Classification of Effective Soil Depth by Using Multinomial Logistic Regression Analysis

• Main Authors: Chien-Hui Chang, 張建輝
• Other Authors: Hsun-Chuan Chan
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/53401667878313499735

碩士 === 國立中興大學 === 水土保持學系所 === 105 === It is known that the slopeland disaster is usually related to the land development, improper land use and torrential rain. The soil depth is an important factor in the shallow landslide. The government of Taiwan also uses soil depth as a factor of slopeland utilizable limitation to regulate the development and conservation of a slopeland. That all shows the soil depth play an important role in shallow landslide and land management. This research aimed to classify the effective soil depth by using multinomial logistic regression with the environmental factors. The upper watershed of the Houlong River located at the central Taiwan was selected as the study areas. The analysis of multinomial logistic regression is performed by the assistance of a Geographic Information Systems (GIS). The effective soil depth was categorized into four levels including deeper, deep, shallow and shallower. The environmental factors of slope, aspect, digital elevation model (DEM), curvature and normalized difference vegetation index (NDVI) were selected for classifying the soil depth. Error Matrix and Kappa index were then used to assess the model accuracy. In the modeling group, the overall accuracy was 76.6% and Kappa index was. 0.65. In the validation group, the overall accuracy was 70.5% and Kappa index was. 0.57. Then, the results of model were compared with ordinary kriging, regression kriging and TWI method. In the ordinary kriging method, the overall accuracy was 45.7% and Kappa index was. 0.15. In the regression kriging method, the overall accuracy was 46.7% and Kappa index was. 0.16. In the TWI method, the overall accuracy was 30.5% and Kappa index was. -0.05. As the results, the multinomial logistic regression method is more accurate on classifying the effective soil depth among the investigated methods in the study areas.