# Probing Geospatial SSL Representations with Environmental Signals

arXiv: https://arxiv.org/abs/2607.05207 · HTML: https://mlsanity.com/paper/2607.05207/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 8 of its 13 references that are in the map.
- First posted on arXiv 6 Jul 2026.

## Builds on (most cited first, 8 of 8)

- [Event-based Vision: A Survey](https://mlsanity.com/paper/1904.08405/) (2019)
- [Event-Based Motion Segmentation by Motion Compensation](https://mlsanity.com/paper/1904.01293/) (2019)
- [Unsupervised Learning of a Hierarchical Spiking Neural Network for Optical Flow Estimation: From Events to Global Motion Perception](https://mlsanity.com/paper/1807.10936/) (2018)
- [Spiking Optical Flow for Event-based Sensors Using IBM's TrueNorth Neurosynaptic System](https://mlsanity.com/paper/1710.09820/) (2017)
- [Secrets of Edge-Informed Contrast Maximization for Event-Based Vision](https://mlsanity.com/paper/2409.14611/) (2024)
- [Modelling Drosophila Motion Vision Pathways for Decoding the Direction of Translating Objects Against Cluttered Moving Backgrounds](https://mlsanity.com/paper/2007.00886/) (2020)
- [A Bio-Inspired Research Paradigm of Collision Perception Neurons Enabling Neuro-Robotic Integration: The LGMD Case](https://mlsanity.com/paper/2501.02982/) (2025)
- [Emergence of robust looming selectivity via coordinated inhibitory neural computations](https://mlsanity.com/paper/2510.00498/) (2025)

## Abstract (arXiv)

Self-supervised learning (SSL) is designed to learn generic, transferable representations rather than representations optimized for a single task. Most geospatial benchmarks evaluate representations solely through downstream tasks, providing limited insight into the information encoded within the representation itself. We ask a different question: do SSL representations of satellite imagery preserve statistical associations with environmental variables that co-vary with the imaging process? To answer this question, we probe SSL representations using co-located ERA5 reanalysis variables, a global dataset of physically consistent environmental variables, including temperature, precipitation, surface solar radiation, surface pressure, and volumetric soil water. These variables are physically related to the spectral reflectance and radar backscatter recorded by Sentinel-1 and Sentinel-2, making them meaningful evaluation targets despite not being used during SSL pretraining. We complement this probing analysis with intrinsic representation metrics to characterize representation geometry and investigate how these properties relate to downstream performance and the encoding of environmental signals. Using DINO, MAE, and MoCo models trained under identical conditions, we show that representation-level metrics distinguish models with similar downstream benchmark performance, providing complementary information beyond task-driven benchmarks. We further find that the linear accessibility of environmental signals is associated with performance on environmentally dependent tasks in the PANGAEA benchmark. Finally, we release ERA5 annotations co-located with the SSL4EO dataset to enable physically grounded representation evaluation for future geospatial foundation models.