Jump to Main Content
Performance of simple temperature-based evaporation methods compared with a time series of pan evaporation measures from a standard 20 m² tank
- Ansorge, Libor, Beran, Adam
- Journal of water and land development 2019 v.41 no.1 pp. 1-11
- data collection, equations, evaporation, evapotranspiration, temperature, time series analysis, water footprint, water management
- Evaporation and evapotranspiration is crucial part of hydrological and water resource management studies e.g. water footprinting. Proper methods for estimating evaporation/potential evapotranspiration using limited climatic data are critical if the availability of climatic data is extremely limited. In a large scale studies are very often used generalized (modelled or gridded) input data. For a large scale water footprint studies is also important to find methods as simple as possible with quantifiable error. In our study, nine simple temperature-based empirical equations were compared with a long term time series of real evaporation data from a 20 m² tank at Hlasivo station. In the first step, we used real temperature measured at Hlasivo station for validation of equations. In the second step, the gridded temperature data (interpolated datasets) derived from the meteorological stations were used. For both datasets, the differences between observed and predicted values were categorized into three groups of accuracy and the statistical indices of each equation were calculated. Very good results were achieved with the Hamon equation from 1961 and the Oudin equation for both datasets with index of agreement (d) higher than 0.9, cross-correlation coefficient (R²) around 0.7 and root mean square error (RMSE) around 0.5 mm∙(24 h)⁻¹ The Kharrufa equation, which was developed for semi-arid or arid areas, also provides results with sufficient accuracy. Comparison of the results with similar studies showed a lower accuracy of very simple equations against more complex equations, which have RMSE lower than 0.25 mm∙(24 h)⁻¹. But for some kind of studies, quantifiable errors with sufficient accuracy can be more important than the absolute accuracy.