Performance Evaluation of Artificial Neural Network Models for Downscaling and Predicting of Climate Variables

Abstract

Atmosphere–ocean coupled global climate models (GCMs) are the main source to simulate the climate of the earth climate. The computational grid of the GCMs is coarse and so, they are unable to provide  reliable  information  for hydrological   modelling.  To  eliminate such limitations, the downscaling methods are used. The present study is focused on simulating the impact  of  climate  change  on  the  behavior  of  precipitation  and  temperature  of  Sirjan  synoptic station in Kerman Province. At first, the capability of artificial neural network to downscaling of climate  variables  that  predicted  by  CanESM is  tested. Then, using  the  most  appropriate models, the  mean  monthly  temperature  and  precipitation  amounts  forecast  for  future periods under RCP scenario. Results of this study for monthly temperature downscaling indicated that the artificial neural network with   hidden layer,   neurons, with Tangent and Log sigmoid activation function was the best model, so that RMSE, NS and R were, and respectively.  Also,  for  precipitation  variable,  the  structure  with hidden  layer  feed  forward perceptron,   neurons, Tangent and Log sigmoid activation function and Levenberg-Marquardt algorithm  had  better  performance,  so  that  RMSE,  NS  and  R were, and, respectively. Results indicate that until , amount of monthly mean temperature under RCP emission  scenario  will  be  increased  by (˙C)  and  the  highest  increase  is  predicted  for August  by (˙C)  and  a  lower  increase  in  April by(˙C).  The  results  also  showed considerable increase of precipitation for June to November and noticeable decrease for March and May months. However, no change occure in annaul scale (inter-annual).