Abstract:
This study presents a new methodology to identify landslide and landslide susceptible locations in interior Alaska using only geomorphic properties from light detection and ranging (LiDAR) derivatives (i.e., slope, profile curvature, roughness) and the normalized difference vegetation index (NDVI). The study specifically focused on the effect of different resolutions of LiDAR images in identifying landslide locations. I developed a semi-automated object-oriented image classification approach in ArcGIS 10.5, and prepared a landslide inventory from visual observation of hillshade images. The multistage workflow included combining derivatives from 1m, 2.5m, and 5m resolution LiDAR, image segmentation, image classification using a support vector machine classifier, and image generalization to clean false positives. I assessed the accuracy of the classifications by generating confusion matrix tables. Analysis of the results indicated that the scale of LiDAR images played an important role in the classification, and the use of NDVI generated better results in identifying landslide and landslide susceptible places. Overall, the LiDAR 5m resolution image with NDVI generated the best results with a kappa value of 0.55 and an overall accuracy of 83%. The LiDAR 1m resolution image with NDVI generated the highest producer accuracy of 73% in identifying landslide locations. I produced a combined overlay map by summing the individual classified maps, which was able to delineate landslide objects better than the individual maps. The combined classified map from 1m, 2.5m, and 5m resolution LiDAR with NDVI generated producer accuracies of 60%, 80%, 86%, and user accuracies of 39%, 51%, 98% for landslide, landslide susceptible, and stable locations, respectively, with an overall accuracy of 84% and a kappa value of 0.58. The proposed method can be improved by fine-tuning segmented image generation, incorporating other data sets, and developing a standard accuracy assessment technique for object-oriented image analysis.
