Regional-Scale Wilting Point Estimation Using Satellite SIF, Radiative-Transfer Inversion, and Soil-Vegetation-Atmosphere Transfer Simulation: A Grassland Study

Abstract

Although water availability strongly controls gross primary production (GPP), the impact of soil moisture content (SMC) (wilting point) is poorly quantified on regional and global scales. In this study, we used 10 years of observations of solar-induced chlorophyll fluorescence (SIF) from the Greenhouse gases Observing Satellite (GOSAT) satellite to estimate the wilting point of a semiarid grassland on the Mongolian Plateau. Radiative-transfer model inversion and soil-vegetation-atmosphere transfer simulation were sequentially conducted to distinguish the drought impacts on plant physiology from the changes in the leaf-canopy optical properties. We modified an existing inversion algorithm and the widely used Soil-Canopy Observation of Photosynthesis and Energy fluxes model to adequately evaluate dryland features, for example, sparse canopy and strong convection. The modified model, with retrieved parameters and calibration to GOSAT SIF, predicted realistic GPP values. We found that (a) the SIF yield estimated from GOSAT showed a clear sigmoidal pattern in relation to drought, and the estimated wilting point matched ground-based observations in the literature within ∼0.01 m3 m−3 for the SMC, (b) tuning the maximum carboxylation rate improved the SIF prediction after considering the changes in the leaf-canopy optical properties, implying that GOSAT detected drought stress in leaf-level photosynthesis, and (c) the surface energy balance significantly impacted the grassland's SIF; the modified model reproduced observed SIF well (mean bias = 0.004 mW m−2 nm−1 sr−1 in summer), whereas the original model predicted substantially low values under weak horizontal wind conditions. Some model-observation mismatches in the SIF suggest that more research is needed for fluorescence parametrization (e.g., photoinhibition) and for additional observation constraints.