Comparative study of conventional and artificial neural network-based ETo estimation models

摘要: Accurate estimation of reference crop evapotranspiration (ETo) is required for several hydrological studies and thus, in the past, a number ETo methods have been developed with different degree complexity data requirement. The present study was carried out to develop artificial neural network (ANN) based models corresponding ASCE’s best ranking conventional (Jensen et al. ASCE Manual Rep. on Engrg. Pract. no. 70, 1990). Among radiation methods, FAO-24 (or Rad) method arid Turc humid region, among temperature Blaney–Criddle BC) were studied. ANN architectures above three less data-intensive four CIMIS (California Irrigation Management Information System) stations, namely, Davis, Castroville, Mulberry, West Side Field station. comprehensive architecture by Kumar (J Irrig Drain Eng 128(4):224–233, 2002) Penman–Monteith (PM) Davis also tried other stations. Daily meteorological period more than 10 years (01 January 1990 30 June 2000) collected from these stations used train, test, validate models. Two learning schemes, standard back-propagation rate 0.2 momentum having term 0.95 considered. performance compared FAO-56 PM method. It found that gave better closeness each category (radiation temperature). Thus can be agreement climatic availability, when not all variables are observed.