Interpreting spatial heterogeneity of crop yield with a process model and remote sensing

A process-based crop growth model (Vegetation Interface Processes (VIP) model) is used to estimate crop yield with remote sensing over the North China Plain. Spatial pattern of the key parameter—maximum catalytic capacity of Rubisco (Vcmax) for assimilation is retrieved from Normalized Difference of Vegetation Index (NDVI) from Terra-MODIS and statistical yield records. The regional simulation shows that the agreements between the simulated winter wheat yields and census data at county-level are quite well with R2 being 0.41–0.50 during 2001–2005. Spatial variability of photosynthetic capacity and yield in irrigated regions depend greatly on nitrogen input. Due to the heavy soil salinity, the photosynthetic capacity and yield in coastal region is less than 50 μmol C m−2 s−1 and 3000 kg ha−1, respectively, which are much lower than that in non-salinized region, 84.5 μmol C m−2 s−1 and 5700 kg ha−1. The predicted yield for irrigated wheat ranges from 4000 to 7800 kg ha−1, which is significantly larger than that of rainfed, 1500–3000 kg ha−1. According to the path coefficient analysis, nitrogen significantly affects yield, by which water exerts noticeably indirect influences on yield. The effect of water on yield is regulated, to a certain extent, by crop photosynthetic capacity and nitrogen application. It is believed that photosynthetic parameters retrieved from remote sensing are reliable for regional production prediction with a process-based model.