Simulating Gross Primary Productivity of vegetation under soil water stress using in-situ and reanalysis soil moisture data inputs

This study utilises in-situ and reanalysis soil moisture data inputs from various sources to evaluate the effect of soil water stress on Gross Primary Productivity (GPP) of different Plant Functional Types (PFTs) using Terrestrial Ecosystem Model in R (TEMIR), which is under development by Tai Group of Atmosphere-Biosphere Interactions (Tai et al. in prep.). An empirical soil water stress function with reference to Community Land Model (CLM) Version 4.5 is employed to quantify water stress experienced by vegetation which hinders stomatal conductance and thus carboxylation rate. The model results are compared against observations at FLUXNET sites in semi-arid regions across the globe at daily timescale where in-situ GPP data is available and water stress inhibits plant functions to some extent. By dividing the soil into two layers (topsoil and root zone), GPP simulation improves significantly comparing with using single layer bulk soil (Modified Nash-Sutcliffe Model Efficiency Coefficient N increases from -0.686 to -0.586). Such upgrade is particularly substantial for vegetation with shallow roots such as grass PFTs. Despite this improvement, the model is characterised by an overall overestimation of GPP when water stress occurs, and inconsistency of accuracy subject to PFTs and degree of water stress experienced. This study informs responses of various PFTs to soil water stress, capacity of TEMIR in simulating the responses, and possible drawbacks of empirical soil water stress functions, and highlights the importance of topsoil moisture data input for vegetation drought monitoring.

Keywords: Soil water stress, Terrestrial model representation, Photosynthesis, In-situ data, Reanalysis data, FLUXNET