Performance and Generalizability Impacts of Incorporating Location Encoders into Deep Learning for Dynamic PM2.5 Estimation

TL;DR

This paper systematically evaluates how different methods of incorporating location information affect deep learning models' accuracy and transferability in estimating daily PM2.5 levels across the US, highlighting the benefits of pretrained location encoders.

Contribution

It provides the first comprehensive analysis of location encoders in dynamic environmental prediction, comparing raw coordinates, static, and pretrained encoders for geospatial deep learning.

Findings

Pretrained location encoders improve both accuracy and generalization.

Raw coordinates enhance within-region interpolation but reduce cross-region transferability.

Different encoder types exhibit varying spatial artifact patterns.

Abstract

Deep learning has shown strong performance in geospatial prediction tasks, but the role of geolocation information in improving accuracy and generalizability remains underexamined. Recent work has introduced location encoders that aim to represent spatial context in a transferable way, yet most evaluations have focused on static mapping tasks. Here, we study the effect of incorporating geolocation into deep learning for a dynamic and spatially heterogeneous application: estimating daily surface-level PM2.5 across the contiguous United States using satellite and ground-based observations. We compare three strategies for representing location: excluding geolocation, using raw latitude and longitude, and using pretrained location encoders. We evaluate each under within-region and out-of-region generalization settings. Results show that raw coordinates can improve performance within regions by supporting spatial interpolation, but can reduce generalizability across regions. In contrast, pretrained location encoders such as GeoCLIP improve both predictive accuracy and geographic transfer. However, we also observe spatial artifacts linked to encoder characteristics, and performance varies across encoder types (e.g., SatCLIP vs. GeoCLIP). This work provides the first systematic evaluation of location encoders in a dynamic environmental estimation context and offers guidance for incorporating geolocation into deep learning models for geospatial prediction.