Evaluation of NASA Satellite- and Model-Derived Weather Data for Simulation of Maize Yield Potential in China

Use of crop models is frequently constrained by lack of the required weather data. This paper evaluates satellite-based solar radiation and model-derived air temperature (maximum temperature, Tmax; minimum temperature, Tmin) from NASA and their utility for simulating maize (Zea mays L.) yield potential at 39 locations in China's major maize-producing regions. The reference data in this evaluation were the corresponding ground-observed weather data and simulated yield using these data. NASA weather data were closely correlated with data from ground weather stations with an r2 > 0.8, but a systematic underestimation of air temperature was found (Tmax of –2.8°C; Tmin of –1.4°C). As a result, use of NASA data alone for yield simulation gave poor agreement with simulated yields using ground weather data (r2 = 0.2). The simulations of yield potential using satellite-derived solar radiation, coupled with temperature data from ground stations, agreed well with simulated results using complete ground weather data in three of the five regions (r2 > 0.9). The agreement in the other two regions was relatively poor (r2 = 0.62 and 0.64). Across all 710 site-years evaluated, the agreement was shown with a mean error (ME) = 0.2 t ha–1, a root mean square error (RMSE) = 0.6 t ha–1, and r2 = 0.9. Our results indicate that combining NASA solar radiation with ground-station temperature data provides an option for filling geospatial gaps in weather data for estimating maize yield potential in China.