Not Every Tree Is a Forest: Benchmarking Forest Types from Satellite Remote Sensing

TL;DR

This work introduces ForTy, a global, multi-temporal benchmark for forest typology mapping that differentiates natural forests, planted forests, and tree crops using 200k time-series patches from multi-modal satellite data. It provides a comprehensive dataset pipeline, including diverse reference data sources, input modalities, and balanced sampling, to support supervised learning and potential self-supervised foundation models. The authors propose MTSViT, a transformer-based architecture that fuses multi-modal, multi-temporal information through image embedding, spatial and temporal encoders, and a cross-modal decoder to achieve superior per-class segmentation performance. Experiments and ablations demonstrate the method’s advantages over strong baselines and highlight the value of temporal and modal diversity for accurate forest-type mapping, with implications for policy and conservation efforts.

Abstract

Developing accurate and reliable models for forest types mapping is critical to support efforts for halting deforestation and for biodiversity conservation (such as European Union Deforestation Regulation (EUDR)). This work introduces ForTy, a benchmark for global-scale FORest TYpes mapping using multi-temporal satellite data1. The benchmark comprises 200,000 time series of image patches, each consisting of Sentinel-2, Sentinel-1, climate, and elevation data. Each time series captures variations at monthly or seasonal cadence. Per-pixel annotations, including forest types and other land use classes, support image segmentation tasks. Unlike most existing land use products that often categorize all forest areas into a single class, our benchmark differentiates between three forest types classes: natural forest, planted forest, and tree crops. By leveraging multiple public data sources, we achieve global coverage with this benchmark. We evaluate the forest types dataset using several baseline models, including convolution neural networks and transformer-based models. Additionally, we propose a novel transformer-based model specifically designed to handle multi-modal, multi-temporal satellite data for forest types mapping. Our experimental results demonstrate that the proposed model surpasses the baseline models in performance.