Flood Risk Assessment Using Digital Elevation & HAND Models

This article demonstrates a Python and Jupyter Notebook workflow for rapid flood risk assessment in northeastern Brazil's rural and small-city areas. Leveraging a digital elevation model (DEM) and the Height Above Nearest Drainage (HAND) model, this method provides a real-time, low-resource solution for identifying inundation likelihood.

Key Questions Addressed:

• DEM data acquisition for flood risk analysis.
• Setting up the Python programming environment.
• DEM preprocessing for drainage extraction.
• Utilizing the HAND model to classify flood risk levels ("very high," "high," "moderate," "low," "very low").

Table of Contents:

• Introduction
• Environment Setup
• Data Acquisition and Preparation
  • Data Acquisition
  • Data Preprocessing
• Flow Direction and Accumulation
  • Flow Direction Calculation
  • Flow Accumulation Calculation
  • Stream Network Extraction
• HAND Model Application
• Flood Risk Classification
• Results and Discussion
• Conclusion
  • References
• FAQ

Environment Setup:

This workflow utilizes a Jupyter Notebook running Python 3.12 and the following libraries: NumPy, WhiteboxTools, GDAL, RichDEM, and Matplotlib.

Data Acquisition and Preparation:

Data Acquisition:

Elevation data is sourced from FABDEM (Forest and Buildings Removed Copernicus DEM), freely accessible via the University of Bristol's website [1]. FABDEM offers a global 1-arc-second resolution DEM (approximately 30 meters at the equator), correcting for building and tree height biases. This study focuses on a 1º x 1º area in northeastern Brazil (6ºS 39ºW to 5ºS 38ºW, WGS84). This region, shown in Figure 1, experienced unusually heavy rainfall in 2024.

Data Preprocessing:

Preprocessing involves filling DEM sinks (depressions) using WhiteboxTools and RichDEM to ensure accurate hydrological modeling.

Flow Direction and Accumulation:

Flow Direction Calculation:

Flow direction is calculated using the D8 method, assigning each pixel a value (1-128) representing the steepest downslope direction. (See Figure 2).

Flow Accumulation Calculation:

Flow accumulation identifies areas of water collection by counting upstream contributing pixels. High accumulation values indicate streams and rivers. (See Figure 3).

Stream Network Extraction:

A threshold (15 in this study) is applied to the flow accumulation raster to delineate the stream network.

HAND Model Application:

The HAND model calculates the height of each DEM pixel above the nearest drainage point. Higher values indicate lower flood risk. (See Figure 4).

Flood Risk Classification:

Based on HAND values, flood risk is classified into five levels (Table 1).

Table 1: Flood Risk Classification

Risk Level	Threshold (m)	Class Value
Very High	0 – 1	5
High	1 – 2	4
Medium	2 – 6	3
Low	6 – 10	2
Very Low	≥10	1

Results and Discussion:

The classified HAND raster (Figure 5) and its GeoTIFF export (Figure 6, visualized in QGIS) highlight high-risk (yellow) and very high-risk (red) areas near the stream network.

Conclusion:

The HAND model provides a computationally efficient and rapid method for flood risk assessment, particularly valuable in resource-constrained settings. This workflow is adaptable to various regions and situations.

Jupyter Notebook available here.

References: (List of references as provided in the original text)

Frequently Asked Questions: (FAQ section as provided in the original text)

The above is the detailed content of Flood Risk Assessment Using Digital Elevation & HAND Models. For more information, please follow other related articles on the PHP Chinese website!