Analysis of cumulative precipitation forecast accuracy and its impact on rice irrigation scheduling in China

Introduction Rising rice demand has intensified pressure on irrigation water resources in China. Improving rainfall water use is an important way to reduce the pressure on irrigation water and avoid wasting water. Cumulative precipitation forecasts are expected to better match rice water needs with rainfall. Their accuracy and effectiveness for rice irrigation scheduling have not been evaluated. Methods In this study, cumulative precipitation forecasts with 3, 5, and 7 days of lead time were defined to improve the use of precipitation forecast. On this basis, these cumulative precipitation forecasts and the daily precipitation forecasts have been applied for irrigation forecast using rules at a China-wide scale. The quality of cumulative precipitation forecast and daily precipitation forecast were compared, and their effects on irrigation compared with conventional flood irrigation evaluated. Results The results showed that the precipitation forecasts from public weather forecasts had a clear seasonal and spatial pattern in the different rice cropping regions. They had better performance in inland gently sloping regions than in highland or coastal regions. Percentage correct ( PC ), defined as the proportion of forecast precipitation grades, correctly predicted compared to measured rainfall data, which ranged from 0.413 to 0.758. Precipitation amount correct rate ( PAC ), defined as the ratio of correctly predicted precipitation amount to observed amount, was higher for cumulative forecasts than that for the daily forecasts. All forecasts granted a higher precipitation utilization efficiency ( PUE ) of the paddy field when compared against conventional irrigation, with the 7-day cumulative precipitation forecast reaching an average PUE of 0.603. Generally, the volume of the water used for irrigation and the drainage one was reduced when employing the cumulative precipitation forecast and the number of irrigation and drainage actions. These reduction effects intensified with longer cumulative forecast horizons. The best performance was obtained for the 7-day cumulative precipitation forecasts, which allowed reducing the irrigation water by 20.81% and the number of irrigation actions by 25.80%. Irrigation scheduling based on cumulative precipitation forecasts had a small effect on yield, approximately 0.01%. Discussion The improved irrigation efficiency observed with cumulative precipitation forecasts, especially the 7-days forecast, suggests that integrating cumulative precipitation predictions into irrigation decisions can significantly reduce water use without compromising rice yield.