Determination of the Areas with Snow Avalanche Risk in Bingöl Province with the Help of GIS

Avalanche susceptibility mapping is one of the first steps to minimize the loss of life and property caused by avalanche disasters. We can consider the factors that play a role in avalanche formation in two groups; topographic features of the terrain (static) and meteorological conditions (dynamic). However, due to its static nature, terrain-related factors are more preferred in avalanche susceptibility mapping. In this study, we will determine the areas susceptible to avalanche disaster in Bingöl by using land morphology and vegetation characteristics.

For the analysis, I used the boundary data downloaded from GADM, the altitude data downloaded from the USGS and the Landsat 8 satellite image of 05.06.2022. You can access the data via the link.

As a result of the literature review, we can list the common features of avalanche-sensitive areas as follows:

• The land slope is between 28 degrees and 55 degrees
• There is a sudden change in slope (10 degrees and above)
• No avalanche-preventing dense vegetation
• It has a surface area of 625 square meters and above

Analysis Process

I used ArcGIS Pro for analysis, but you can also complete the work with ArcMap. After adding the data to the project, we will create a slope map from the elevation data. Click ‘Slope’ in the Geoprocessing menu.

Fig. 1: Required parameters for slope map

Our slope map will appear as follows;

Fig. 2: Bingol province slope map

Fig. 3: Final version of the layer window

We run the ‘Raster Calculator’ tool from the Geoprocessing menu to determine regions with slopes between 28° and 55°;

Fig. 4: The code in the figure will give us the fields we want

Our resulting map should be as follows;

Fig. 5: Areas with a slope of 28°-55° were determined

Fig. 6: Desired areas are displayed with ‘1’ in the layer window

In the next step, we will determine the areas with 10° or more slope changes. For this, we will use the ‘Focal Statistics’ tool from the Geoprocessing menu.

Fig. 7: We fill in the parameters as in the figure

With this tool, the largest of the slope values of its neighboring cells will be assigned to each cell in our slope data. Thus, the difference between the new map to be created and our slope map will give us the maximum slope change.

Fig. 8: Result map

Fig. 9: Final version of the layer window

We use the ‘Raster Calculator’ to calculate the slope change.

Fig. 10: The code in the figure will give us the slope change

Fig. 11: Slope change map

Fig. 12: Final version of the layer window

We want to select the areas with more than 10° slope changes as these places are more suitable for avalanche events.

Fig. 13: The places where the slope change is more than 10° will be selected.

Fig. 14: Result map

Fig. 15: Layer window

Another criterion is the density of vegetation. We will generate NDVI (Normalized Difference Vegetation Index) using our satellite imagery to detect the vegetation density in our study area.

NDVI calculates vegetation density from the difference between reflected (near infrared) and absorbed (red) rays by vegetation.

NDVI = (NIR — Red) / (NIR + Red)

In the NDVI map, values range from -1 to 1. Values close to 1 indicate dense vegetation, and values close to -1 indicate water. Values close to 0 indicate that there are no plants in that area and may be an urban area.

In Landsat 8 images, NIR (near infrared) corresponds to band 5 and Red (red) corresponds to band 4.

Click Imagery -> Indices -> NDVI for NDVI calculation.

Fig. 16: Index calculations can be done via the Imagery menu.

We make our band selection as follows:

Fig. 17: Band selection for NDVI

Our resulting map will look like the one below.

Fig. 18: Result NDVI map

Fig. 19: Final version of the layer window

Using the literature and experimenting to identify areas with dense vegetation, I chose a value of 0.4. We assume that regions above this value can prevent avalanche events. We will use the ‘Raster Calculator’ to determine these regions.

Fig. 20: We select places with dense vegetation

Regions marked with ‘1’ in our result map have the potential to prevent avalanche events.

Fig. 21: Result map

Fig. 22: Layer window

When we complete our raster analysis, we see that the areas we do not want in our result maps are shown with ‘0’ and we want to delete these areas. We will use the ‘Set Null’ function for this.

Fig. 23: We delete regions whose slope value is not between 28–55

Our new result map will look like this:

Fig 24: Areas we don’t want are no longer visible

We do the same for our other result maps.

We produced the necessary result maps for our first 3 criteria. We will combine these outcome maps before moving on to our final criterion. First, we find the intersection of regions with a slope value of 28°-55° and regions with a gradient change of 10° and above:

Fig. 25: We write the code in the figure

A new raster data containing the intersection regions is generated. We will subtract those with dense vegetation from the areas shown in this data:

Fig. 26: With the code in the figure, we extract regions with dense vegetation.

Our resulting map will look like this:

Fig. 27: Map prepared using the first 3 criteria

Fig. 28: Zoomed in the result map

Fig. 29: Final version of the layer window

Our last criterion states that regions with a surface area of 625 m² and above have avalanche potential. We convert our “cig_alanlar” map to a polygon shapefile so that we can calculate the surface area:

Fig. 30: We will display our avalanche zones as polygons

Fig. 31: Avalanche zones shown as polygons

We open the attribute table of the created layer and click ‘Add Field’ in the window that appears.

Fig. 32: We will add a new column with area data

We fill in the required information for our new column as follows:

Fig. 33: New column parameters

Fig. 34: New field column created

Right click on our newly added column and select ‘Calculate Geometry’.

Fig. 35: We will calculate the surface area for each polygon.

Fig. 36: We fill the parameters in the window as in the figure.

Fig. 37: Surface area calculation completed

We will use the ‘Select by Attributes’ tool to get areas of 625 m² and above.

Fig. 38: Accessible from the top menu

Fig. 39: We fill in the parameters as in the figure

The selected regions will appear as follows:

Fig. 40: Desired areas are marked

In order to save the selected regions in a separate layer, right-click on the ‘cig_pol.shp’ layer and click Data -> Export.

Fig. 41: We save the desired regions in a separate layer

When the save process is complete, our new layer will be loaded into our project:

Fig. 42: Avalanche-prone areas

Fig. 43: Final version of our layer window

The advantage of the method used in this study is that it can be applied in different study areas and there is no need for an avalanche inventory map. Our result map shows the avalanche-hazardous regions in Bingöl in general terms, but it would be useful to include meteorological factors (temperature, precipitation, snow density, wind) in the analysis for a more detailed and high-accuracy result.

References

An ArcGIS Model for Avalanche Risk Assessment in the North Cascades, Washington

Snow avalanche risk mapping using GIS-based multi-criteria decision analysis: the case of Van, Turkey

Avalanche susceptibility mapping with the use of frequency ratio, fuzzy and classical analytical hierarchy process for Uzungol area, Turkey

GISGeography — What is NDVI (Normalized Difference Vegetation Index)?

USGS — Landsat Normalized Difference Vegetation Index