Comprehensive evaluation of global CI, FVC, and LAI products and their relationships using high-resolution reference data

Abstract

In recent decades, several global clumping index (CI), fractional vegetation cover (FVC), and leaf area index (LAI) products have been generated using a range of optical satellite sensors. It is essential for the application community to understand the accuracy and uncertainty of these products. However, current validation studies have limited spatial coverage and temporal continuity, and the relationships between different products are not clearly understood. This study aims to validate the most recent global CAS-CI, GEOV2 FVC, and MODIS LAI products using the Ground-Based Observations for Validation (GBOV) and DIRECT 2.1 datasets. The relationships between these products, the leaf projection function in the nadir direction (G(0)), and the vertical structural characteristics of the forest were also analyzed. The GEOV2 FVC and MODIS LAI show good performance, whereas CAS-CI shows a slight underestimation (bias = −0.10) and strong seasonal variations. FVC and LAI are strongly correlated at different spatial scales [field, high-resolution (30 m and upscaled 3 km), and moderate-resolution (3 km)], particularly for forests [mean Pearson's correlation coefficients (r) = 0.90]. The CI-FVC/LAI relationship is more consistent across spatial scales for grass and shrubs than for forests. The derived G(0) overestimates the reference value (bias = 0.28) mainly due to inconsistencies between the CI, FVC, and LAI products. The overstory CI is very close to the overall forest CI, indicating that the understory may be neglected when estimating the overall CI. The overstory G(0) is similar to the overall G(0) for most forest types, except for the evergreen needleleaf forest. The overall CI-FVC/LAI relationship is mainly influenced by the overstory, while the FVC-LAI relationship is affected by both the overstory and understory. Overall, the CI-FVC/LAI relationship should be further evaluated and applied cautiously across different scales. The forest vertical structural characteristics revealed in this study are important for field measurement, remote sensing estimation, and modeling studies.