On the consistency and stability of vegetation biophysical variables retrievals from Landsat-8/9 and Sentinel-2

Abstract

Systematic decametric resolution global mapping of vegetation biophysical variables, including fraction of absorbed photosynthetically active radiation (fAPAR), fraction of vegetation cover (fCOVER), and leaf area index (LAI), is required to support various activities, including climate adaptation, crop management, biodiversity monitoring, and ecosystem assessments. The Canada Centre for Remote Sensing (CCRS) version of the Simplified Level 2 Prototype Processor (SL2P-CCRS) enables global mapping of these variables using freely available medium resolution multispectral satellite data from Sentinel-2 (S2) and Landsat-8/9 (LS) data. In this study, fiducial reference measurements (RMs) from the National Ecological Observatory Network (NEON) supplemented with regional measurements from CCRS were used to evaluate the consistency between SL2P-CCRS estimates of fAPAR, fCOVER and LAI from LS and S2 data and to quantify their temporal stability. SL2P-CCRS estimates of fCOVER (Accuracy (A) ∼ 0.03, Uncertainty (U) ∼ 0.13) and fAPAR (A ∼ −0.03, U ∼ 0.13) from LS and S2 were unbiased, and generally similar between sensors, based on 6569 LS-RMs and 4932 S2-RMs matchups. However, LAI estimates, especially for woody wetlands, deciduous forest, and mixed forest, were underestimated, with better estimates obtained using S2 (A ∼ −0.33, U ∼ 0.98) than LS (A ∼ −0.43, U ∼ 1.13). For all variables, SL2P-CCRS LS estimates were highly correlated to S2 estimates overall (R2 0.80 to 0.82) but up to 35 % lower for LAI over broadleaf and mixed forests and between lower 10 % and 20 % otherwise. The inter-annual stability of SL2P-CCRS estimates from both LS and S2 fell within the Global Climate Observing System (GCOS) requirements with the mean (standard deviation) values of −0.01 yr⁻¹ (0.06 yr⁻¹) for LS LAI, 0.02 yr⁻¹ (0.09 yr⁻¹) for S2 LAI, and 0 yr⁻¹ (0.01 yr⁻¹) for fCOVER and fAPAR from both LS and S2. The stability of both S2 and LS vegetation biophysical products indicate that are well suited for quantify the physical response of vegetation to climate variability, disturbances and regeneration.