Database for The Article: Leaf Area Index (LAI) Partitioning in Dryland Forests Using Dual-Sensor Sentinel Imagery

Leaf Area Index (LAI) is a key biophysical variable for understanding forest structure and function, particularly in dryland ecosystems where vegetation responses to environmental stressors are highly dynamic. Traditional satellite-derived LAI products are often inaccurate, as they typically represent the effective Plant Area Index (PAIeff), excluding necessary corrections such as consideration of leaf clumping and exclusion of woody elements. Moreover, while most studies report total ecosystem LAI, it is increasingly important to partition LAI into its structural components, such as overstory and understory layers, to better understand forest structure and function. In this study, we developed and validated a novel approach to estimate forest LAI (LAIEcosystem) and partition it into overstory (LAIOverstory) and understory (LAIUnderstory) components using an integrated model based on Sentinel-2 (S2) multispectral and Sentinel-1 SAR data (S1). Field LAI measurements were conducted in two Aleppo pine forests in Israel, HaKedoshim and Yatir, which span an aridity gradient and include multiple thinning treatments. Partial Least Squares Regression (PLS-R) models were calibrated and validated using extensive field data collected between 2018 and 2019. The combined S1+S2 model significantly improved LAIEcosystem predictions (R² = 0.85, RMSE = 0.52) and enabled fine-scale monitoring of vertical forest structure. Notably, S1 data substantially enhanced the prediction of understory LAI (R² = 0.87, RMSE = 0.26), highlighting the value of active sensing in capturing low-canopy structure. Comparisons with commonly used satellite LAI products (i.e., GCOM-C SGLI, Globe PROBA-V, Sentinel-3 OLCI, the MODIS, and the Simplified Level 2 Prototype Processor (SL2P) for S2 at 20 m and 10 m resolutions) revealed strong agreements for PAIeff, underscoring the importance of methodologies for correcting LAI estimation. This study presents an operational approach for LAI partitioning in dryland forests, providing novel insights into the structural responses of forests to climatic aridity and forest management. The methodology holds significant potential for enhancing global forest monitoring and improving the precision of ecosystem models under changing environmental conditions.叶面积指数（Leaf Area Index, LAI）是理解森林结构与功能的关键生物物理变量，在植被对环境胁迫的响应极具动态性的旱地生态系统中尤为重要。传统的卫星反演LAI产品往往精度不足，因为它们通常仅表征有效植物面积指数（effective Plant Area Index, PAIeff），未考虑叶片簇集校正以及排除木质组分等必要修正步骤。此外，尽管多数研究报告的是生态系统总LAI，但将LAI拆分为冠层上层和冠层下层等结构组分的需求日益增长，这有助于更深入地解析森林结构与功能。本研究开发并验证了一种全新方法，基于Sentinel-2（S2）多光谱与Sentinel-1 SAR（S1）数据构建集成模型，用以估算森林LAI（LAIEcosystem）并将其拆分为冠层上层（LAIOverstory）与冠层下层（LAIUnderstory）组分。野外LAI测量在以色列的两处阿勒颇松林地开展，分别为哈克多希姆（HaKedoshim）与亚提尔（Yatir），这些样地横跨干旱梯度，并包含多种疏伐处理方案。研究基于2018至2019年间采集的大量野外数据，对偏最小二乘回归（Partial Least Squares Regression, PLS-R）模型进行了校准与验证。结合S1与S2数据的模型显著提升了LAIEcosystem的预测精度（决定系数R²=0.85，均方根误差RMSE=0.52），并实现了森林垂直结构的精细化监测。尤为值得注意的是，S1数据大幅提升了冠层下层LAI的预测效果（R²=0.87，RMSE=0.26），凸显了主动遥感在捕捉低冠层结构方面的应用价值。将该方法与常用卫星LAI产品（即GCOM-C SGLI、Globe PROBA-V、Sentinel-3 OLCI、MODIS，以及适用于20m与10m分辨率S2数据的简化二级原型处理器（Simplified Level 2 Prototype Processor, SL2P））进行对比后发现，其与PAIeff的一致性良好，这进一步强调了LAI估算校正方法的重要性。本研究提出了一种可业务化应用的旱地森林LAI拆分方法，为解析森林对气候干旱及森林经营措施的结构响应提供了全新视角。该方法在提升全球森林监测能力、改善环境变化下生态系统模型的模拟精度方面，具备显著应用潜力。