Landslide Susceptibility Prediction Modeling Based on Remote Sensing and a Novel Deep Learning Algorithm of a Cascade-Parallel Recurrent Neural Network.

Li Zhu,Jiawu Chen,Lianghao Huang,Jinsong Huang,Linyu Fan,Zihe Zhang,Yuhao Wang,Faming Huang

doi:10.3390/s20061576

Li Zhu, Jiawu Chen + Show 6 more

Open Access

https://doi.org/10.3390/s20061576

Copy DOI

Journal: Sensors (Basel, Switzerland)	Publication Date: Mar 12, 2020
Citations: 72	License type: CC BY 4.0

Affiliation: Nanchang University

Abstract

Landslide susceptibility prediction (LSP) modeling is an important and challenging problem. Landslide features are generally uncorrelated or nonlinearly correlated, resulting in limited LSP performance when leveraging conventional machine learning models. In this study, a deep-learning-based model using the long short-term memory (LSTM) recurrent neural network and conditional random field (CRF) in cascade-parallel form was proposed for making LSPs based on remote sensing (RS) images and a geographic information system (GIS). The RS images are the main data sources of landslide-related environmental factors, and a GIS is used to analyze, store, and display spatial big data. The cascade-parallel LSTM-CRF consists of frequency ratio values of environmental factors in the input layers, cascade-parallel LSTM for feature extraction in the hidden layers, and cascade-parallel full connection for classification and CRF for landslide/non-landslide state modeling in the output layers. The cascade-parallel form of LSTM can extract features from different layers and merge them into concrete features. The CRF is used to calculate the energy relationship between two grid points, and the extracted features are further smoothed and optimized. As a case study, the cascade-parallel LSTM-CRF was applied to Shicheng County of Jiangxi Province in China. A total of 2709 landslide grid cells were recorded and 2709 non-landslide grid cells were randomly selected from the study area. The results show that, compared with existing main traditional machine learning algorithms, such as multilayer perception, logistic regression, and decision tree, the proposed cascade-parallel LSTM-CRF had a higher landslide prediction rate (positive predictive rate: 72.44%, negative predictive rate: 80%, total predictive rate: 75.67%). In conclusion, the proposed cascade-parallel LSTM-CRF is a novel data-driven deep learning model that overcomes the limitations of traditional machine learning algorithms and achieves promising results for making LSPs.

Highlights

Landslides are one of the most common geological disasters worldwide and cause considerable damage to public infrastructure and human life every year [1,2]
Based on studies between landslide and environmental factors conducted by researchers in different regions, the environmental factors consist of five main categories: topography and geomorphology, hydrologic environment, basic geology, land cover, and human activities [70]
The landslide susceptibility prediction (LSP) of Shicheng County was implemented using each of the cascade-parallel long short-term memory (LSTM)-conditional random field (CRF), logistic regression (LR), It is necessary to prepare the training dataset and the test dataset for model training and testing

Summary

Introduction

Landslides are one of the most common geological disasters worldwide and cause considerable damage to public infrastructure and human life every year [1,2]. The landslide susceptibility model based on geological environmental conditions can supply the government with an important theoretical basis for land resource planning and disaster prevention and reduction. The process of landslide susceptibility prediction (LSP) modeling primarily includes a catalog of landslides, environmental factors extraction, model architecture construction, model training, landslide susceptibility mapping, and model evaluation [6,7]. The catalog of landslides (landslide area, boundary, locations) are measured using global positioning systems and put into a geographic information system (GIS) for landslide storage and management [8,9]. The environmental factors are extracted from the remote sensing (RS) images, such as Landsat TM image, digital elevation model (DEM), aerial imagery, and LiDAR, based on the GIS spatial analysis, including terrain analysis, hydrological analysis, and map algebra [10]. The LSP modeling is built on the platform of GIS because of the spatial big data analysis, storage, mapping, and management abilities [11]

Objectives

Methods

Results

Discussion

Conclusion