Climate Data

• Rainfall

• Humidity

• Temperature

Hydrological & Environmental Data

• SPI

• Drainage Density

• Land Use/Land Cover (LULC)

• Soil Type

Rainfall

Humidity

Temperature

SPI

Drainage Density

LULC

Soil Type

Data obtained from:

• Remote Sensing: Sentinel, Landsat, MODIS

Processing steps:

• Standardize coordinate system

• Spatial resampling

• Data cleaning & normalization

• Flood inventory map creation

• Labeled flood / non-flood training dataset generation

Tree-Based Models

• XGBoost

• Random Forest

• CatBoost

Regression Models

• Lasso

• Ridge

• Elastic Net

Deep Learning Models

• MLP, ANN, RNN, LSTM, CNN, DNN

SVM-Based Models

• SVR, MARS, KNN

XGBoost

Random Forest

CatBoost

Lasso

Ridge

Elastic Net

MLP, ANN, RNN, LSTM, CNN, DNN

SVR, MARS, KNN

• Cross-validation & testing

• Evaluation metrics:

AUC, ROC, F1-Score, RMSE, NSE

• Ensemble stacking

• Ensemble blending

• Re-validate ensemble models

• Flood Susceptibility Maps (Present & Future)

• Variable Importance Analysis (SHAP, Permutation)

• Model Accuracy Assessment

• Uncertainty Analysis