Evaluation of an ecosystem model for a wheat–maize double cropping system over the North China Plain

Abstract

A process-based ecosystem model (Vegetation–atmosphere Interface Processes (VIP) model) is expanded, and then validated against three years’ biometric, soil moisture and eddy-covariance fluxes data over a winter wheat–summer maize cropping system in the North China Plain (NCP). The results show that the model is capable of simulating satisfactorily the evolution of crop biomass, phenological development and soil moisture. The computed 30-min estimates of CO2, water and heat fluxes agree well with the eddy-covariance measurements. At daily scale, the root mean square errors (RMSEs) of net radiation, latent heat flux and net ecosystem productivity (NEP) are 1.0MJm−2day−1, 1.8MJm−2day−1 and 2.6gCm−2day−1, respectively. However, systematic errors in sensible heat flux estimates are identified in times of season when daily sensible heat flux is negative due to the horizontal advection. Annually, about 55% of evapotranspiration (ET) is emanated from winter wheat and 45% from maize. The annual NEP varies noticeably, with relative biases of 18, 9 and −29% in each year from 2003 to 2005, respectively. Sensitivity analysis illustrates that ET is quite sensitive to soil resistance parameters contributing to soil evaporation, and NEP to quantum efficiency of photosynthesis. The uncertainties of annual ET and NEP are 16.5% and 35.6% respectively when the key parameters are randomly sampled in their uncertainty ranges. Errors on eddy-covariance measurements and uncertainty on the model parameters may partly explain the discrepancy between the simulations and the measurements.