Short-term forecasting of daily reference evapotranspiration using the reduced-set Penman-Monteith model and public weather forecasts.

Highlights • Four types of wind speed data were used in reduced set Penman-Monteith model for ETo forecasting. • Three modified reduced set Penman-Monteith models were proposed. • ETo forecast performance can be significantly improved through considering wind speed. • Wind speed data type was recommended for each station in different climate zone. Abstract Farm irrigation management and real-time irrigation decisions are based on a series of parameters including irrigation water demand predictions which estimated by the short-term daily reference evapotranspiration (ETo). Based on the temperature data, a simplest empirical model called the temperature-based method, was developed to forecast near-future ETo. Although temperature is one of the most influential weather variables in ETo forecasting, wind velocity also has non-ignorable impact on the forecast performance of ETo. Therefore, we employed the Reduced-set Penman-Monteith (RPM) model for short-term ETo forecasting using temperature data derived from public weather forecasts and wind speed data. Four different types of wind speed data were adopted as model inputs, including wind speed assumed as a default value, i.e., u 2 = 2.0 m s−1 (W1), forecasted wind speed (W2), long-term daily average wind speed (W3) and annual average wind speed (W4). Based on the above-mentioned wind speed inputs, we presented four types of RPM models (RC, RF, RD and RM models respectively corresponding to W1, W2, W3 and W4) to forecast daily ETo for lead times up to 7 days for eight weather stations in China. The results confirmed that the highest precision wind speed input was W3, followed by W4, W1 and W2, and the most accurate ETo forecasts of four RPM models was provided by RC, followed with RD, RM and RF. By comparing with one of the most common used temperature-based models, the Hargreaves-Samani (HS), it was found that the performance of RPM approaches were better than that of HS model for arid and sub-arid regions whereas opposite results were found in sub-tropical area, suggesting inclusion of wind speed parameter makes positive impact on ETo forecast for arid and semi-arid regions. Further, it was found that the greatest type of RPM model, the RC model, was superior to HS method for most climate regions, including arid, humid, sub-arid and sub-humid areas, thus, the W1 wind speed type could be highly recommended for ETo forecasting for most climate regions. [ABSTRACT FROM AUTHOR]