Comparing the use of past and forecast weather data for estimating reference evapotranspiration

• Published in: Agricultural and forest meteorology 2020-12, Vol.295, p.108196, Article 108196
• Main Authors: Vanella, D., Intrigliolo, D.S., Consoli, S., Longo-Minnolo, G., Lizzio, G., Dumitrache, R.C., Mateescu, E., Deelstra, J., Ramírez-Cuesta, J.M.
• Format: Article
• Language: English
• Subjects: Irrigation; Numerical weather predictions; Weather ground-based observation
• ISSN: 0168-1923
• DOI: 10.1016/j.agrformet.2020.108196

•The weather variable that offered the most precise forecasts was air temperature.•Weather forecasts provided reliable reference evapotranspiration (ET0) estimates.•The forecast performance followed a decreasing trend with lead time.•ET0 was more accurate estimated from weather forecast scenario. The reliability of short-term weather forecast provided by COSMO model in simulating reference evapotranspiration (ET0) was evaluated in 7 study sites distributed in 4 countries (Italy, Norway, Romania and Spain). The main objective of the study was to assess the optimal scenario for calculating ET0, using the FAO-56 Penman-Monteith (PM) equation, by separately considering the accuracy in the use of “past” and “forecast” data input. Firstly, each forecasted variable (air temperature, Tair; relative humidity, RH; wind speed, u2; solar radiation, Rs) and ET0 were compared with in situ observations at hourly and daily scales. Moreover the seasonality effect in the forecast performance was evaluated. Secondly, simulated ET0 were computed every three days with: (i) a “past scenario” that used the observed data input measured in situ during the previous three days, (ii) a “forecast scenario” that used the forecasted input variables for the next three days; and compared with (iii) actual ET0 obtained from the in situ measured data. A general good agreement was found between observed and forecasted agro-meteorological parameters at the different explored time-scales. The best performance was obtained for Tair and Rs, followed by RH and u2. Globally, the comparison between ET0 from the measured and forecasted data input showed high performance, with R2 and RMSE of 0.90 and 0.68 mm d−1. ET0 simulations resulted more accurate using the “forecast scenario” (1.7% overestimation), rather than using the “past scenario” (2.6% underestimation). These results open promising perspectives in the use of forecast for ET0 assessment for different agriculture practices and particularly for irrigation scheduling under water scarcity conditions.