The Bolivian Altiplano (high plateau between 3600 and 4000 m a.s.l.) is one of the highest agricultural areas in the world. Due to low levels of rainfall, high evapotranspiration rate and soils with low water retention capacity, water stress is a major constraint to crop production. Under these conditions, irrigation would be an asset to reduce the increased risk for agriculture. For that purpose, reliable reference evapotranspiration (E0) estimates for the design and management of irrigation systems are necessary but not available. In this study, E0 calculated by means of the Thornthwaite, Hargreaves-Samani and FAO Penman-Monteith equations is compared with measured (lysimeters) grass crop evapotranspiration (Egrass) during the growing period (October-April) at the Bolivian Highlands. The E0 estimated by means of the FAO Penman-Monteith method agrees well with E grass at the four locations. The temperature-based Hargreaves-Samani formula is able to estimate the reference evapotranspiration at the northern locations of the Altiplano but not at south due to the exclusion of aerodynamic factors affecting evapotranspiration. The temperature (mean temperature)-based Thornthwaite formula strongly underestimates E0 at all locations. The mean bias error (Merror) for estimates of grass crop evapotranspiration applying the Penman-Monteith, Hargreaves-Samani and Thornthwaite and compared to the lysimetric measurements, were (on average for the four locations) of -0.2, -0.4 and 2.2 mm per day, respectively, when the average lysimetric grass crop evapotranspiration was 4.3 mm per day for the growing season, demonstrating the suitability of the application of the FAO Penman-Monteith equation in the Altiplano. To overcome the problem of the availability of climatic parameters for the application of this equation, applications of procedures for estimating E0 by means of the FAO Penman-Monteith method with a limited data set of climatic data revealed that the difference between E0 obtained with a full and limited data set both applying this equation, is smaller than deviations resulting from the use of another method. Both the Merror and the root mean square error (Rerror) of the comparison of full and limited sets of data are less than 0.4 mm per day leading to small errors in the E0 estimates. The higher deviations occur when the only available information is minimum and maximum temperature.
Bibliographical noteFunding Information:
This work was possible thanks to the financial support of the International Foundation for Science through the grant C-3019.