Point estimation of soil moisture characteristic curve using artificial neural networks and its optimizing by genetic algorithm in Agro-Industries of Khouzestan

Document Type : Research

Authors

1 Master's student in soil science, Ramin Khuzestan University of Agriculture and Natural Resources

2 Associate Professor, Department of Soil Science, Ramin Khuzestan University of Agriculture and Natural Resources

3 Assistant Professor, Department of Soil Science, Ramin Khuzestan University of Agriculture and Natural Resources

4 Assistant Professor of Soil Science Department of Shahrekord University

10.22092/wmej.2016.112501

Abstract

Soil hydraulic properties have key role in sugar cane cultivation management. The purpose of this study is to estimate soil moisture characteristic curve using an artificial neural network and its optimization with genetic algorithm. Therefore, based on the cultivation operations management and soil properties included: organic matter content, soil texture, electrical conductivity, sodium adsorption ratio, 4 land unit tracts in Debel-Khozaii, Amir-Kabir, Karoon and Haft-Tapeh agro-industries were selected. A total of 310 soil samples from both 0-40 and 40-80 cm of soil profiles were collected. In this study, five models were arranged in hierarchy to estimate soil hydraulic properties with ANNs. The performances of the models were evaluated using Spearman's correlation coefficient (r) between the observed and the estimated values, normalized mean square error (NMSE), and mean absolute error (MAE). Owing to the fact that the selection of each of the variable parameters of neural network necessitated recurring trails and errors, and consequently teaching a large number of networks with various topologies, genetic algorithm method was utilized for finding the optimization of these parameters and the efficiency of this method was examined in terms of the optimization of neural network. Results showed that the neural network has a high degree of accuracy in modeling and estimating soil moisture characteristic curve (R =0.943, MAE=0.019, NMSE=0.054). Also, combining artificial neural networks with genetic algorithm for optimizing the conditions of the artificial neural networks implementation was positive and combining approach indicated its superiority over non-optimized implementation of artificial neural networks in all cases (R=0.985, MAE=0.01, NMSE =0.0151).

Keywords

Watershed Management Research
Volume 29, Issue 3 - Serial Number 112
October 2016
Pages 101-112

Files

History

  • Receive Date: 22 December 2012
  • Revise Date: 19 April 2013
  • Accept Date: 21 May 2013

Share

How to cite

Statistics