Sensitivity Analysis of Desertification Status to the Input Parameters

Document Type : Research Article

Authors

M. Akbari ¹

E. Ranaee ²

S.H. Badiee ³

¹ Department of Dry Lands and Desert Management, Faculty of Natural Resources and Environment, Ferdowsi University of Mashhad

² Millan University of Tecknology

³ Natural Resources and Watershed Head Office of Khorasan Razavi Province

https://doi.org/10.22067/jsw.v0i0.9486

Abstract

Abstract
Desertification could cause reduction of the ecological and biological potential of land which may occur both naturally and artificially. Identifying and evaluating the effective factors in development of desertification is very important for better management of land. The aim of this research is to evaluate the sensitive input parameters in the desertification condition by using of Artificial Neural Networks. The study area with 118658 hectars is located in south of Neishabour Township. During the past years, this area has been faced to increase in desertification rate due to some long consecutive periodic droughts, destruction of vegetation, converting of pasture lands to dry farms, water and wind erosion and unsuitable management of land use. After field studies preparation of aerial photos and satellite image, we prepared and analyzed the required layers in Geographic Information System. FAO-UNEP method (1984) was used for assess the Desertification rate. In this study vegetation condition, pasture condition, water and wind erosion and salinity has been defined and categorized as the factors in the status of desertification. After introducing the information to GIS, based on FAO-UNEP approach, the effective criteria were studied and map of desertification condition was achieved. The results showed that the desertification in northern parts of this area was serve which originated from the reduction of canopy, destruction of vegetation and water erosion and 62 and 30 percents of the whole studied area could be classified in moderate and slight conditions, respectively. Furthermore, to compare the results and quantifying the weight of input parameters, a mathematical model of artificial neural networks was used. The result showed that the effect of vegetation, wind erosion and water erosion could not be ignored and should be seriously considered. But salinity parameter is less effective than other factors in desertification and confirmed with FAO-UNEP method output. Analysis of different error criteria especially Mean Square Error with the value of 0.25 confirmed the accuracy of results.

Keywords: Desertification, FAO-UNEP, GIS, Artificial Neural Network, Neishabour Township

20.1001.1.20084757.1390.25.2.20.7

---