Abstract:Landslides, resulting from disturbances in slope equilibrium, pose a significant threat to landscapes, infrastructure, and human life. Triggered by factors such as intense precipitation, seismic activities, or volcanic eruptions, these events can cause extensive damage and endanger nearby communities. A comprehensive understanding of landslide characteristics, including spatio-temporal patterns, dimensions, and morphology, is vital for effective landslide disaster management. Existing remote sensing approaches mostly use either optical or synthetic aperture radar sensors. Integrating information from both these types of sensors promises greater accuracy for identifying and locating landslides. This study proposes a novel approach, the ML-LaDeCORsat (Machine Learning-based coseismic Landslide Detection using Combined Optical and Radar Satellite Imagery), that integrates freely available Sentinel-1, Palsar-2, and Sentinel-2 imagery data in Google Earth Engine (GEE). The approach also integrates relevant spectral indices and suitable bands used in a machine learning-based classification of coseismic landslides. The approach includes a robust and reproducible training and validation strategy and allows one to choose between five classifiers (CART, Random Forest, GTB, SVM, and Naive Bayes). Using landslides from four different earthquake case studies, we demonstrate the superiority of our approach over existing solutions in coseismic landslide identification and localization, providing a GTB-based detection accuracy of 87–92%. ML-LaDeCORsat can be adapted to other landslide events (GEE script is provided). Transfer learning experiments proved that our model can be applied to other coseismic landslide events without the need for additional training data. Our novel approach therefore facilitates quick and reliable identification of coseismic landslides, highlighting its potential to contribute towards more effective disaster management.

What problem does this paper attempt to address?

The main problem this paper attempts to address is the more accurate detection of coseismic landslides by combining optical and radar satellite imagery with machine learning algorithms on the Google Earth Engine (GEE) platform. Specifically, the paper proposes a new method called ML-LaDeCORsat, which integrates satellite data from Sentinel-1, Palsar-2, and Sentinel-2, and uses machine learning classifiers to improve the accuracy and reliability of landslide detection. ### Main Research Questions: 1. **How to improve the accuracy of landslide detection using multi-sensor data?** - Explore whether the combination of optical and radar satellite imagery can improve the accuracy of landslide detection. 2. **What are the performance and processing speed of different machine learning classifiers in landslide detection?** - Compare the performance of different machine learning classifiers available in GEE (such as CART, Random Forest, Gradient Boosting Trees, Support Vector Machine, and Naive Bayes) in landslide detection. 3. **What is the effectiveness of TOA and SR products in early landslide detection?** - Investigate the advantages and disadvantages of using TOA (Top of Atmosphere) radiance products versus SR (Surface Reflectance) products in landslide detection. 4. **Which spectral and terrain features are most important for landslide detection?** - Analyze the importance of different spectral bands and terrain features in landslide detection. 5. **What other factors affect the accuracy of landslide detection based on satellite imagery?** - Explore how factors other than spectral and terrain features affect the accuracy of landslide detection. 6. **How to achieve automation of landslide detection based on GEE?** - Study how to fully automate the landslide detection method to facilitate adjustment and application in any spatiotemporal scenario. ### Research Background: Landslides are a severe natural disaster triggered by various factors such as heavy rainfall, earthquakes, and volcanic eruptions. Accurate and timely detection of landslides is crucial for disaster management and rescue. Existing remote sensing methods mainly rely on a single type of sensor (optical or radar), while the combination of multi-sensor data is expected to improve detection accuracy. The paper validates the effectiveness of the new method through case studies of four different earthquake events. ### Main Contributions: 1. **Detailed comparison of the performance of existing landslide detection methods**: - Evaluate the accuracy of existing landslide detection methods using ground truth datasets from four different case locations. 2. **Propose a new landslide detection solution**: - Utilize multi-band optical and radar imagery (10-meter spatial resolution), including S1, S2, P2, and terrain-derived bands. - Implement a landslide-specific training and validation sampling strategy based on a novel slope mask method. - Apply machine learning classifiers with optimized parameters to improve performance and processing speed. - Introduce new pseudo-bands (such as slope curvature, aspect, P2 SAR bands, S1 SAR band combination VH-VV). 3. **Investigate the importance of each landslide condition band in the ML model**: - Analyze the impact of each band on landslide detection through experiments. 4. **Comprehensive comparison of ML classifiers in GEE**: - Conduct a comprehensive evaluation of the performance of different machine learning classifiers in earthquake-induced landslide detection. 5. **Transfer learning across geographic regions**: - Study the applicability and effectiveness of transfer learning in landslide detection across different geographic regions. Through these studies, the paper provides significant theoretical and technical support for remote sensing-based landslide detection, particularly in the application of multi-sensor data fusion and machine learning algorithms.

Detecting Coseismic Landslides in GEE Using Machine Learning Algorithms on Combined Optical and Radar Imagery

Machine learning powered high-resolution co-seismic landslide detection

A New Deep-Learning-Based Approach for Earthquake-Triggered Landslide Detection From Single-Temporal RapidEye Satellite Imagery

Generating landslide density heatmaps for rapid detection using open-access satellite radar data in Google Earth Engine

Rapid landslide identification using synthetic aperture radar amplitude change detection on the Google Earth Engine

Automatic recognition of active landslides by surface deformation and deep learning

Automatic detection of earthquake triggered landslides using Sentinel-1 SAR imagery based on deep learning

Deep Learning for Rapid Landslide Detection using Synthetic Aperture Radar (SAR) Datacubes

Globally vs. Locally Trained Machine Learning Models for Landslide Detection: A Case Study of a Glacial Landscape

A novel framework for landslide displacement prediction using MT-InSAR and machine learning techniques

Landslide Hazard Identification Based on Deep Learning and Sentinel-2 Remote Sensing Imagery

Dynamic Earthquake-Induced Landslide Susceptibility Assessment Model: Integrating Machine Learning and Remote Sensing

Detection of slow‐moving landslides through automated monitoring of surface deformation using Sentinel‐2 satellite imagery

A Deep-Learning-Based Algorithm for Landslide Detection over Wide Areas Using InSAR Images Considering Topographic Features

ML-CASCADE: A machine learning and cloud computing-based tool for rapid and automated mapping of landslides using earth observation data

Enhancing Seismic Landslide Susceptibility Analysis for Sustainable Disaster Risk Management through Machine Learning

Landslide Susceptibility Evaluation Based on Potential Disaster Identification and Ensemble Learning

Automatic Detection of Forested Landslides: A Case Study in Jiuzhaigou County, China

Global Dynamic Rainfall-Induced Landslide Susceptibility Mapping Using Machine Learning

Change Detection-Based Co-Seismic Landslide Mapping Through Extended Morphological Profiles and Ensemble Strategy

Landslide Identification and Gradation Method Based on Statistical Analysis and Spatial Cluster Analysis