Mapping Forests in the Chesapeake Bay Watershed

The approach used to map tree cover was similar to that used to create maps of impervious surface cover, except instead of using GIS planimetric data sets to train the regression tree algorithm, we used fine scale maps of tree cover derived from digital orthophoto quadrangles (DOQ) and from high resolution Ikonos satellite imagery.

Tree cover map of Montgomery County, Maryland dervied from Ikonos imagery. In areas of persistent cloud cover, forest cover from a GIS data set is shown.

While the tree cover maps derived from Ikonos were ultimately used as a source of training data to derive a Bay-wide map of tree cover from Landsat imagery for the entire watershed, they provide a highly detailed and accurate view of forest extent for a 1800 km² area in central Maryland.

Like the impervious surface maps, tree cover for the entire watershed is expressed as a continuous value between 0-100% that describes the amount of forested area within each 30 meter pixel. The fine scale information conveyed by these maps allows us to monitor tree cover even within relatively narrow riparian (stream) buffer zones, where pollutants from agricultural and impervious areas can be filtered by root systems that also resist erosion by holding stream banks in place.

Tree cover in Montgomery County, derived from Landsat 7 satellite imagery. Darker shades of green indicate higher proportions of tree cover.

Tree cover, derived from Landsat imagery, across the Chesapeake Bay watershed.

Access tree cover data here:

ftp://ftp.whrc.org/GIS-RS/cbw/treecover

©Woods Hole Research Center, 2007