Flood Susceptibility Mapping Data with Ensemble Feature Selection for Khuzestan, Iran

19 environmental factors were sourced from Google Earth Engine for flood susceptibility mapping in Khuzestan Province, Iran. The dataset contains 1,000 sample points (500 flood and 500 non-flood) used to train a model, with results generalized to the entire study area. Author Mohammad Kazemi published the data on figshare in 2026 under a CC-BY-4.0 license.

Use Cases

• Comparing the effectiveness of nine ensemble feature selection methods (e.g., Boruta, RFE) for geospatial predictors.
• Training and validating metaheuristic-optimized deep learning models (e.g., LSTM-WOA) for flood prediction.
• Identifying critical environmental variables like NDVI and Daily Minimum Temperature for flood susceptibility in arid regions.
• Creating high-precision flood susceptibility maps for specific geographic areas based on consensus feature selection.

Strengths

• Applies a formalized consensus rule, retaining variables only if selected by a majority of nine feature selection methods.
• Model validation reported specific performance metrics, including an F1-Score of 0.88 and Cohen’s Kappa of 0.75 for the best model.
• The methodology is clearly described, involving 1,000 sample points and 19 initial factors from a defined source (Google Earth Engine).

Limitations

• The dataset is very small at 5.5 KB, indicating limited scope, likely containing summary results rather than raw training data.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for large-scale modeling.

Provenance

Source

Google Earth Engine, author Mohammad Kazemi.

Collection Method

Environmental factors were sourced and processed; feature selection and model training were conducted on 1,000 sample points.

Freshness

Last updated 2026-04-29 17:42:14; freshness should be verified.

Geography

Khuzestan Province, Iran.