Abstract: The precision of cropland soil moisture prediction is closely related to the accuracy of weather forecast. To clarify the influence of weather forecast accuracy on the precision of soil moisture prediction, we used soil water balance model of cropland to simulate the daily soil moisture in Zhengzhou, based on soil moisture data and daily meteorological data of automatic observation station from 2010 to 2020. We analyzed the simulation precision of the model in different months and different crops. The variations of daily average temperature (±1-2 ℃) and precipitation (±10%-20%) were set up to analyze the effects of temperature and precipitation forecast bias on the precision of soil moisture simulation. The results showed that annual variation of R² presented a trend of “big-small-big”, while that of NRMSE and MARE presented a “small-big-small” pattern. The simulation performance was better in November to the next March than in other months. Simulation precision for winter wheat soil moisture was higher than that of summer maize, with R² of 0.910, NRMSE of 15.14% and MARE of 8.57% for winter wheat and 0.841, 27.62%, and 15.49% for summer maize, respectively. A deviation of ±1 ℃/2 ℃ in temperature forecast had a slight impact on NRMSE and MARE, but with stable R². The simulation precision of soil moisture was higher under the higher forecast value of temperature, and vice versa. When the precipitation forecast value was more than 10%-20%, the precision of soil moisture simulation was lower than that when the precipitation was normal. Conversely, the simulation precision was improved, but the maximum amplitudes of NRMSE and MAREwere 6.16% and 3.23%, respectively, both were smaller than the bias range of precipitation forecast. When there were deviations in both temperature and precipitation forecasts, the variation of soil moisture simulation precision was mainly determined by the bias of precipitation forecast, i.e., when the forecast value of precipitation was higher than the actual value, the simulation precision decreased, and vice versa. The maximum variations of NRMSE and MARE were 6.19% and 3.23%, respectively.

Key words: temperature, precipitation, soil moisture, water balance, evapotranspiration.