Rooting depth projections of global plant functional types and driving factors analysis based on a hybrid modeling framework
Qinggong Han, Bo Liu, Yunuo Liu, Jielin Zhang, Yongxia Ding, Shouzhang Peng
Abstract
Rooting depth serves as a critical parameter in earth system and hydrological models. However, the lack of rooting depth gridded dataset has hindered investigations into the interactions between rooting depth and terrestrial hydrological and biogeochemical processes worldwide. In this study, we developed a hybrid model that combines random forest algorithm with ecosystem data simulated by LPJ-GUESS model and environmental variables to predict global rooting depth, based on 1,184 integrated rooting depth observations. Using this framework, we generated the first global gridded dataset of rooting depth for 13 plant functional types and embedded uncertainty, at 0.05° spatial resolution (∼5 km) from 2015 to 2100 with five-years interval under different climate scenarios. Our results indicated that the hybrid model performs with high accuracy, and vegetation types exhibited different responses under future climate scenarios. The mean rooting depth in 2100 changed by 19.5 %–82.5 % for tree, by 13.4 %–47.9 % for shrub, and by 3.0 %–23.5 % for grass compared to the 2015 baseline. Rooting depth in different periods showed similar spatial heterogeneity, which was mainly related to factors such as plant functional types, climate, elevation, and soil. The results of this study provided new methodological perspectives and data support for terrestrial biosphere modeling.