# Multi-Year Geospatial Reasoning using Interannually-Consistent Historical Predictions as a Free Input Modality

arXiv: https://arxiv.org/abs/2608.05979 · HTML: https://mlsanity.com/paper/2608.05979/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 19 of its 31 references that are in the map.
- First posted on arXiv 6 Aug 2026.

## Builds on (most cited first, 19 of 19)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Distilling the Knowledge in a Neural Network](https://mlsanity.com/paper/1503.02531/) (2015)
- [Class-Balanced Loss Based on Effective Number of Samples](https://mlsanity.com/paper/1901.05555/) (2019)
- [A Transformer-based Framework for Multivariate Time Series Representation Learning](https://mlsanity.com/paper/2010.02803/) (2020)
- [SatMAE: Pre-training Transformers for Temporal and Multi-Spectral Satellite Imagery](https://mlsanity.com/paper/2207.08051/) (2022)
- [Foundation Models for Generalist Geospatial Artificial Intelligence](https://mlsanity.com/paper/2310.18660/) (2023)
- [Panoptic Segmentation of Satellite Image Time Series with Convolutional Temporal Attention Networks](https://mlsanity.com/paper/2107.07933/) (2021)
- [Lightweight, Pre-trained Transformers for Remote Sensing Timeseries](https://mlsanity.com/paper/2304.14065/) (2023)
- [Entity Embeddings of Categorical Variables](https://mlsanity.com/paper/1604.06737/) (2016)
- [Temporal Convolutional Neural Network for the Classification of Satellite Image Time Series](https://mlsanity.com/paper/1811.10166/) (2018)
- [Self-attention for raw optical Satellite Time Series Classification](https://mlsanity.com/paper/1910.10536/) (2019)
- [AnySat: One Earth Observation Model for Many Resolutions, Scales, and Modalities](https://mlsanity.com/paper/2412.14123/) (2024)
- [Lightweight Temporal Self-Attention for Classifying Satellite Image Time Series](https://mlsanity.com/paper/2007.00586/) (2020)
- [BreizhCrops: A Time Series Dataset for Crop Type Mapping](https://mlsanity.com/paper/1905.11893/) (2019)
- [TerraFM: A Scalable Foundation Model for Unified Multisensor Earth Observation](https://mlsanity.com/paper/2506.06281/) (2025)
- [EuroCrops: A Pan-European Dataset for Time Series Crop Type Classification](https://mlsanity.com/paper/2106.08151/) (2021)
- [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)
- [Deploying Geospatial Foundation Models in the Real World: Lessons from WorldCereal](https://mlsanity.com/paper/2508.00858/) (2025)
- [Crop Rotation Modeling for Deep Learning-Based Parcel Classification from Satellite Time Series](https://mlsanity.com/paper/2110.08187/) (2021)

## Abstract (arXiv)

Machine learning, and deep networks in particular, are increasingly used to derive higher-level Earth observation (EO) products such as annual land-cover and crop-type maps. Many are generated operationally: each year a new acquisition is processed, typically with the same model, extending a multi-year archive. In the process these systems accumulate two kinds of useful signal that are almost never fed back into the model: the system's own archive of past predictions, and ancillary layers produced by other partners in a processing consortium. Both are normally used outside the network, as rule-based post-processing or a fixed input mask. Using the Copernicus Land Monitoring Service High Resolution Layer (HRL) Croplands crop-type product as a testbed, we show that bringing both signals inside the model turns a single-year, single-task pixel classifier into one that reasons across years. We introduce a Crop Type (CTY) embedding encoder that represents each past prediction as a confidence-scaled, time-ordered categorical token and attends over the year axis, and we study how the externally provided Base Vegetation Layer (BVL) mask should be represented in the model's inputs and outputs. To compare designs fairly when they relabel non-crop pixels, we evaluate on the 18 crop classes only and report precision and recall separately. On a pan-European dataset of about 5.4M labelled pixels, adding the prediction history raises crop-only F1 by 1.6 percentage points (pp) and, more importantly, corrects a recall-skewed error profile, with the largest gains on perennial and tree crops (olives +4.6, fruits +3.7, nuts +3.2 pp). Representing the BVL mask consistently in both the history and the target year adds about 2.5 pp on the crop classes. The approach is a low-cost recipe for any recurring geospatial or foundation model that emits class maps.