Preparation of waterway erosion sensitivity map using data mining integrated entropy-informational value model (IOE-IV) (Case Study: The Upper Watershed of Tajan River)
Document Type : Original Article
Authors
1 PhD graduated student of Watershed Management Sciences and Engineering, Water, Faculty of Natural Resources and Earth Sciences, University of Kashan
2 Department of Nature Engineering, Faculty of Natural Resources and Earth Sciences, University of Kashan, Kashan, Isfahan. Iran
3 PhD Student of Watershed Management Sciences and Engineering, Water and Soil Conservation, Faculty of Natural Resources and Earth Sciences, University of Kashan
10.22034/gmpj.2023.419035.1456
Abstract
Introduction
Soil is one of the most vital issues for the sustainability of the environment, which provides human needs and livelihood on the surface of the earth. Therefore, reducing the phenomenon of land degradation is one of the leading challenges for the sustainable development of the environment and economic activities, which is why comprehensive planning and management against erosion is essential. Today, several methods have been developed for preparing soil erosion maps, which can be mentioned using experimental methods; that these methods include equal weights for all parameters in calculating the average erosion; While each criterion has a point value that is affected by environmental, geomorphological and physical factors related to soil erosion. Among other methods, we can mention machine learning, which is an advanced method that requires high-performance computing systems. However, it is necessary to use statistical methods for fast, understandable and accurate modeling, and there is no requirement for high-performance systems in these methods and Among these methods, Entropy Index, Frequency Ratio, Information Value, Certainty Factor, and Witness Weight methods can be mentioned. Therefore, the aim of the current research is to identify the most important criteria and sub-criteria that are effective in preparing the erosion sensitivity map of the watershed upstream of the Tajen River using the Integration of two data mining models Entropy - Information Value.
Methodology
The erosion inventory map with 252 erosion points for the upper watershed of Tejn River was identified using Google Earth images, and 70% of these erosion points were classified for training and validation of the model. Then, to evaluate the sensitivity of erosion and also to identify the most important driving factors in the occurrence of erosion, the criteria of elevation, slope direction, soil type, land use, Rainfall Erosivity, distance to stream and Topographic Wetness Index were used and The erosion sensitivity map of the studied area was obtained from the integrated entropy-information value model. In order to evaluate the success rate and predict the model, Receiver Operating Characteristics and Area Under the Curve were used.
Results and Discussion
The results of the present study showed that the effect of elevation on soil erosion is increasing, so that this effect reaches its maximum value in the elevation class of 2745-3742 meters with an Information Value of 1.65. The influence of the slope directions on soil erosion showed that the south-west slope direction with an Information Value of 1.04 has the greatest effect on the erosion of the region. The results of overlapping the distance to stream layer with the erosion map of the region indicate that the distance layer of 0-456 meters has the highest Information Value of 0.63. The barren land use class with an Information Value of 2.74 has the greatest effect on the erosion potential of the region. The results of the overlap between the Rainfall Erosivity map and the regional erosion map showed that the 302-325 layer of rain erosion has the highest Information Value of 1.28. Examining the relationship between types of soil groups with regional erosion showed that the Inceptisols soil group with an Information Value of 0.66 has the greatest effect on soil erosion in the region. The results of the overlap between the Topographic Wetness Index layer and soil erosion in the region indicated that the 6.2-8.6 layer of the Topographic Wetness Index with an information value of 0.25 has the highest erosion potential. According to the curve (ROC-AUC), the success and prediction rate of the combined model of Entropy Index - Information Value was obtained as 0.831 and 0.837, respectively which indicates the good performance of the model for training and validation courses.
Conclusion
Soil erosion is one of the most important issues of land destruction on a global scale, which causes muddying of waters and lakes, loss of fertile agricultural topsoil and biodiversity; Therefore, it is essential to evaluate the sensitivity of soil erosion. The current research was conducted with the aim of identifying the most important criteria and sub-criteria effective in soil erosion with the integrated model of Entropy - Information Value in the upper watershed of Tajen River. The results of the present study showed that out of the total area of 693.23 square kilometers of the study area, 156.75 and 94.71 square kilometers are in the high and very high categories, respectively. The prepared soil erosion sensitivity map in the studied area can be a useful tool for management and planning in line with soil protection measures.
Keywords: Tajen River, data mining model, ROC curve ,waterway erosion sensitivity
Soil is one of the most vital issues for the sustainability of the environment, which provides human needs and livelihood on the surface of the earth. Therefore, reducing the phenomenon of land degradation is one of the leading challenges for the sustainable development of the environment and economic activities, which is why comprehensive planning and management against erosion is essential. Today, several methods have been developed for preparing soil erosion maps, which can be mentioned using experimental methods; that these methods include equal weights for all parameters in calculating the average erosion; While each criterion has a point value that is affected by environmental, geomorphological and physical factors related to soil erosion. Among other methods, we can mention machine learning, which is an advanced method that requires high-performance computing systems. However, it is necessary to use statistical methods for fast, understandable and accurate modeling, and there is no requirement for high-performance systems in these methods and Among these methods, Entropy Index, Frequency Ratio, Information Value, Certainty Factor, and Witness Weight methods can be mentioned. Therefore, the aim of the current research is to identify the most important criteria and sub-criteria that are effective in preparing the erosion sensitivity map of the watershed upstream of the Tajen River using the Integration of two data mining models Entropy - Information Value.
Methodology
The erosion inventory map with 252 erosion points for the upper watershed of Tejn River was identified using Google Earth images, and 70% of these erosion points were classified for training and validation of the model. Then, to evaluate the sensitivity of erosion and also to identify the most important driving factors in the occurrence of erosion, the criteria of elevation, slope direction, soil type, land use, Rainfall Erosivity, distance to stream and Topographic Wetness Index were used and The erosion sensitivity map of the studied area was obtained from the integrated entropy-information value model. In order to evaluate the success rate and predict the model, Receiver Operating Characteristics and Area Under the Curve were used.
Results and Discussion
The results of the present study showed that the effect of elevation on soil erosion is increasing, so that this effect reaches its maximum value in the elevation class of 2745-3742 meters with an Information Value of 1.65. The influence of the slope directions on soil erosion showed that the south-west slope direction with an Information Value of 1.04 has the greatest effect on the erosion of the region. The results of overlapping the distance to stream layer with the erosion map of the region indicate that the distance layer of 0-456 meters has the highest Information Value of 0.63. The barren land use class with an Information Value of 2.74 has the greatest effect on the erosion potential of the region. The results of the overlap between the Rainfall Erosivity map and the regional erosion map showed that the 302-325 layer of rain erosion has the highest Information Value of 1.28. Examining the relationship between types of soil groups with regional erosion showed that the Inceptisols soil group with an Information Value of 0.66 has the greatest effect on soil erosion in the region. The results of the overlap between the Topographic Wetness Index layer and soil erosion in the region indicated that the 6.2-8.6 layer of the Topographic Wetness Index with an information value of 0.25 has the highest erosion potential. According to the curve (ROC-AUC), the success and prediction rate of the combined model of Entropy Index - Information Value was obtained as 0.831 and 0.837, respectively which indicates the good performance of the model for training and validation courses.
Conclusion
Soil erosion is one of the most important issues of land destruction on a global scale, which causes muddying of waters and lakes, loss of fertile agricultural topsoil and biodiversity; Therefore, it is essential to evaluate the sensitivity of soil erosion. The current research was conducted with the aim of identifying the most important criteria and sub-criteria effective in soil erosion with the integrated model of Entropy - Information Value in the upper watershed of Tajen River. The results of the present study showed that out of the total area of 693.23 square kilometers of the study area, 156.75 and 94.71 square kilometers are in the high and very high categories, respectively. The prepared soil erosion sensitivity map in the studied area can be a useful tool for management and planning in line with soil protection measures.
Keywords: Tajen River, data mining model, ROC curve ,waterway erosion sensitivity
Keywords
)