Navigating Spatial and Temporal Modalities for the Development of an AI-Ready Dataset for Urban Heat Research

The urban heat island effect is a well-studied phenomenon that describes how urban areas create microclimates that are warmer than their less developed and greener surroundings. The implications of this effect are broad, given that over half of the world’s population lives in urban areas, including over eighty percent of the United States’ population. These percentages are only expected to increase in future decades. Access to heat-mitigating green spaces has also been identified as an environmental justice issue, deepening the implications of potential harms to society. It is important to understand how we can mitigate urban heat to better serve the increasing number of urban residents. Artificial Intelligence (AI) is a tool that can help with this research topic, given that it can capture nonlinearities and complex relationships between variables, and is also flexible in its ability to leverage datasets of differing modalities (e.g., in-situ point data and satellite imagery). However, a considerable problem hindering research advances in urban heat is the current lack of datasets that are ready to implement in AI frameworks. Such datasets involve complex reformating and processing to be ready to use for AI applications, including reprojecting and spatiotemporal aligning. “AI-ready datasets” focus on making data FAIR (findable, accessible, interoperable, and reusable) to reduce the amount of time spent on data preprocessing and accelerate research progress. Here we will document efforts involving the creation of an urban heat AI-ready dataset, containing remote sensing images and in situ observations from ground-based instruments. We will describe challenges faced and ways to overcome them. Complex relationships between various physical variables included in the dataset will also be visualized using self-organizing maps (SOMs). Additional applications of the developed AI-ready dataset, such as cloud in-painting and super-resolution, will also be discussed.