نوع مقاله : مقاله پژوهشی
نویسندگان
1 مرکز تحقیقات و آموزش کشاورزی و منابع طبیعی سیستان، سازمان تحقیقات، آموزش و ترویج کشاورزی
2 ، مرکز تحقیقات و آموزش کشاورزی و منابع طبیعی سیستان، سازمان تحقیقات، آموزش و ترویج کشاورزی
3 موسسه تحقیقات جنگلها و مراتع، سازمان تحقیقات، آموزش و ترویج کشاورزی
چکیده
کلیدواژهها
عنوان مقاله [English]
نویسندگان [English]
Introduction: Wind erosion is one of the most serious environmental issues in the world. The most important effects resulting from wind erosion include the formation of wind deposits, sparsity of vegetation cover, changes in soil texture, reduction of soil fertility, land degradation, and air pollution. Studying of wind erosion is one of the effective steps in the managing and control of this phenomenon. In this regard, the tools that have become importance for identifying and monitoring of wind erosion include remote sensing technology and artificial intelligence, which is designed based on numerous algorithms. The aim of this study is to identify and monitor areas susceptible and sensitive to wind erosion using Landsat satellite imagery, machine learning technology, and the Random Forest algorithm over the years 2013 to 2023 in the study area.
Materials and Methods: In this study, three remote sensing indices-including the Soil Adjusted Vegetation Index (SAVI), the Normalized Difference Moisture Index (NDMI), and the Land Surface Temperature (LST) index, were used for the identification of areas sensitive to wind erosion during the period from 2013 to 2023. In this regard, Landsat 8 satellite images and OLI sensor data from the month of June were used. For the identify and monitor areas susceptible to wind erosion, the machine learning method and the Random Forest (RF) algorithm were utilized. The wind erosion assessment resulting from the Random Forest algorithm was classified into four classes: low, moderate, severe, and very severe. Additionally, the Random Forest algorithm also showed the relative importance of input indices for identifying areas susceptible to wind erosion during the 2013-2023 period. The accuracy and validation of the Random Forest algorithm model were determined using statistical indices: coefficient of determination (R²), Root Mean Squared Error (RMSE), Mean Squared Error (MSE), and Area Under the ROC Curve (AUC).
Results and Discussion: The results of the wind erosion assessment in the 2013-2023 period showed that the SAVI index increased from 2013, with maximum values of 0.78 and minimum values of -0.41, to 2018, with maximum values of 0.79 and minimum values of -0.09. From 2018 to 2022, the amount of vegetation cover showed a significant decrease, with maximum values of 0.00015 and minimum values of -0.00005. The NDMI index from 2013 with a maximum value of 0.91 and a minimum value of -1, to 2016 with a maximum value of 0.51 and a minimum value of -0.99, indicates a decreasing trend in surface soil moisture. From 2016 to 2018, with a maximum value of 1 and a minimum of -0.77, it indicated an increasing trend in surface soil moisture. Also, from 2018 to 2023, with maximum values of 0.84 and a minimum of -1, it indicates a decreasing trend in soil surface moisture. The LST (Land Surface Temperature) index also fluctuated between 2013 and 2023, based on vegetation cover and soil surface moisture indices. The most critical status in terms of wind erosion was estimated to be in the year 2022, with an area of 3,530,221 hectares (99.86%) classified as very severe. The validation results of the Random Forest algorithm for identifying areas susceptible to wind erosion in the study area during the years 2013 to 2023 showed that the Correlation coefficients (R2) were estimated to be between 0.4 and 0.87, the Root Mean Square Error (RMSE) between 0.022 and 0.069, the Mean Square Error (MSE) between 0 and 0.0048, and the Area Under the ROC Curve (AUC) value to be greater than 0.923. The results indicate the high efficiency of the Random Forest algorithm for identifying areas sensitive to wind erosion in the study area. The SAVI index was found to be one of the most effective indices examined for wind erosion in the study area.
Conclusion: The results indicate that the year 2022 was found to be the most critical year in terms of the wind erosion phenomenon, classified as very severe. The influential factors on wind erosion during the years 2013 to 2023 are the status of vegetation cover and surface soil moisture. In this research, the information obtained from the monitoring and identification of erosion-susceptible areas can be utilized for project planning, with the aim of managing and controlling wind erosion in the study area.
کلیدواژهها [English]
)