Effects of Forest Disturbance on Water Quality in Appalachian Watersheds
Common sources of forest disturbance in the Appalachian Mountain region of the eastern U.S.
Predominantly-forested watersheds in the Appalachian Mountain region of the eastern U.S. are increasingly subjected to both natural and man-induced disturbances that can cause major changes in ecosystem and hydrologic functions. While the effects of low frequency, catastrophic disturbances—such as fire, floods, hurricanes, and insect defoliations—have been described in the literature, much less is known about transient, higher frequency, disturbances that tend to be:
- ubiquitous in the region;
- “stealthy” and poorly characterized both spatially and temporally;
- highly variable in extent, frequency, and intensity; and
- rarely associated with demonstrable and quantifiable ecosystem effects at scales ranging from small plots to river basins.
Forest disturbance often results in leakage of nitrogen from disturbed watersheds. At White Oak Run watershed in Shenandoah National Park, Virginia, nitrate export increased by more than 4 times following major defoliation years. Click on the graph to view a simulation demonstrating the relationship between gypsy moth defoliation
and nitrate export from 1986 through 2000. One of the most symptomatic effects of forest disturbance in the eastern U.S. is the release or “leakage” of dissolved nitrogen (N) from forests to surface waters (see figure at left). Our research is addressing the following research questions pertaining to forest disturbance in this region:
- At what scales can these N leakage responses to disturbance be observed?
- What methods can be used to detect disturbances and quantify responses?
- How do these responses scale in space and time?
- What modeling approaches are useful for prediction/regionalization?
The Fifteenmile Creek watershed (area = 159.57 km2) is located in a rural area of the Ridge and Valley Province in Maryland and Pennsylvania and is a tributary of the Potomac River; 85% of the watershed is forested, of which a significant portion is in Green Ridge (MD) and Buchanan (PA) State Forests. Two subwatersheds of FC are presently gaged: Deep Run and Terrapin Run We tested the hypothesis that interannual variations in streamwater total dissolved N (TDN) and nitrate-N concentrations within the Fifteenmile Creek watershed can be predicted from indices of forest disturbance derived from satellite data.
We used quarterly synoptic stream chemistry surveys, beginning in 2001, to measure nitrogen export from streams in the Fifteenmile Creek watershed.
Disturbance indices (DI) were calculated using the methodology of Healey et al. (2005) to provide mean aggregated indices of forest disturbance for selected years for the subwatersheds drained by the synoptic stream survey. It was computed as a linear combination of six Landsat TM reflectance bands (Tasseled Cap transformation) providing an aggregated index of brightness (B), greenness (G), and wetness (W) for each 30m x 30m pixel. Mean DI values for each subwatershed were computed from the complete set of DI values for forested pixels within each subwatershed (see below left).
Disturbance index calculations based on the Healey et al. (2003) method. Mean DI for a forested area for a particular year was then computed from the entire set of DI values for the area of interest.
Linear regression was used to test the null hypothesis by relating the change in spring baseflow TDN (and nitrate-N) concentrations from 2001 to 2005 to the change in DI.
DI calculated for Fifteenmile Creek watershed in 2000 and 2005.
Overall, a decrease in DI from 2000 to 2005 was observed (see map at right).
In Boxes 1 and 2, the increase in DI was a result of forest harvesting.
In Box 3, the decrease in DI was due to re-greening from insect defoliation. In
Boxes 4 and 5 the observed decrease in DI was due to re-greening from harvesting.
We the compared differences in DI from 2005 to 2001 and differences in stream TDN (total dissolved nitrogen) from 2005 to 2001 and observed a statistically significant relationship at sites that have more than 90% forest cover (see figure below left). This robust relationship did not hold for sites with less than 90% forest cover.
A statistically significant relationship was observed between streamwater nitrogen and DI at sites with more than 90% forest cover. Conclusions:
- Systematic water quality monitoring can be used to detect responses to ecosystem disturbances in watersheds.
- Robust “signals” include elevated TDN (and nitrate-N) concentrations in streamwater under spring baseflow conditions.
- More evidence that forest disturbances contribute to N leakage from Appalachian watersheds.
- Disturbance of mid-Atlantic forests is detectable at multiple scales using satellite (Landsat TM) imagery.
Project PI:
Dr. Keith Eshleman
This research was funded with grant support from NSF-Ecosystems, EPA-STAR, NASA-IDS and The Nature Conservancy.
Return to Top