نوع مقاله : مقاله پژوهشی
نویسندگان
1 گروه علوم و مهندسی خاک، دانشکده کشاورزی، دانشگاه شهید چمران اهواز، اهواز، ایران
2 گروه جغرافیای انسانی، دانشکده علوم جغرافیایی و برنامه‌ریزی، دانشگاه اصفهان، اصفهان، ایران
چکیده
کلیدواژه‌ها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
Desertification is a form of land degradation resulting from both natural processes and human activities. In this regard, the use of remote sensing indicators to prepare desertification maps can be an efficient approach. The goal of this study is to evaluate the potential of soil spectral indices and surface albedo in modeling desertification intensity in the western Khuzestan Province. Initially, we extracted indices including BSI, BI, SI, IFe2O3, NSMI, TGSI, CLI, CI, SCI, and Albedo from Landsat 8 satellite images. In this research, soil spectral indices were calculated using the band calculation tool available in ENVI software, and raster maps of these indices were generated. To create feature space models and identify correlations between the variables, we extracted the surface albedo index values from the satellite images.
The maximum and minimum values of this index were 37518.1 and 8338.75, respectively. Next, we randomly selected spectral values from 753 locations on the soil spectral index map and extracted their corresponding values from the surface albedo index map. To ensure comparability, the index values were standardized through data normalization. After normalization, scatter plots of the spectral pixel value densities were generated in the SAGA 9.5 software environment, and best-fit line equations were determined. Linear regression analysis was then conducted to examine the correlation between the soil spectral indices and Albedo.
In this study, all soil spectral indices were considered independent variables, while the surface albedo index was treated as the dependent variable. Desertification intensity was classified by dividing the feature space perpendicular to the trend of desertification change. Subsequently, the Jenks Natural Breaks method was applied to classify the data values into five desertification classes: areas without desertification, areas with low desertification intensity, areas with moderate desertification intensity, areas with high desertification intensity, and areas with very high desertification intensity.
The results of the correlation analysis of the indices showed that the BSI, BI, SI, NSMI, TGSI, CLI, and SCI indices were positively correlated with the Albedo index; so with the increase in the BSI, BI, SI, NSMI, TGSI, CLI, and SCI indices, the Albedo variable increases. While the IFe2O3 and CI indices were negatively correlated with the Albedo index, such that an increase in IFe2O3 and CI indices resulted in a reduction of the Albedo variable. An analysis of the correlation coefficients between the indices revealed that the BI, SI, and TGSI indices exhibited the strongest correlation with the Albedo index, with correlation coefficients of 0.985, 0.909, and 0.850, respectively. Conversely, the IFe2O3, CLI, and NSMI indices showed the weakest correlation with the Albedo index, with correlation coefficients of -0.026, 0.106, and 0.110, respectively. Therefore, the Albedo-BI, Albedo-SI, and Albedo-TGSI models, due to their highest correlation with the variables, were considered a suitable criterion for evaluating and classifying desertification in the region, given its arid climate.
Following the classification of desertification intensity, the accuracy of the Albedo-BI, Albedo-SI, and Albedo-TGSI feature space models was assessed using the error matrix, overall accuracy, and the kappa coefficient. The Albedo-BI model achieved an overall accuracy of 98.23% and a kappa coefficient of 0.97; the Albedo-SI model yielded 94.45% accuracy and a kappa value of 0.92; and the Albedo-TGSI model recorded 90.50% accuracy with a kappa coefficient of 0.86. These results indicate that the Albedo-BI model provides the highest classification accuracy among the three models. Furthermore, analysis of the models suggests that the eastern parts of the study area exhibit higher desertification intensity than the western regions.
کلیدواژه‌ها [English]
)