No Location Left Behind: Measuring and Improving the Fairness of Implicit Representations for Earth Data

TL;DR

This paper introduces FAIR-Earth, a dataset for evaluating fairness in Earth data representations, reveals disparities in current models, and proposes spherical wavelet encodings to improve equitable modeling across diverse regions.

Contribution

The paper presents FAIR-Earth, a novel dataset for fairness assessment in Earth INRs, and introduces spherical wavelet encodings to enhance model robustness and equity.

Findings

Significant performance disparities in existing INRs across subgroups.

High-frequency signals like islands and coastlines are poorly modeled.

Spherical wavelet encodings improve performance consistency across scales.

Abstract

Implicit neural representations (INRs) exhibit growing promise in addressing Earth representation challenges, ranging from emissions monitoring to climate modeling. However, existing methods disproportionately prioritize global average performance, whereas practitioners require fine-grained insights to understand biases and variations in these models. To bridge this gap, we introduce FAIR-Earth: a first-of-its-kind dataset explicitly crafted to examine and challenge inequities in Earth representations. FAIR-Earth comprises various high-resolution Earth signals and uniquely aggregates extensive metadata along stratifications like landmass size and population density to assess the fairness of models. Evaluating state-of-the-art INRs across the various modalities of FAIR-Earth, we uncover striking performance disparities. Certain subgroups, especially those associated with high-frequency signals (e.g., islands, coastlines), are consistently poorly modeled by existing methods. In response, we propose spherical wavelet encodings, building on previous spatial encoding research. Leveraging the multi-resolution capabilities of wavelets, our encodings yield consistent performance over various scales and locations, offering more accurate and robust representations of the biased subgroups. These open-source contributions represent a crucial step towards the equitable assessment and deployment of Earth INRs.