SRM INSTITUTE OF SCIENCE AND TECHNOLOGY

SCHOOL OF COMPUTING
DEPARTMENT OF COMPUTING TECHNOLOGIES
18CSP109L - MAJOR PROJECT

Landslide inventory
mapping and post seismic
landslide risk reduction
model
Guide name: Dr. Snehasish Batch ID:
Ghosh
Designation: Assistance Student 1 Reg. No: RA2011003011221
Student 1 Name: Kshitij Singh
Professor Department: CTech

Student 2 Reg. No: RA2011003011206

Student 2 Name: Priyanshu Singh
 Landslide inventory mapping
and post seismic landslide risk
reduction model
Abstract
• Develop a comprehensive approach integrating satellite imagery, topographic data, and
meteorological information for accurate landslide detection.

• Utilize feature extraction and selection techniques to process multi-sensor data, followed
by training a machine learning model (e.g., support vector machine or convolutional
neural network) to predict landslide-prone areas.

• Implement a feedback loop for continuous model updates with real-time data, ensuring
the system's adaptability to changing environmental conditions.

• Design a user-friendly interface for easy implementation by authorities and emergency

responders, providing real-time visualization of landslide risk areas to facilitate proactive
decision-making and resource deployment.
 Landslide inventory mapping
and post seismic landslide risk
reduction model
Introduction
• Landslide inventory maps serve multiple purposes, encompassing
various scales from small to large watersheds, regions, states, and
nations..

• These maps play a crucial role as a preliminary step towards

assessing landslide susceptibility, hazard, and risk, contributing to
proactive risk management.

• Landslide inventory maps are essential for investigating the

distribution, types, and patterns of landslides concerning
morphological and geological characteristics.

• They contribute to the study of landscape evolution in areas

 Landslide inventory mapping
and post seismic landslide risk
reduction model
Motivation
• Purpose of Landslide Inventory Maps: Records location, date,
and types of mass movements in an area.Utilizes techniques
like aerial photographs, satellite images, LiDAR elevation data,
and GIS technology. Technique selection based on study's
purpose, area extent, base map scale, and available resources.

• Comprehensive Information:Inventory maps include location

and types of landslides.Other geomorphological features like
escarpments, alluvial fans, badlands, and alluvial deposits may
be portrayed.Enhances understanding of slope processes and
related phenomena.Contributes to a comprehensive
 Landslide inventory mapping
and post seismic landslide risk
reduction model
Literature Review
 Landslide inventory mapping
and post seismic landslide risk
reduction model
Challenges and Limitations
• Limited Labeled Data:Challenges arise due to the scarcity of
labeled data for training deep learning models specifically tailored
for landslide detection.

• Data Imbalance:Imbalances in the distribution of landslide and

non-landslide data can affect model performance.

• Computational Complexity:Deep learning models, particularly

convolutional neural networks (CNNs) used for image-based
detection, can be computationally demanding

• Generalization Challenges: Deep learning models trained on one

geographic region or under specific environmental conditions may
 Landslide inventory mapping and
post seismic landslide risk
reduction model
Objectives
• Accuracy Enhancement: Implement advanced machine learning
algorithms to improve the accuracy and reliability of landslide
detection, ensuring precise identification of landslide-prone areas.
• Multi-Source Data Integration: Integrate diverse datasets, such as
satellite imagery, topographic information, and meteorological data,
to create a comprehensive model that leverages multiple
information sources for more effective landslide detection.
• Real-Time Monitoring: Establish a system capable of real-time
monitoring and detection of landslide events, facilitating timely
response and proactive risk management strategies.
• Adaptability to Dynamic Conditions: Develop a model that can adapt
to changing environmental conditions ensuring continued
effectiveness in landslide detection under varying circumstances,
including seasonal changes and land use modifications.
 Landslide inventory mapping and
post seismic landslide risk
reduction model
Innovation idea of the Project
• Integrated Sensor Network:Deploy a network of advanced sensors,
including but not limited to IoT devices, ground-based sensors, and
satellite imagery, to continuously monitor environmental factors such as
rainfall, soil moisture, and topography.
• Machine Learning Algorithms:Utilize cutting-edge machine learning
algorithms, including deep neural networks, to analyze real-time data and
identify early indicators of potential landslide events. Train the model on
diverse datasets to enhance adaptability to various geographical and
climatic conditions.

• Predictive Analytics:Implement predictive analytics to forecast landslide

risks based on the continuous monitoring of environmental parameters.
The system should provide accurate predictions with varying time
horizons, from short-term alerts to long-term risk assessments.

• Dynamic Risk Mapping:Develop a dynamic risk mapping system that

visualizes the evolving landslide risks in real-time. The mapping should be
accessible through a user-friendly interface, allowing authorities,
emergency responders, and the public to make informed decisions.
 Landslide inventory mapping and
post seismic landslide risk
reduction model
Scope and applications of the Project
• Geographical Coverage: The scope of this project encompasses
diverse geographical regions prone to landslides, ranging from
small-scale localities to large watersheds and potentially
expanding to national or global scales.

• Data Types: The study includes the integration and analysis of

various data types, such as satellite imagery, topographic
information, meteorological data, and any other relevant
datasets contributing to the accurate detection of landslide-
prone areas.

• Environmental Conditions: The model's scope involves

adaptability to diverse environmental conditions, considering
 Landslide inventory mapping and
post seismic landslide risk
reduction model
• Proposed Architecture

Conceptualization of generating the

patches for training the models
 Landslide inventory mapping and
post seismic landslide risk
reduction model
Proposed System:
Data Integration and Preprocessing:
• Aggregate diverse data sources, including satellite imagery, topographic
data, and meteorological information, and preprocess the data for
uniformity.

Feature Extraction and Selection:

• Extract relevant features from preprocessed data to create a
comprehensive set of inputs for the machine learning model.

Machine Learning Model Development:

• Train a machine learning model capable of identifying patterns indicative
of landslide-prone areas.

Real-Time Monitoring and Adaptation:

• Enable continuous monitoring of environmental conditions and implement
a feedback loop to adapt the model in real-time.

Automation and Scalability:

• Implement automation to streamline the landslide detection process and
ensure scalability for large geographical areas.
 Landslide inventory mapping and
post seismic landslide risk reduction
model
Modules Description
Data Preprocessing Module:
• Prepare and clean diverse data types, including satellite imagery,
topographic information, and meteorological data, ensuring
consistency and compatibility for subsequent analysis.

Feature Extraction Module:

• Extract relevant features from preprocessed data to capture essential
information related to slope stability and potential landslide factors.

Real-Time Monitoring Module:

• Enable continuous monitoring of environmental conditions and
update the machine learning model in real-time for adaptive landslide
detection.

Automation and Scalability Module:

• Implement automation in the landslide detection process to enhance
efficiency, allowing the system to scale for larger geographical areas
and datasets.