Flooding is among the most destructive of natural hazards. It can cause loss of life, destroy infrastructure and property, and impede economic and social activities. Floods are primarily caused by heavy rainfall resulting in river overflows, flash floods, and coastal flooding. The risk of flooding is expected to increase, especially due to climate change, resulting in an increased frequency of precipitation events such as storm surges and ever-rising sea levels. Many experts now agree that absolute flood protection is not possible. This has steered a shift of focus from flood control using civil engineering structures to Flood Risk Management (FRM), which now aims at developing the resilience of communities to flooding events.
The approaches to FRM are largely made possible by the availability of geospatial technology that can reliably model flood risk. These models span from 1D models for simple terrains to more complex 2D and 3D models for complex terrains, processes, and built-up areas. The key factor in getting the most accurate results is choosing the correct model for a particular location. Generally, the parameters used to quantify flood hazard include land use/land cover, elevation, slope, topographic wetness index (TWI), surface runoff, landform, lithology, distance from the main river, soil texture, and soil drainage.
The steps below describe the process of flood risk modelling in ArcGIS. This tutorial is written using ArcMap, but the same workflow will translate to ArcGIS Pro with minimal adjustments. The parameters used for this demonstration are slope, elevation, rainfall, land use/land cover (LULC), and Euclidean distance.
Step 1: Data Acquisition and Preparation
Datasets needed: Area of interest (AOI), digital elevation model (DEM), land use/land cover (LULC), precipitation
The first stage of flood risk modelling is collecting the relevant data that will be used to create the model. Most of the data required can be found free of charge at various geospatial data repositories online.
The AOI can be obtained from an administrative boundary shapefile, or you can digitise a custom area of interest yourself. Its main purpose is to clip a defined area from the larger datasets that will be used in the modelling process.
The DEM can be obtained from the USGS EarthExplorer website. It contains elevation data that is useful for calculating hydrological networks and deriving the slope of the land. If your AOI cannot be covered by a single DEM scene, download all the raster tiles that cover your AOI and merge them in ArcMap using the “Mosaic to New Raster” tool, found under Data Management Tools – Raster – Raster Dataset. USGS provides SRTM data at 30 m resolution, which may be too coarse for detailed local analysis. If a higher-resolution dataset is available for your area, it is preferable to use it.
Rainfall data can be obtained from the CHRS Data Portal, maintained by the Center for Hydrometeorology and Remote Sensing at UC Irvine, which provides the PERSIANN family of global precipitation products. LULC data is available from ESRI’s Sentinel-2 Land Use/Land Cover Downloader. These datasets provide the rainfall and LULC parameters to be included in the modelling process during overlay.
At this point, it is best practice to create a geodatabase for the project in which all outputs will be stored. This helps with locating outputs quickly and organising results from different processing stages.
Step 2: Hydrological Analysis
- Fill Sinks (Fill tool)
The Fill tool removes sinks from the DEM — sinks are cells with no defined drainage direction because all neighbouring cells are at a higher elevation. Filling these depressions produces a hydrologically conditioned surface that is essential for accurate flow analysis. To run the Fill tool, go to Toolbox – Spatial Analyst Tools – Hydrology – Fill. Under “Input surface raster” load your DEM, then specify a storage location under “Output surface raster” and click OK.
Figure 1: Fill
- Determine Flow Direction (Flow Direction tool)
The Flow Direction tool can be found under the Hydrology toolset in ArcGIS. It uses the D8 algorithm to determine the direction of flow from each cell to its steepest downslope neighbour among the eight surrounding cells. To run the tool, set “Input surface raster” to the result of the Fill tool, specify a storage location under “Output flow direction raster”, and click OK.
Figure 2: Flow Direction
- Computing Accumulation (Flow Accumulation tool)
The Flow Accumulation tool determines the cells into which flow accumulates, based on the flow direction grid. Still under the Hydrology toolset, set “Input flow direction raster” to the result of the Flow Direction tool, choose a storage location, and click OK.
Figure 3: Flow Accumulation
After obtaining the flow accumulation result, the next step is to threshold it so that only cells accumulating flow from more than 500 upstream cells are retained as stream cells. This can be done in the Raster Calculator or using the “Con” tool under Conditional tools in the Spatial Analyst toolset. This tutorial uses the Raster Calculator: load the flow accumulation raster, add the expression “> 500” after it, and click OK. The threshold value of 500 can be adjusted to produce a denser or sparser stream network depending on the size and characteristics of the AOI.
Figure 4: Raster Calculator
- Calculating Linear Networks (Stream Order tool)
The Stream Order tool assigns an order value to each segment of the stream network. By default it uses the Strahler method, in which stream order increases only when two streams of the same order meet. This is useful for identifying higher-order, higher-risk stream channels. Still under the Hydrology toolset, open the Stream Order tool, set “Input stream raster” to the result of the Raster Calculator step, set “Input flow direction raster” to the result of the Flow Direction tool, choose a storage location, and click OK.
Figure 5: Stream Order
- Creating Features from a Linear Network (Stream to Feature tool)
The Stream to Feature tool converts the raster stream network into a vector feature class. Also located under the Hydrology toolset, open Stream to Feature, set “Input stream raster” to the result of the Stream Order tool, set “Input flow direction raster” to the result of the Flow Direction tool, choose a storage location, and click OK.
Figure 6: Stream to Feature
Step 3: Deriving Euclidean Distance
Euclidean distance measures the straight-line distance from each cell to the nearest stream feature, providing a proxy for proximity to flood-prone channels. To calculate it, expand “Distance” under Spatial Analyst Tools and open “Euclidean Distance”. Set “Input raster or feature source data” to the result of the Stream to Feature tool, choose a storage location, and click OK.
Figure 7: Euclidean Distance
Step 4: Deriving Slope
Slope measures the rate of change in elevation between adjacent cells in the AOI. Flat, low-lying areas with gentle slopes accumulate water more readily and are therefore at greater flood risk. To derive slope, go to Spatial Analyst Tools, expand “Surface”, and open “Slope”. Set “Input raster” to the DEM of the AOI, specify a storage location, and click OK.
Figure 8: Slope
Step 5: Reclassification
Reclassification standardises the number of classes in each dataset to a uniform scale so that the layers can be combined in the weighted overlay. In this tutorial a classification scale of 1–5 is used to represent flood risk: 1 – Very Low, 2 – Low, 3 – Moderate, 4 – High, and 5 – Very High. Each parameter must therefore be reclassified into five classes.
To perform reclassification, expand “Reclass” under Spatial Analyst Tools and double-click “Reclassify”. Load the raster to be reclassified under “Input raster”, click the “Classify” button, and choose 5 classes. Because low slopes are more susceptible to flooding than high slopes, use “Reverse New Values” to ensure that low slope values map to high flood-risk class values. If this is not done, the output will incorrectly imply that low slopes carry low flood risk. The same logic must be applied consistently to all rasters: rainfall values should not be reversed (higher rainfall → higher risk), while elevation and Euclidean distance values should both be reversed (lower elevation and shorter distance to streams → higher risk). Applying sound physical geography reasoning at this stage is essential to producing a valid flood risk model.
Figure 9: Reclassify
Step 6: Overlaying Weights
After standardising the number of classes in each dataset to five, it is possible to overlay the influence of each parameter by assigning weights. Weights should be assigned based on the relative influence of each factor on flood occurrence in the specific AOI. There is no universal set of weights that applies to all locations, as each area responds differently to these parameters. For this tutorial, the following weights were used: Slope – 40%, Elevation – 40%, Rainfall – 10%, LULC – 5%, Euclidean Distance – 5%. Note that all weights must sum to 100%.
To overlay weights, open “Weighted Overlay” under Overlay tools in the Spatial Analyst toolset. Click the “+” button to add the reclassified parameters one by one. Under “% Influence”, enter the weight for each parameter. Set the “Evaluation scale” to “1 to 5 by 1”. Ensure that the “Scale Value” entries in the overlay table match the reclassified values in the “Field” column for each layer. Then click OK.
Figure 10: Weighted Overlay
Step 7: Presenting Results
The Weighted Overlay tool outputs a flood risk map with values 1–5 corresponding to the classification scale defined during reclassification. Rename the values: 1 – Very Low through to 5 – Very High, as defined earlier. Switch to layout view in ArcMap and add standard map elements such as a neat line, title, scale bar, legend, and north arrow. The map can then be exported to a format of choice (PDF, PNG, JPEG, and so on). The final output represents the relative flood susceptibility of each location within the AOI.
Figure 11: Flood Risk Model
Want to learn more about the science behind flood modelling? Listen to our podcast episode on Flood Monitoring from Space with Dr. Shay Strong of ICEYE.
