Comparing efficiency of Iranian desert hazard assessment models, namely MICD and IMDPA (Case study: Sabzevar region)

There have been several models presented for evaluation and provision of the prevailing conditions of desertification. Two Iranian desertification models, namely MICD and IMDPA, were used to assess desertification hazard and select the best model for Sabzevar region. Firstly, work unit map (geomorphologic facies) was created using slope, geology, vegetation cover, land use, Landsat 5 satellite imagery and google earth data. Once scoring was at hand indicators, the relative hazard maps were produced. According to field observatories and incorporating advices from experienced connoisseurs of this field, ground truth map was prepared. In the next step, efficiency of the models was evaluated. In order to ascertain the level of compatibility between the hazard assessment maps produced by means of the MICD and IMDPA models and what was learned in the field, the Kappa index came to help. Differences between the frequencies of classes defined in the maps were investigated by applying the Chi-square test and along this, the most efficient model was chosen based on Kappa index as well as Spearman and Kendall’s correlations indices. Results from statistical comparison made between hazard assessment classes defined in the models and the ground truth showed significant differences (Pvalue<0.01). The matchup between the out results of models and the ground truth was evaluated as average to low extents. On the other hand, drawing an analogy between the Spearman and Kendall’s Correlation indices outcomes in each model and the ground truth showed comparable results in which the Kappa index and Chi square index was included, and this led to this conclusion that the MICD performs well above the IMDPA. Finally, the MICD was chosen as the efficient model.