Assessment of evapotranspiration models for the humid tropical region of Tavanur

Abstract

World is facing an acute water crisis due to the increase of world population, droughts, land degradation, and food demand. This increases the concern over conservation of water. One of the most important factors related to water management is crop evapotranspiration. In the present research work, the reference crop evapotranspiration (ETo) is estimated by using ten empirical models which are widely used in Indian conditions namely, Thornthwaite (1948), Hargreaves et al., (1985), Turc (1961), Christiansen (1968) Pan Evaporation, FAO-24 Blaney-Criddle (1977), FAO-24 Modified Penman (1977), FAO-24 Open Pan (1977), Preistly-Taylor, Makkinik and FAO-56 Penman-Monteith (1991). The accuracy of these reference evapotranspiration models were evaluated by comparing it with FAO-56 Penman-Monteith using six years monthly average meteorological data for the period January, 2011-December, 2016. Then the models were validated with lysimetric data. The weekly water balance studies were conducted in lysimeter to find the actual reference evapotranspiration. The model values were estimated using weekly meteorological data for the period January-May 2017 during which the lysimeter study was conducted. Then best fit relations were developed between the estimated values (EToEST) and observed values (EToLYM) for the humid tropical region. Among the different empirical models, Turc model showed the highest ETo value (14.92 mm/day) while the Priestly-Taylor showed the lowest (0.62 mm/day). Thornthwaite, Blaney-Criddle and Modified Penman model gave closer values to each other 7.32, 8.9 and 7.09 mm/day respectively. While Christiansen, Penman-Monteith, Open Pan and Makkinik models gave values like 3.08, 3.23, 3.24 and 2.22 mm/day respectively which were slightly lower compared to the values obtained from the Hargreaves model (4.7 mm/day). The statistical comparison was made by considering FAO-56 PMM as the standard model using six year average monthly meteorological data. The Modified Penman model gave the best performance with R2 of 0.96 with RMSE 3.95 and RelRMSE 1.22 followed by Hargreaves model. The Open Pan method ranked the third one. The models, Christiansen, Priestly-Taylor and Makkinik were underestimated while Thornthwaite, Turc and Blaney-Criddle models overestimated. For validation of the models, weekly ETo estimated from models were compared with ETo observed from lysimeter for the period January-May, 2017. The Hargreaves model showed the best performance with R2 0.83 and RMSE 0.82. The Turc model was highly over estimated while Blaney-Criddle and Modified Penman models were only slightly overestimated. The Penman- Monteith and Makkinik models were slightly underestimated while Priestly- Taylor highly underestimated with R2 0.56 and the RMSE 4.29. Hence it is concluded that Hargreaves (HAM), Open Pan (OPM) and Christiansen (CHM) models were found to be in close agreement with lysimetric data and hence these models were suggested for use in this humid tropical region. Therefore relationships were developed between these empirical model output and the lysimetric data (LYM). The relationships developed were as follows: ETo LYM = 0.79HAM + 0.45, EToLYM = 0.79CHM + 1.60 and EToLYM = 0.63OPM + 2.04. Finally the results of this research can be recommended for humid tropical region for irrigation scheduling, selection of cropping pattern, optimum allocation of water resources and efficient use of water.