Debris flow modeling data using EDDA model

Description:

This dataset contains the numerical simulation results and corresponding topographic data used in the study published in the paper: "Multi-task super-resolution deep learning overcomes computational bottlenecks in high-fidelity debris flow modeling".

The data is stored in GeoTIFF format and organized into a hierarchical folder structure based on the location and resolution of the debris flow simulation results. The repository includes data for the primary study area (Lantau) at multiple resolutions and a separate, independent test area (C210) for model inference validation.

1. Study Area: Lantau (lantau_*/)

Data for the Lantau Island, used primarily for model training and validation. It includes simulations at a fine resolution (5m) and two coarser resolutions (15m, 30m). Each subfolder contains the following raster files:

lantau_5m/(Fine-Grid Simulation Outputs):

This directory stores the core simulation outputs at 5-meter resolution.

File Naming Convention: c_{catchment_id}_d_{time}.tif(Debris Flow Depth) and c_{catchment_id}_v_{time}.tif(Debris Flow Velocity).

Examples: c_25_d_0.0.tif(Depth at catchment 25, t=0.0s), c_43_v_1320.0.tif(Velocity at catchment 43, t=1320.0s).

lantau_5m_mask/(Inundation Extent):

Binary masks indicating the final inundation extent derived from the 5m simulations (values typically 0 for non-inundated and 1 for inundated cells).

lantau_5m_topo/(Topographic Derivatives):

Input features for the deep learning model derived from the 5m DEM.

Files: c_{catchment_id}_acc.tif(Flow Accumulation), c_{catchment_id}_ele.tif(Elevation), c_{catchment_id}_slp.tif(Slope), c_{catchment_id}_profile.tif(Profile Curvature), c_{catchment_id}_plan.tif(Plan Curvature).

Note:The curvature files were generated to support the input parameter sensitivity analysis performed during the manuscript revision phase.

lantau_15m/& lantau_30m/(Coarse-Grid Simulation Outputs):

These folders contain the corresponding depth and velocity simulation results at 15-meter and 30-meter resolutions, respectively. They serve as the "low-fidelity" data used to demonstrate the super-resolution capabilities of the proposed model.

2. Independent Test Area: C210 (c210_*/)

This directory contains data for catchment ID 210, which was held out during the training process and used exclusively for the final model inference and application testing.

Structure: The structure mirrors the Lantau dataset, containing c210_5m, c210_15m, and c210_30mfolders.

Purpose: These files allow users to validate the model's generalization capabilities on unseen geographical terrain.