Broad Consistency Between Satellite and Vegetation Model Estimates of Net Primary Productivity Across Global and Regional Scales

Abstract

Drought has the potential to significantly decrease net primary production (NPP). However, the drought sensitivity of NPP remains unclear globally. Here we investigated the response of NPP to drought for the period 1982–2011 using NPP derived from satellite models, eight dynamic global vegetation models (DGVMs), and several drought indices. Our results indicate that the sign of drought response of NPP is consistent across satellites and DGVMs globally. Both the model and satellite data show a unimodal distribution of drought sensitivity across climatic gradients and indicate that semiarid regions are the most drought-sensitive areas. In addition, a decreasing drought trend was significantly correlated with an increasing trend in satellite-derived NPP or DGVM-derived NPP globally. However, the DGVM-derived NPP is more sensitive to drought than the satellite-derived NPP over most vegetated land areas and presents a stronger correlation with drought, higher drought sensitivity, and larger slopes between NPP and drought trends. The overresponse of the DGVM-derived NPP to drought relative to the satellite-derived NPP could be attributed to the higher drought sensitivity of the simulated leaf area index and overresponse of the DGVM-derived NPP to the vapor pressure deficit. Overall, large uncertainties remain in the response of NPP to drought and additional long-term observations are required to better represent NPP responses to water stress in DGVM models.