Flood Susceptibility Mapping: Model Design

Aim

Workshop Objectives

Workshop Structure

Day 1: Geospatial Intelligence & Multi-Source Data Fusion
Focus: Establishing a robust, high-resolution foundational database.

• Advanced Spatial Stratigraphy: Leveraging Multi-Criteria Decision Analysis (MCDA) and high-resolution Digital Elevation Models (DEMs).
• Hydro-Climatic Synthesis: Integrating CMIP6 climate projections and satellite altimetry (GRACE/Sentinel) into susceptibility frameworks.
• Geomorphological Influencers: Deep dive into Slope, Aspect, TWI (Topographic Wetness Index), and SPI (Stream Power Index) dynamics.

Hands-On Sessions

• Hands-On I: Automated Data Pipelines—API-based acquisition of global environmental datasets.
• Hands-On II: Topographic Morphometry—Generating secondary hydrological derivative layers using GIS automation.

Day 2: Algorithmic Architecture & Predictive Modeling
Focus: Navigating the transition from statistical inference to AI-driven forecasting.

• Comparative Model Paradigms: Evaluating Frequency Ratio (FR) and Weight of Evidence (WoE) against non-linear ML approaches.
• Advanced ML Ensembles: Deploying Random Forest (RF), Support Vector Machines (SVM), and Gradient Boosting Machines (XGBoost) for spatial classification.
• Optimization Strategies: Hyperparameter tuning using Bayesian optimization to minimize structural risk.
• The Hybrid Frontier: Introduction to Neuro-Fuzzy systems and Deep Learning (CNNs) for complex flood-pattern recognition.

Hands-On Sessions

• Hands-On I: Scripting Statistical Benchmarks—Implementing Bivariate and Multivariate regression models.
• Hands-On II: Neural Network Implementation—Building and training a supervised ML model for spatial prediction.

Day 3: Validation Rigor, Uncertainty, and Policy Integration
Focus: Ensuring model reliability for publication and industrial application.

• Diagnostic Metrics: Beyond accuracy—interpreting the Receiver Operating Characteristic (ROC) curve and Area Under the Curve (AUC).
• Sensitivity Analysis: Identifying dominant predictors using Jackknife tests and Variance-based Sobol indices.
• Uncertainty Quantification (UQ): Assessing error propagation in spatial modeling to enhance decision-making confidence.
• Future-Proofing: Scaling models for real-time monitoring and integration into Early Warning Systems (EWS).

Hands-On Sessions

• Hands-On I: Performance Analytics—Generating Confusion Matrices and F1-Scores for model comparison.
• Hands-On II: Robustness Testing—Conducting Monte Carlo simulations for spatial uncertainty assessment.

Who Should Enrol?

Important Dates

Workshop Outcomes