Quantifying the Robustness of Chlorophyll-Sensitive Vegetation Indices to Multi-Source Disturbances in Sparse Canopies Using the PROSPECT-LESS Model

Leaf chlorophyll content (LCC) is a key biochemical trait that regulates vegetation photosynthetic capacity and reflects ecological status, making it a critical parameter for carbon cycle modeling, crop growth monitoring, and climate change research. However, retrieving LCC accurately from canopy reflectance remains challenging due to the spectral coupling between leaf biochemical properties, canopy structure, and soil background. These challenges are particularly acute under sparse canopy conditions, where interactions among soil reflectance, canopy parameters such as crown width, minimum plant spacing, crop row spacing, ridge width, and leaf spectral characteristics are especially pronounced. In this study, we coupled the PROSPECT leaf radiative transfer model with the LESS canopy model to generate 20 sparse canopy scenarios (12 eucalyptus stands and 8 maize fields). Across these scenarios, canopy reflectance was simulated under gradients of LCC, minimum tree spacing, leaf area index, row spacing, ridge width, and five levels of soil brightness. A total of 20 broadband vegetation indices (VIs) were then calculated, linearly normalized, and quantitatively assessed against four robustness criteria: resistance to saturation, resistance to canopy structure perturbations, resistance to soil background disturbances, and resistance to observation geometry variability. Our results indicate that Maccioni's 2001 Index (Macc01), Datt's 1999 Index (Datt99), and Red Edge Position (REP) outperform all other indices, with R² values ranging from 0.81 to 0.83 and disturbance-induced mean standard deviations (STD) below 0.05, identifying them as the most reliable options for accurate LCC retrieval in sparse canopies. Red-edge indices such as MERIS Terrestrial Chlorophyll Index (MTCI), Canopy Structure Index (CSI), Chlorophyll Vegetation Index (CVI), and Transformed Chlorophyll Absorption in Reflectance Index / Optimized Soil-Adjusted Vegetation Index (TCARI/OSAVI) also exhibit strong robustness. Traditional vegetation indices such as Normalized Difference Vegetation Index (NDVI), Structure Insensitive Pigment Index (SIPI), Enhanced Vegetation Index (EVI), and Soil- Adjusted Vegetation Index (SAVI) are highly sensitive to canopy structure and soil background and saturated early (R2< 0.5), limiting their applicability for LCC retrieval under sparse conditions. We therefore recommend prioritizing indices including Macc01, Datt99, REP, MTCI, CVI, CSI, and TCARI/OSAVI for remote sensing-based chlorophyll retrieval in sparse vegetation, thereby minimizing errors induced by soil and canopy structure disturbances and providing a sound basis for index selection in complex canopy conditions.