Forecasting daily reference evapotranspiration for Australia using numerical weather prediction outputs

Abstract

Farmers and irrigation system operators make real-time irrigation decisions based on a range of
factors including crop water requirement and short-term weather forecasts of rainfall and air temperature.
Forecasts of reference crop evapotranspiration (ETO) can be calculated from numerical weather prediction
(NWP) forecasts and ETO has the advantage of being more directly relevant to crop water requirements than
air temperature. This paper aims to discuss the forecasting ability of ETO using outputs from the Bureau of
Meteorology's operational NWP forecasts derived from the Australian Community Climate and Earth System
Simulator – Global (ACCESS-G). The daily ETO forecasts were evaluated for the Shepparton Irrigation Area
in Victoria. Forecast performance for ETO was quantified using the root mean squared error (RMSE),
coefficient of determination (r2), anomaly correlation coefficient (ACC) and mean square skill score (MSSS).
Lead times of daily ETO forecasts up to 9 days were compared against ETO calculated using hourly
observations from the Shepparton airport automatic weather station. It was found that forecasting daily ETO
was better than using the long-term monthly mean ETO for lead times up to 6 days and beyond that the longterm
monthly mean was better. The average MSSS of ETO forecasts varied between 64% and 4 % for 1 to 6
day lead times, respectively. The most influential forecast weather variable for daily ETO forecasts was mean
wind speed, air temperature and incoming solar radiation for 1, 2-3 and 4-9 day lead times respectively. Also,
it was found that the forecast performance for incoming solar radiation and mean wind speed was relatively
poor compared with the air and dew point temperatures