# Transferability and operational reliability of a Prithvi crop classification foundation model under phenological and geographic shift across three continents

arXiv: https://arxiv.org/abs/2610.08810 · HTML: https://mlsanity.com/paper/2610.08810/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 20 of its 45 references that are in the map.
- First posted on arXiv 20 Sep 2026.

## Builds on (most cited first, 20 of 20)

- [On Calibration of Modern Neural Networks](https://mlsanity.com/paper/1706.04599/) (2017)
- [Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learning](https://mlsanity.com/paper/1506.02142/) (2015)
- [SatlasPretrain: A Large-Scale Dataset for Remote Sensing Image Understanding](https://mlsanity.com/paper/2211.15660/) (2022)
- [Foundation Models for Generalist Geospatial Artificial Intelligence](https://mlsanity.com/paper/2310.18660/) (2023)
- [Prithvi-EO-2.0: A Versatile Multi-Temporal Foundation Model for Earth Observation Applications](https://mlsanity.com/paper/2412.02732/) (2024)
- [PANGAEA: A Global and Inclusive Benchmark for Geospatial Foundation Models](https://mlsanity.com/paper/2412.04204/) (2024)
- [TerraTorch: The Geospatial Foundation Models Toolkit](https://mlsanity.com/paper/2503.20563/) (2025)
- [TimeMatch: Unsupervised Cross-Region Adaptation by Temporal Shift Estimation](https://mlsanity.com/paper/2111.02682/) (2021)
- [A Sentinel-2 multi-year, multi-country benchmark dataset for crop classification and segmentation with deep learning](https://mlsanity.com/paper/2204.00951/) (2022)
- [No One Knows the State of the Art in Geospatial Foundation Models](https://mlsanity.com/paper/2605.12678/) (2026)
- [From parcel to continental scale -- A first European crop type map based on Sentinel-1 and LUCAS Copernicus in-situ observations](https://mlsanity.com/paper/2105.09261/) (2021)
- [AgriFM: A Multi-source Temporal Remote Sensing Foundation Model for Agriculture Mapping](https://mlsanity.com/paper/2505.21357/) (2025)
- [Deploying Geospatial Foundation Models in the Real World: Lessons from WorldCereal](https://mlsanity.com/paper/2508.00858/) (2025)
- [On the Generalizability of Foundation Models for Crop Type Mapping](https://mlsanity.com/paper/2409.09451/) (2024)
- [SHRUG-FM: Reliability-Aware Foundation Models for Earth Observation](https://mlsanity.com/paper/2511.10370/) (2025)
- [Towards Global Crop Maps with Transfer Learning](https://mlsanity.com/paper/2211.04755/) (2022)
- [From Time-series Generation, Model Selection to Transfer Learning: A Comparative Review of Pixel-wise Approaches for Large-scale Crop Mapping](https://mlsanity.com/paper/2507.12590/) (2025)
- [FARM: Fine-Tuning Geospatial Foundation Models for Intra-Field Crop Yield Regression](https://mlsanity.com/paper/2510.26609/) (2025)
- [Do Foundation Model Embeddings Improve Cross-Country Crop Yield Generalisation? A Leave-One-Country-Out Evaluation in Sub-Saharan Africa](https://mlsanity.com/paper/2605.08113/) (2026)
- [Land cover and flood type govern the detection limits of satellite-based flood mapping across diverse global flood events](https://mlsanity.com/paper/2606.07780/) (2026)

## Abstract (arXiv)

Fine-tuned geospatial foundation models (GeoFMs) pretrained on large satellite archives have been shown to improve crop classification accuracy and geographic transferability. However, their operational performance beyond the training distribution remains poorly characterized. We evaluated the out-of-distribution performance of a widely adopted GeoFM [Prithvi-EO-2.0] across 37 events in 12 countries on three continents and validated against regional reference products. Results indicated that the mean overall accuracy (OA) declined from 0.65 in the United States to 0.40 in Europe. Beyond accuracy metrics, we assessed five key aspects of model performance: whether model confidence indicates signal failure, sensitivity to observation windows, the effect of coarsening class schemes, and robustness to both band loss and cloud- and shadow-contamination. Accuracy collapsed when the observation window misaligned with local crop phenology, while deterministic confidence remained high. Expected calibration error increased for seven of eight paired events, and 12-51% of each affected scene was confidently mislabeled at near-zero precision. Monte Carlo dropout entropy registered the shift in all eight, indicating that much of the apparent cross-continent decline reflected phenological misalignment rather than spatial transfer. Two adjustments recovered accuracy without retraining. Consolidating 13 classes into 10, based on the model's dominant confusions, raised the mean OA by 8.4 percentage points. Compressing the window toward near-real-time use preserved accuracy across a 45- to 90-day plateau, peaking near 75 days, though arms tighter than 30 days fell about 0.11 below that plateau. Fine-tuned crop GeoFMs therefore transfer usefully only where observation windows match local growing seasons. We translate these findings into operational guidance for the reliable deployment of the released model.