Evaluation of remote sensing data to predict soil organic carbon using statistical and artificial neural network methods

Document Type : Original Article

Authors

1 Department of Environment Science, Faculty of Natural Resources, University of Kurdistan, Sanandaj, Iran

2 Environmental Scientist, ESG Environmental, Melbourne, Australia

Abstract

The aim of this study was to evaluate the remote sensing data and evaluate the capability of artificial neural network in estimating soil organic carbon in Abidar and Toos Nozar forest parks in Sanandaj. Soil samples were prepared from 120 points at a depth of 0-30 cm and soil organic carbon was determined by walk-block method. Remote sensing data set was performed based on two statistically significant methods of correlation coefficient and stepwise linear regression. The artificial neural network MLP was used to estimate soil organic carbon. The results indicate that using the full potential of the electromagnetic spectrum can be effective in improving the accuracy of estimating soil organic carbon. lowest error rate in the training phase (0.001) was related to the stepwise linear regression method and the highest error rate (0.036) was related to the fixed number of input parameters. the artificial neural network MLP showed that it has a high ability to extend the experimental data to other areas.

Keywords