Increasing Transportation Planning Resiliency with Landslide Vulnerability Modeling


April showers bring May flowers has been a saying from as far back as the 1500s. Lately, the showers just keep coming—and so does the stormwater that follows. All this rain is leading to an increase in landslides, damaging roads and highways in southwestern Pennsylvania. Last year, millions of dollars were allocated for landslide remediation in this region.

Increasing resiliency in transportation planning is a priority for the Southwestern Pennsylvania Commission (SPC), an agency responsible for planning the use of federal and state transportation funds for the Pittsburgh region. SPC sought our help to develop a landslide susceptibility geographic information system (GIS) predictive analytics model based on a similar model recently released by the Minnesota Department of Transportation. 

Doing the Math

I came to the project with my GIS hat on as a GeoDecisions professional but knew this model would rely heavily on my civil engineering background to make it come to life. In its simplest form, the model uses a variation of Coulomb’s law of friction to develop a mathematical equation that identifies how much rain could cause a landslide. By combining this information with the National Oceanic and Atmospheric Administration (NOAA) National Weather Service Precipitation Frequency Data, we can identify areas vulnerable to landslides. Running the model against historical landslide data, we can determine the possibility of future landslides with a high degree of accuracy. 

Creating the GIS Model with the Math

Much of my work involved finding available sources of data to plug into the equation’s variables and converting it to raster data – pixelating the data to become imagery that could be viewed on a map. I found the compatible, free data sources I needed. Converting to raster data and analyzing the many maps and layers were facilitated easily using Esri®’s ArcGIS Pro, Esri’s latest professional desktop GIS software. The actual calculation, using the mathematical equation and the raster data, required the ArcGIS Map Algebra Raster Calculator included with the ArcGIS Spatial Analyst Tools. 

Machine Learning – Improving the Accuracy

In its present form, transportation planning agencies will benefit greatly from this model. However, using machine learning can attain even greater accuracy. Engaging key algorithms, our predictive model and more historical landslide data with additional explanatory variables will give us these results.

Contact me to learn more about how predictive analytics can assist with your transportation planning. Hany Hassaballa

Learn more about our transportation solutions.