Using Ikonos Satellite Imagery to Map Riparian Forest Buffers

Stream corridor vegetation, particularly forest, plays a significant role in protecting water quality and other biological resources of streams and their surrounding habitat (Virginia Cooperative Extension 2000). Measuring forest buffers along streams may be an indicator of overall watershed health, and help to identify problem areas and potential opportunities for stream restoration. Federal, State and local government agencies in the Chesapeake Bay Watershed have entered the Chesapeake 2000 Agreement, which recommends establishment of 2010 miles of riparian forest buffer by the year 2010. In order to assess the success of the program and to target plantings it is necessary to inventory and monitor the extent and distribution of riparian forest buffers (Figure 1). To date only a rudimentary framework is in place, thus RESAC is working with the Chesapeake Bay Program to implement a strategy for riparian buffer mapping and monitoring.

Traditionally, orthophotographic imagery was used to create forest maps, but this approach is time consuming and expensive. Classification of Landsat Thematic Mapper (TM) satellite data provides another means to assess buffer extent, but TM spatial resolution limits its utility to detecting buffers greater than ~90 meters (the functional minimum mapping unit of TM imagery). Ikonos multispectral imagery, at a nominal spatial resolution of 4 meters (multispectral), provides a capability for mapping riparian forest buffers of 12 m or more (Figure 2). Incorporating 1 m Ikonos panchromatic imagery could reduce the minimum buffer width mapping to just a few meters. Trade-offs of higher resolution mapping include increased data processing requirements, associated computing and data storage capabilities, and a host of image preprocessing considerations required to ensure consistent mapping across a range of conditions.

Forest Cover and Riparian Buffer Assessment

As a case study we conducted a supervised classification using an Ikonos image acquired within Montgomery County, Maryland (Figure 3). The maximum likelihood classification distinguished forested from nonforested areas (Figure 4). The resulting map was used with two stream coverages to calculate the proportion of streams buffered by forest. One stream coverage was based on existing hydrographic data provided in the Montgomery County Parks & Planning Commission (MNCPPC) planimetric database - the other coverage was created from a digital elevation model (DEM) derived from 5 foot topographic contours (also provided by MNCPPC). The former provided a more conservative estimate of the perennial stream network; and the latter extended the stream network to smaller drainages in the upper stream reaches. A 100 ft buffer was created around the arcs of the stream coverages and the percentage of forest cover in each was calculated for a small subwatershed (Figure 5).

Our exploratory work demonstrated the efficacy of a methodology using Ikonos data for determining forested area within a 100 ft riparian buffer zone. Based on these results we are implementing a county-wide riparian buffer mapping and assessment project that will enable us to monitor the extent of riparian forests, identify potential restoration opportunities, and demonstrate the feasibility of riparian buffer mapping using satellite imagery. We will then be better able to assess the utility of coarser resolution imagery including Landsat TM.

Figure 1. This Ikonos image (top) depicts a coverage of tributary streams buffered by forest in an agricultural setting. An Ikonos image (bottom) with overlaid GIS coverages shows building footprints encroaching on the 100 ft stream buffer.

Top

Figure 2. An image degradation of a high resolution multispectral (1m ADAR) image shows the appearance of 1 m, 2 m, 4 m, 10 m, 20 m, 30 m spatial resolutions. This exercise seems to indicate that the 30 m resolution of thematic mapper imagery is too coarse to capture buffers measuring less than 30 m (100 ft).

Top

Figure 3. This Ikonos image in Montgomery County shows where the initial supervised classification was performed.

Top

Figure 4. A map of the resulting classification shows forest and nonforest in several subwatershed areas.

Top

Figure 5. Analysis of a 100 ft buffer was conducted at the subwatershed level. Two stream networks were used: one provided a more conservative estimate of the perennial stream network; another extended the stream network to smaller drainages in the upper stream reaches. We calculated the ratio of non-forest vs. forest cover within the 100 ft buffer zone. This kind of analysis allowed us to correlate subwatershed water quality (assessed by independent stream life surveys) with acreage of forest within the riparian forest buffer.

Top

Literature Cited

Klapproth, Julia C, and James E. Johnson. 2000. Understanding the science behind riparian forest buffers: effects on water quality. Virginia Cooperative Extension. Publication Number 420-151. Url: http://www.ext.vt.edu/pubs/forestry/420-151/420-151.html