Retrieving Woody Components from Time-Series Gap-Fraction and Multispectral Satellite Observations over Deciduous Forests

Abstract

Highlights What are the main findings? A novel method was developed to estimate the woody fraction in deciduous forests by integrating time-series gap-fraction (PAI) measurements with Sentinel-2 NDVI data. The estimated woody fraction accurately captured the expected seasonal dynamics across forest types, increasing in deciduous forests during winter following leaf senescence, while remaining relatively stable in evergreen forests. What is the implication of the main finding? The proposed method offers a scalable framework for quantifying woody components across broad regions using satellite observations. It advances understanding of forest structural dynamics and enhances the application of gap-fraction data in vegetation monitoring and ecosystem modeling.Highlights What are the main findings? A novel method was developed to estimate the woody fraction in deciduous forests by integrating time-series gap-fraction (PAI) measurements with Sentinel-2 NDVI data. The estimated woody fraction accurately captured the expected seasonal dynamics across forest types, increasing in deciduous forests during winter following leaf senescence, while remaining relatively stable in evergreen forests. What is the implication of the main finding? The proposed method offers a scalable framework for quantifying woody components across broad regions using satellite observations. It advances understanding of forest structural dynamics and enhances the application of gap-fraction data in vegetation monitoring and ecosystem modeling.Abstract Leaf area index (LAI) is essential for understanding vegetation dynamics, ecosystem processes, and land-atmosphere interactions. Various measurement methods exist, but gap-fraction-based indirect methods are preferred due to their reduced labor and field survey time in comparison to direct methods. However, gap-fraction-based field observations, often referred to as plant area index (PAI), frequently overestimate LAI because they include woody components. To mitigate this issue, the woody-to-total-area ratio (alpha) can be utilized to exclude these woody components from PAI, yielding more accurate LAI estimates (hereafter referred to as LAIadjusted). In this study, we demonstrate a novel method to estimate alpha using Sentinel-2-based normalized difference vegetation index (NDVI) and time-series PAI measurements. The alpha estimates effectively reduce the influence of woody components in PAI within deciduous broadleaf forests (DBF). Moreover, LAIadjusted shows good agreement with the Sentinel-2 LAI, which represents effective LAI derived from the PROSAIL model. Notably, the spatial distribution of alpha effectively captured the expected seasonal dynamics across various forest types. In DBF, alpha values increased during winter due to leaf fall when compared to the growing season, while seasonal variations were relatively small in evergreen needleleaf forest (ENF). We further confirmed that our method demonstrates greater robustness with NDVI than with other vegetation indices that are more susceptible to topographic variation. Ultimately, this framework presents a promising pathway to mitigate biases in most gap-fraction-based PAI measurements, thereby enhancing the accuracy of vegetation structural assessments and supporting broader ecological and climate-related applications.