This story map requires JavaScript, but running JavaScript is not currently allowed by your web browser. If you wish to view this story, please enable JavaScript in this browser or try a different browser.

Impervious Surface Analysis

Many local governments use impervious surface calculations to determine the storm water bills for properties. This story map shows the steps involved to use high resolution multispectral imagery to derive a pervious/impervious surface map for a residential area near Louisville, Kentucky. The map is then used to compute the total square footage of impervious features per parcel. We’ll be using an object-oriented feature extraction method to accomplish this.

Data Courtesy: the Louisville/Jefferson County Information Consortium.

What is Impervious?

Impervious surfaces have the following characteristics:

 

  • Constructed surfaces such as buildings, roads, parking lots, brick, asphalt, and concrete.
  • Areas of man-made compacted soil or material such as mining or unpaved parking lots (no vegetation present) can also be considered impervious.

 

Non-impervious or "pervious" surfaces can be defined as:

 

  • All vegetated areas, both natural and man-made.
  • Water bodies and wetland areas.
  • Ski runs.
  • Natural occurring barren areas (including rocky shores, sand, and bare soil).

 

In this map of our study area, you'll see many of these types of features present. Calculating impervious surfaces manually for even a relatively small area such as this would be tedious and cost prohibitive. Instead, we'll use multispectral imagery and analysis methods to automate a process that can be used here and is also easily scalable to much larger areas.

Choosing the Right Bands

When using multispectral imagery, different bands of the spectrum can be used to identify different features on the ground. When performing analysis, it is important to choose a band combination to visually discriminate your features of interest. False-color imagery combines these bands, and it a good choice to start with for identifying pervious/impervious features:

 

  • Infrared (IR) is useful for to detecting , extracting, and eliminating vegetation.
  • Red is important for discriminating bare soil.
  • Blue is important for discriminating urban features, especially concrete.

 

Segmenting the Imagery

 

Segmentation is the process where pixels in close proximity and having similar spectral characteristics are grouped together into a segment. Segments exhibiting certain shapes, and spectral and spatial characteristics can be further grouped into objects. During this classification process, we’re classifying the segments, not pixels.

 

We’ll use both raster functions and geoprocessing tools in this workflow. Raster functions are used to assess and preview the results. When we’re satisfied we’ll use geoprocessing to save the results.

 

Creating Training Samples

Next, we'll create "training samples" over select features in the image. This process of identifying what sample segments represent leads to a more accurate automated classification of the image.

 

 

First we'll identify impervious features such as roofs, driveways, and roads, as well as pervious features such as trees, grass, and exposed dirt. The Training Sample Manager tool will be used to create and save our training samples. The training samples will be used for classification purposes at a later point.

 

 

 

Classifying the Image

We’ll now create a classified image based on the segmented image and the SVM classifier. We will do this the same way that we segmented the image to assess and preview the results using raster functions, and when we’re satisfied we’ll use geoprocessing to save the results.

 

Reclassifying to Pervious/Impervious

Next, we’ll use the Reclassify tool to reclassify the scene so we can see it as a classified image with only two classes, representing impervious and pervious features. Features classified as impervious are shown in magenta.

 

Calculating Impervious Square Footage Per Parcel

 

The final step uses tools in the Spatial Analyst Zonal toolbox to calculate the amount of impervious surface per parcel, which can then be used by the local government to determine the storm water bills for each property.

 

 

 

 

Impervious Surface Analysis

Many local governments use impervious surface calculations to determine the storm water bills for properties. This story map shows the steps involved to use high resolution multispectral imagery to derive a pervious/impervious surface map for a residential area near Louisville, Kentucky. The map is then used to compute the total square footage of impervious features per parcel. We’ll be using an object-oriented feature extraction method to accomplish this.

Data Courtesy: the Louisville/Jefferson County Information Consortium.

Tap for details Swipe to explore

LEARN MORE

Tap to go back Swipe to explore

What is Impervious?

Impervious surfaces have the following characteristics:

 

  • Constructed surfaces such as buildings, roads, parking lots, brick, asphalt, and concrete.
  • Areas of man-made compacted soil or material such as mining or unpaved parking lots (no vegetation present) can also be considered impervious.

 

Non-impervious or "pervious" surfaces can be defined as:

 

  • All vegetated areas, both natural and man-made.
  • Water bodies and wetland areas.
  • Ski runs.
  • Natural occurring barren areas (including rocky shores, sand, and bare soil).

 

In this map of our study area, you'll see many of these types of features present. Calculating impervious surfaces manually for even a relatively small area such as this would be tedious and cost prohibitive. Instead, we'll use multispectral imagery and analysis methods to automate a process that can be used here and is also easily scalable to much larger areas.

Tap for details Swipe to explore

LEARN MORE

Tap to go back Swipe to explore

Choosing the Right Bands

When using multispectral imagery, different bands of the spectrum can be used to identify different features on the ground. When performing analysis, it is important to choose a band combination to visually discriminate your features of interest. False-color imagery combines these bands, and it a good choice to start with for identifying pervious/impervious features:

 

  • Infrared (IR) is useful for to detecting , extracting, and eliminating vegetation.
  • Red is important for discriminating bare soil.
  • Blue is important for discriminating urban features, especially concrete.

 

Tap for details Swipe to explore

LEARN MORE

Tap to go back Swipe to explore

Segmenting the Imagery

 

Segmentation is the process where pixels in close proximity and having similar spectral characteristics are grouped together into a segment. Segments exhibiting certain shapes, and spectral and spatial characteristics can be further grouped into objects. During this classification process, we’re classifying the segments, not pixels.

 

We’ll use both raster functions and geoprocessing tools in this workflow. Raster functions are used to assess and preview the results. When we’re satisfied we’ll use geoprocessing to save the results.

 

Tap for details Swipe to explore

LEARN MORE

Tap to go back Swipe to explore

Creating Training Samples

Next, we'll create "training samples" over select features in the image. This process of identifying what sample segments represent leads to a more accurate automated classification of the image.

 

 

First we'll identify impervious features such as roofs, driveways, and roads, as well as pervious features such as trees, grass, and exposed dirt. The Training Sample Manager tool will be used to create and save our training samples. The training samples will be used for classification purposes at a later point.

 

 

 

Tap for details Swipe to explore

LEARN MORE

Tap to go back Swipe to explore

Classifying the Image

We’ll now create a classified image based on the segmented image and the SVM classifier. We will do this the same way that we segmented the image to assess and preview the results using raster functions, and when we’re satisfied we’ll use geoprocessing to save the results.

 

Tap for details Swipe to explore

LEARN MORE

Tap to go back Swipe to explore

Reclassifying to Pervious/Impervious

Next, we’ll use the Reclassify tool to reclassify the scene so we can see it as a classified image with only two classes, representing impervious and pervious features. Features classified as impervious are shown in magenta.

 

Tap for details Swipe to explore

LEARN MORE

Tap to go back Swipe to explore

Calculating Impervious Square Footage Per Parcel

 

The final step uses tools in the Spatial Analyst Zonal toolbox to calculate the amount of impervious surface per parcel, which can then be used by the local government to determine the storm water bills for each property.

 

 

 

 

Tap for details Swipe to explore

LEARN MORE

Tap to go back Swipe to explore

An error has occurred