A prognostic vegetation phenology model to predict seasonal maximum and time series of global leaf area index using climate variables

Abstract

Vegetation phenology models provide us with tools for understanding and predicting vegetation responses to a changing climate. Vegetation phenology can be characterized by periodic changes in leaf area index (LAI), which quantifies leaf areas per unit ground surface areas. As an essential variable in land surface studies, LAI is generally used to simulate a range of fundamental vegetation processes and is essential to the simulations of land surface processes and land-atmosphere fluxes. Yet few studies directly provide one key variable required in simulating LAI, the annual seasonal maximum leaf area index (LAISM). There is a need to develop a method for predicting LAISM and allow for simulating daily LAI time series (LAITS). In this study, we developed a Prognostic Vegetation Phenology Model (PVPM) that predicts both LAISM and daily LAITS across biomes at the site and global scales, using climate variables. The method predicts LAISM based on an improved Budyko-like function and then predicts LAITS by integrating a previously developed semi-prognostic model. Experimental results show that PVPM is suitable for simulating vegetation dynamics and provides two parameterization schemes across different biomes at both site and global levels. For different biomes, the model could explain a large variance in satellite-derived LAISM and daily LAITS at the site scale, and well capture the spatial patterns and temporal trends of satellite-derived LAISM and LAITS across the globe. Compared to results obtained from other two process-based climate models (i.e., the GFDL-CM4 and CESM2), PVPM could make predictions on LAITS that match global satellite observations well (r = 0.9 and RMSE < 0.82). This study provides a prognostic vegetation phenology model that can be used in the land surface models to forecast global LAI time series for modeling vegetation processes.