Predicting of vegetation map based on geomorphological factors using artificial neural networks (case study: Sabzkouh, Chaharmahal-Va-Bakhtiari Province)

Authors

1 grajuted msc

2 - Assistant Professor, Department of Rangeland and watershed management, Faculty of Natural Resources and Geoscience, Shahrekord University

3 Assistant Professor, Department of Rangeland and watershed management, Faculty of Natural Resources and Geoscience, Shahrekord University

Abstract

Vegetation maps are considered as fundamental tool of managing vegetation. Several methods of vegetation mapping are suggested, each of one has founded on some basic information and so has different accuracy. One of this method called geomorphological method that the aim of this research is to investigate its accuracy. In this research, three maps of geomorphological facies, land-forms and lithology were considered as independent and vegetation map (physiognomic-floristic) was used as independent in a model of multi-layer perceptron neural-network to predict the pattern of vegetation distribution. the results showed that the model can explain 39.1% of the variation of vegetation map and land forms of the study area. This was not the case for all vegetation types and land forms so that Sparse forest (sparse oak forest) and Astragalus morinus -Acantholimon festucaceum - Acanthophyllum bracteatum with 64.4 and 61.5% had the highest whereas, agricultural lands and gardens, which only cover 0.3% of the study area, with 1.2% had the lowest prediction power discernibility respectively. Generally, it can be concluded that the method had relatively acceptable results in predicting vegetation distribution. However, moderate prediction accuracy of the model could be related to other affective factors (not considered in this study) such as soil and climate, accuracy of input variables, incompliance of the current situation with potential due to anthropological affects,low resolution because of overlapping of ecological niche of vegetation types and land cover. Therefore, we recommend that this method should be used for discrimination of relatively natural and intact vegetation type

Keywords