Mapping Impervious Surface |
| |
| This example shows two images are the same area in southwestern Kennesaw, Georgia. Vegetation in these images is red. |
|
|
|
| 1 meter color infrared aerial photo |
30 meter pixel satellite imagery
Landsat ETM+ sensor |
back to top |
|
Step one:
We create a binary map of impervious surface using the 1 meter aerial photograph. A 30 meter grid is placed over the photograph and the percentage of impervious surface for each cell is calculated. This represents the 30 meter cell size of the Landsat image. |
|
|
| back to top |
|
Step two:
We then build a regression model using the photo interpreted impervious surface to predict the percentage of impervious surface found in each grid cell of the Landsat image. |
|
|
Yellow area is actual
impervious surface. |
Predicted impervious surface.
|
| back to top |
|
Animation of process.
|
| back to top |
|
|
|
|
This example shows two images of the same area in northern Newton county. |
|
|
|
| 1 meter color infrared aerial photo |
30 meter pixel satellite image of same area
|
| back to top |
|
Step one:
We created a binary map of tree canopy and no tree canopy, then placed a 30m grid over the new map to calculate percentage of tree canopy for each 30 meter grid cell. |
|
|
| back to top |
|
Step two:
We then build a regression model using the photo interpreted tree canopy to predict to percentage of tree canopy found in each Landsat grid cell. |
|
|
| Actual tree canopy based on aerial photo interpretation. |
Predicted percentage of tree canopy.
The darker the grid cell, the higher
the percentage of tree canopy. |
| back to top |
|
Animation of process. |
| back to top |
|
