Study of growth and yield of winter wheat using WOFOST model based on wireless sensor network data

This study aims to research the applicability of the WOFOST model and predict winter wheat yield on a regional scale. The automatic information in study area could be achieved from the wireless sensor network (WSN) deployed in cropland. The observed results including the phenological dates, LAI and the yield are compared with the simulated results in WOFOST. It is found that the maximum deviation of observed and simulated days taken to anthesis is seven days, and the maximum deviation of observed and simulated days taken to maturity is five days. The simulated days to anthesis and days to maturity indicate that WOFOST model underestimates the phenological period of wheat in both years. Through the analysis of LAI, it can be seen that simulated results are agree well with the observed results, except the early stage of growth. The phenomenon can be attributed to the simulated growth period is earlier than the actual test. Moreover, the observed mean yields are 5410 and 5536 kg ha(-1) during 2006-2007 and 2007-2008, respectively. The simulated mean yields are 5936 and 6027 kg ha(-1) during 2006-2007 and 2007-2008, respectively. Based on the relative root mean square error, the modified WOFOST model presented a very good to fair estimation of wheat yields. The results of this study provide an important theoretical reference for WOFOST model monitoring and estimating winter wheat growth status on a regional scale. It indicates that WOFOST model is a useful tool in yield forecasting.