Multi-source data modeling of the spatial distribution of winter wheat yield in China from 2000 to 2015

Abstract

Yield gridded datasets are essential for agricultural land management, food security and harmonious human-land relationships. Many studies have developed yield spatialization models that are based on cropland areas. However, crop planting areas, phenological dates, and net primary production (NPP) have received minimal attention. This study proposes a novel method to simulate winter wheat yields in China from 2000 to 2015 using crop phenological datasets, phenological observations, and NPP. The results showed that the NPP in the growing season and statistical yield showed a significant positive correlation (R2 = 0.93, p < 0.01). The mean prediction error of the gridded yield dataset was 12.01%. The relative errors of the gridded yield dataset for approximately half of the samples were between ??10% and 10%. Furthermore, the yield distribution was high in the east and low in the west. The high yield was primarily concentrated in the North China Plain, while low yield was observed in eastern Gansu, central Shanxi, southern Hebei, and eastern Sichuan. From 2000 to 2015, the yield mainly showed an increasing trend in the study area, with the average rate of 0.17 t ha-1 yr-1, especially in the North China Plain. This study suggests that NPP is a key indicator to evaluate the yield of winter wheat. Furthermore, this method can be used to generate gridded yield maps along with providing credible and fundamental data for climate change and sustainable agricultural development.