arXiv:2512.11524v3 Announce Type: replace-cross 
Abstract: Fine-scale forest monitoring is essential for understanding canopy structure and its dynamics, which are key indicators of carbon stocks, biodiversity, and forest health. Deep learning is particularly effective for this task, as it integrates spectral, temporal, and spatial signals that jointly reflect the canopy structure. To address this need, we introduce THREASURE-Net, a novel end-to-end framework for Tree Height Regression And Super-Resolution. The model is trained on Sentinel-2 time series using reference height metrics derived from LiDAR HD data at multiple spatial resolutions over Metropolitan France to produce annual height maps. We evaluate three model variants, producing tree-height predictions at 2.5 m, 5 m, and 10 m resolution. THREASURE-Net does not rely on any pretrained model nor on reference very high resolution optical imagery to train its super-resolution module; instead, it learns solely from LiDAR-derived height information. Our approach outperforms existing state-of-the-art methods based on Sentinel data and is competitive with methods based on very high resolution imagery. It can be deployed to generate high-precision annual canopy-height maps, achieving mean absolute errors of 2.63 m, 2.70 m, and 2.88 m at 2.5 m, 5 m, and 10 m resolution, respectively. These results highlight the potential of THREASURE-Net for scalable and cost-effective structural monitoring of temperate forests using only freely available satellite data. The source code for THREASURE-Net is available at: https://github.com/Global-Earth-Observation/threasure-net.

تم تقديم إطار عمل جديد يسمى THREASURE-Net لقياس ارتفاع الأشجار والتكبير الفائق، باستخدام بيانات السلاسل الزمنية من Sentinel-2 وبيانات مرجعية من LiDAR HD جوي عبر فرنسا الحضرية. يهدف هذا النموذج إلى تحسين مراقبة الغابات على نطاق دقيق من خلال إنتاج خرائط ارتفاع سنوية بدقة 2.5 م و5 م و10 م دون الاعتماد على نماذج مدربة مسبقًا أو صور بصرية عالية الدقة.

Se ha introducido un nuevo marco llamado THREASURE-Net para la regresión de la altura de los árboles y la super-resolución, utilizando datos de series temporales de Sentinel-2 y datos de referencia de LiDAR HD aéreos en toda la Francia metropolitana. Este modelo tiene como objetivo mejorar el monitoreo forestal a escala fina produciendo mapas de altura anuales con resoluciones de 2.5 m, 5 m y 10 m sin depender de modelos preentrenados o imágenes ópticas de muy alta resolución.

Un nouveau cadre nommé THREASURE-Net a été introduit pour la régression de la hauteur des arbres et la super-résolution, utilisant des données temporelles Sentinel-2 et des données de référence LiDAR HD aériennes à travers la France métropolitaine. Ce modèle vise à améliorer le suivi forestier à fine échelle en produisant des cartes de hauteur annuelles avec des résolutions de 2,5 m, 5 m et 10 m sans s'appuyer sur des modèles pré-entraînés ou des images optiques de très haute résolution.

A new framework named THREASURE-Net has been introduced for tree height regression and super-resolution, utilizing Sentinel-2 time series data and airborne LiDAR HD reference data across Metropolitan France. This model aims to enhance fine-scale forest monitoring by producing annual height maps with resolutions of 2.5 m, 5 m, and 10 m without relying on pre-trained models or high-resolution optical imagery.