data. I decided to use Svelte for the front end because of its reputation of high performance, and the Python library FastAPI for the data API because of the vast data science tools that already exists in the Python ecosystem. This combination allowed for a fast user experience while delivering full visualizations. The app uses the Highcharts library to display data, as we found it provides the cleanest for plotting gridded data that needs to be interpolated across a surface. Even though my position has officially ended, I still work to maintain the open source project, ensuring fixes and additions are made when needed.
Research Abstract
Paleoclimate reconstructions are important for placing recent observed climatic changes within the context of long-term variability. Because the instrumental climate record is short – particularly in remote data-sparse regions of the world such as Antarctica – various gridded products that combine proxy observations (e.g., ice core records) with climate model simulations have been developed to extend the instrumental record (Hakim et al, 2017; O’Connor et al. 2021; Dalaiden et al., 2021). Using the reconstructions developed by O’Connor et al. 2021, we developed a web application for viewing, analyzing, and downloading the data. The web app has a Python-based backend application programming interface ( API) hosted on a cloud-based server. We find that this framework allows the users to specify a wide range of inputs, such as data source, climate variables, map choices, time series locations, trends, and statistics, with a rapid response time. We custom built the API to allow users to query data without downloading the entire dataset, which reduces request payload size and bandwidth usage. We built the front-end using Svelte, Tailwind CSS, and HighCharts frameworks, which enabled us to efficiently design a web application that displays interactive maps with adjacent time series plots corresponding to user-specified inputs from the back-end API. The code used to build the web app is open-source and adaptable for including additional gridded climate datasets. The web app is available at pv.jortuck.com.