Subtopic Deep Dive

Landslide Susceptibility Mapping with GIS
Research Guide

What is Landslide Susceptibility Mapping with GIS?

Landslide Susceptibility Mapping with GIS uses Geographic Information Systems to model spatial probabilities of landslide occurrence through statistical and machine learning methods validated by ROC-AUC metrics.

Researchers apply logistic regression, support vector machines, and decision trees to GIS datasets for susceptibility zonation (Ayalew and Yamagishi, 2004; Pradhan, 2012). Over 10 highly cited papers since 2004 demonstrate model comparisons across regions like Japan and Iran (Bui et al., 2015; Pourghasemi et al., 2012). Validation relies on multi-scale inventories and global occurrence data (Guzzetti et al., 2012; Froude and Petley, 2018).

Curated Papers

Key Challenges

Why It Matters

Susceptibility maps guide land-use planning in mountainous regions, as outlined in zoning guidelines (Fell et al., 2008). They support early warning systems by integrating with global fatal landslide inventories (Froude and Petley, 2018). Quantitative risk analysis frameworks enable regional hazard assessment for policy (Corominas et al., 2013). Machine learning advancements improve predictive accuracy over traditional logistic models (Merghadi et al., 2020; Pradhan, 2012).

Key Research Challenges

Data Quality Variability

Landslide inventories suffer from incompleteness and scale inconsistencies across regions (Guzzetti et al., 2012). GIS layers like DEM resolution vary, impacting model inputs (Jaboyedoff et al., 2010). Achieving consistent multi-scale validation remains difficult (Froude and Petley, 2018).

Model Transferability Limits

Statistical models like logistic regression perform variably across geologies, as shown in comparative studies (Ayalew and Yamagishi, 2004). Machine learning algorithms require site-specific tuning despite high AUC scores (Bui et al., 2015). Generalization to ungauged areas challenges global application (Merghadi et al., 2020).

Validation Metric Gaps

ROC-AUC overlooks class imbalance in rare landslide events (Pradhan, 2012). Quantitative risk zoning demands multi-hazard integration beyond susceptibility (Corominas et al., 2013). Standardized verification protocols are lacking for ML ensembles (Fell et al., 2008).

Essential Papers

Landslide inventory maps: New tools for an old problem

Fausto Guzzetti, Alessandro Mondini, Mauro Cardinali et al. · 2012 · Earth-Science Reviews · 2.0K citations

Global fatal landslide occurrence from 2004 to 2016

Melanie Froude, David N. Petley · 2018 · Natural hazards and earth system sciences · 1.9K citations

Abstract. Landslides are a ubiquitous hazard in terrestrial environments with slopes, incurring human fatalities in urban settlements, along transport corridors and at sites of rural industry. Asse...

The application of GIS-based logistic regression for landslide susceptibility mapping in the Kakuda-Yahiko Mountains, Central Japan

Lulseged Ayalew, Hiromitsu Yamagishi · 2004 · Geomorphology · 1.9K citations

Guidelines for landslide susceptibility, hazard and risk zoning for land use planning

Robin Fell, Jordi Corominas, C. Bonnard et al. · 2008 · Engineering Geology · 1.4K citations

Spatial prediction models for shallow landslide hazards: a comparative assessment of the efficacy of support vector machines, artificial neural networks, kernel logistic regression, and logistic model tree

Dieu Tien Bui, Trần Anh Tuấn, Harald Klempe et al. · 2015 · Landslides · 1.2K citations

Recommendations for the quantitative analysis of landslide risk

Jordi Corominas, C.J. van Westen, Paolo Frattini et al. · 2013 · Bulletin of Engineering Geology and the Environment · 1.2K citations

This paper presents recommended methodologies for the quantitative analysis of landslide hazard, vulnerability and risk at different spatial scales (site-specific, local, regional and national), as...

A comparative study on the predictive ability of the decision tree, support vector machine and neuro-fuzzy models in landslide susceptibility mapping using GIS

Biswajeet Pradhan · 2012 · Computers & Geosciences · 1.2K citations

Reading Guide

Foundational Papers

Start with Ayalew and Yamagishi (2004) for GIS-logistic regression baseline, then Guzzetti et al. (2012) for inventory essentials, and Fell et al. (2008) for zoning guidelines.

Recent Advances

Study Merghadi et al. (2020) for ML algorithm performance overview and Bui et al. (2015) for SVM vs. logistic tree comparisons.

Core Methods

Core techniques: logistic regression on GIS factors (Ayalew 2004), SVM/decision trees (Pradhan 2012; Bui 2015), fuzzy logic/AHP (Pourghasemi 2012), validated by ROC-AUC.

How PapersFlow Helps You Research Landslide Susceptibility Mapping with GIS

Discover & Search

Research Agent uses searchPapers and citationGraph to trace logistic regression origins from Ayalew and Yamagishi (2004) to ML comparisons in Bui et al. (2015), revealing 10+ high-citation clusters. exaSearch uncovers GIS-specific datasets; findSimilarPapers extends to Pourghasemi et al. (2012) fuzzy-AHP hybrids.

Analyze & Verify

Analysis Agent applies readPaperContent to extract ROC-AUC values from Pradhan (2012), then verifyResponse with CoVe cross-checks claims against Guzzetti et al. (2012) inventories. runPythonAnalysis recreates logistic models via NumPy/pandas on susceptibility data, with GRADE grading for evidence strength in multi-model comparisons (Merghadi et al., 2020).

Synthesize & Write

Synthesis Agent detects gaps in transferability between statistical and ML models (Ayalew 2004 vs. Bui 2015), flagging contradictions in validation metrics. Writing Agent uses latexEditText for map figures, latexSyncCitations for 1961-citation Guzzetti paper, and latexCompile for zoning guideline reports (Fell et al., 2008); exportMermaid visualizes model comparison workflows.

Use Cases

"Reproduce logistic regression susceptibility model from Ayalew 2004 with Python."

Research Agent → searchPapers(Ayalew) → Analysis Agent → readPaperContent → runPythonAnalysis(scikit-learn logistic on GIS raster sample) → matplotlib AUC plot output.

"Draft LaTeX report comparing SVM vs. neuro-fuzzy for Iran landslide mapping."

Research Agent → citationGraph(Pradhan 2012) → Synthesis → gap detection → Writing Agent → latexEditText(sections) → latexSyncCitations(Pourghasemi) → latexCompile → PDF with ROC curves.

"Find GitHub repos implementing GIS-based decision tree landslide models."

Research Agent → citationGraph(Bui 2015) → Code Discovery → paperExtractUrls → paperFindGithubRepo → githubRepoInspect(QGIS plugins) → runnable Jupyter notebook output.

Automated Workflows

Deep Research workflow conducts systematic review of 50+ GIS-landslide papers, chaining searchPapers → citationGraph → GRADE grading for Pradhan (2012) models. DeepScan applies 7-step CoVe verification to Merghadi (2020) ML overview, checkpointing AUC claims against Froude inventories. Theorizer generates hybrid fuzzy-logistic theory from Ayalew baselines.

Try Doxa for Landslide Susceptibility Mapping with GIS Research

Frequently Asked Questions

What defines landslide susceptibility mapping with GIS?

It models spatial landslide probabilities using GIS-integrated statistical methods like logistic regression and ML classifiers, validated by ROC-AUC on inventory data (Ayalew and Yamagishi, 2004).

What are common methods?

Logistic regression (Ayalew 2004), SVM, decision trees, neuro-fuzzy (Pradhan 2012; Bui 2015), and fuzzy-AHP (Pourghasemi 2012) process DEM, slope, and land-use layers.

What are key papers?

Foundational: Guzzetti et al. (2012, 1961 cites) on inventories; Ayalew (2004, 1852 cites) on GIS-logistic. Recent: Merghadi et al. (2020, 1014 cites) ML overview; Bui et al. (2015, 1225 cites) on kernel methods.

What open problems exist?

Transferability across regions, handling inventory biases (Guzzetti 2012), and integrating quantitative risk beyond susceptibility (Corominas 2013; Fell 2008).