Applicability Of A 4 Inputs ANN Model For ETo Prediction In Coastal And Inland Locations
Price
Free (open access)
Transaction
Volume
112
Pages
10
Page Range
167 - 176
Published
2008
Size
412 kb
Paper DOI
10.2495/SI080171
Copyright
WIT Press
Author(s)
P. Martí, A. Royuela, J. Manzano & G. Palau
Abstract
Artificial Neural Networks (ANNs) are simplified models of the central nervous system that can be used as effective tools to model nonlinear problems. This paper reports the study of the applicability of an ANN-based ETo predicting-model in different geographical contexts of the Valencia region. The proposed model only demands the measurement of the maximum and minimum daily air temperatures and the calculation of extraterrestrial radiation and daylight hours. The model provides acceptable approximations of Penman-Monteith ETo values, better than already existing ETo predicting tools (Hargreaves), for coastal locations, where the sea contributes to hinder drastic climatological fluctuations. Nevertheless, the mapping capability of this model in other places with higher indexes of continentality is to be questioned. Furthermore, the possibility of achieving good ETo predictions in different inland locations to those used to train the network might be looked with uncertainty, because of the local uniqueness of the complex relationships between temperature and ETo. On the other hand, models trained in coastal locations might be preferable to carry out predictions in inland locations. The proposed 4-inputs ANN can be useful and preferable to other methods when ETo models which demand a high number of variables cannot be used. Keywords: ETo prediction, artificial neural networks, maximum and minimum temperatures, index of continentality. 1 Introduction The precise quantification of cropping evapotranspiration has become a very important task, due to the current water shortage and the subsequent rise of water price. On the other hand, the already existing models that provide precise enough
Keywords
ETo prediction, artificial neural networks, maximum and minimum temperatures, index of continentality.