Abstract

CERVANTES-OSORNIO, Rocío et al. Reference Evapotranspiration Estimation by Hargreaves Priestley-Taylor and Artificial Neural Networks Models. Ing. invest. y tecnol. [online]. 2013, vol.14, n.2, pp.163-176. ISSN 2594-0732.

Directly measuring evapotranspiration accurately through a lysimeter is difficult, and because of data lack the Penman-Monteith method modified by FAO (ET₀ FAO-56 P-M) was used to obtain the reference evapotranspiration observed. The objective of the present study was to perform a comparison of empirical models like: Hargreaves, Hargreaves calibrated and Priestley-Taylor with the artificial neural network radial basis function (RNA BR) model with the same entry variables, in the estimation of reference evapotranspiration. The estimations of ET₀ were evaluated in four stations of District 075, Valle del Fuerte in Sinaloa, México. RNABR3 y RNABR7 used same entry variables (or less) than HARGC and P-T conventional methods, respectively. HARGC and P-T RMSE's at fitting changed from 0.7092 to 0.7848 and from 0.4178 to 0.8207, and on validation changed from 1.1898 to 0.6914 and from 0.3800 to 0.6889, respectively. RMSE's from RNABR3 and RNABR7 at fitting changed from 0.5295 to 0.6737 and from 0.3574 to 0.4809, and on validation from 1.3096 to 0.6254 and from 0.3470 to 0.4919, respectively. RNABR3, RNABR7 RMSE's obtained fitting as well as validation defined that RNA BR were better on the estimation of ET₀ FAO-56 P-M than conventional methods.

Keywords : basis radial function; Hargreaves; models; water necessities; Priestley and Taylor; forecasting.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )