Comparison and Calibration of Temperature Based Methods of Reference Evapotranspiration for Abbay River Basin

Abstract

Accurate estimation of reference evapotranspiration (ETo) has been a challenge for hydrologists and engineers, mainly because of the space & time variability of the environmental and physical parameters governing ETo. In addition, most accurate ETo methods depend on intensive meteorological information for ETo estimation. Such data are not often readily available at the desired spatial and temporal scales. This limitation has necessitated the development of simple models that are less data intensive and provide ETo estimates with acceptable level of accuracy. In this study, the original coefficients of the Thornthwaite, Blaney-Criddle and Hargreaves were modified for Abbay (Blue Nile) river basin, taking the Penman-Monteith equations of FAQ as a standard. For this purpose meteorological data from fifteen stations in the basin were used for comparison and calibration of the methods. The comparison was first made by using the original constant coefficients in the above three methods and subsequently using recalibrated constant values. In the comparison the indices of regression were analyzed together with least square error and relative error indicators. The results of the study indicated that the use of the three temperature based methods have got limitations if employed without calibration in the study area. The original coefficients either underestimate or overestimate the ETo value as compared to that of the estimate of the FAQ Penman-Monteith method. The Thornthwaite method underestimates the ETo by average value of about 40%, Blaney-Criddle overestimates by average value of about 23% and Hargreaves overestimates or underestimates by average value of about 16%. The rigorous recalibration of the three methods enabled to maintain better relationship between the standard FAQ Penman-Monteith and the aforementioned temperature based methods. Statistical evaluation of the deviation of the estimated value o f the three models from the FAQ Penman-Monteith model revealed that the estimates from Hargreaves deviated by only about 3%, Thornthwaite by 5% and Blaney-Criddle by 6%. The above result depicts that Hargreaves method gives relatively better estimation than Thornthwaite and the latter give better result than Blaney-Criddle methods after calibration. Therefore, before these models are employed for reference evapotranspiration estimation it is crucial to calibrate the models using available local Climatological data/or all major basins and sub-basins.