Geoscience Reference
In-Depth Information
class 82 (Cultivated Crops). The NLCD percent impervious surface cover map was
used to identify areas of human development as a more continuous variable
(ranging from 10 to 100 % for each 30 m image pixel), while the continuous
percent tree cover reflected the comparable proportion of land cover in a more
naturally forested state (again ranging from 10 to 100 %).
4.3 Approach
4.3.1 Watershed Delineation
Our research involved both local high-resolution elevation data sets and coarser
nationwide data sets to delineate the catchment areas upstream of our sample
points. Where high-resolution elevation data were available, including Rhode
Island and southeastern MA, a digital elevation model of 5 m horizontal spatial
resolution was built. Where these data were unavailable, we used the National
Elevation Data set (
http://ned.usgs.gov
), which has a 30 m spatial resolution.
The areas upstream of each sample point in the HBI and EPT analyses were
defined in two ways. One was to generate the area of contribution (catchment) to
each sample point, with additional areas of contribution for sample point upstream
along the same stream reach. The other was to combine (aggregate) those sample
points along the same stream reach and link only the biological sample data to the
lowest sample point (highest stream order) along the reach. These are referred to as
the
catchment
and
aggregated
watershed sampling schemes throughout the remain-
der of the paper. Each of the watershed sampling schemes were generated
using standard GIS hydrology functions/tools available in ESRI
®
GIS software.
These tools derive watershed boundaries based on topographic variables (slope,
aspect and elevation) derived from digital elevation data sets.
4.3.2 Landscape Distance Weighting
Weighting land cover metrics with some form of inverse distance weighting has
been explored elsewhere in the context of land cover assessments of stream biota
(e.g. King et al.
2005
; Baker et al.
2006
; Snyder et al.
2005
). A landscape distance
weighting methodology was designed to capture the relative importance of land
cover information based on its proximity to the stream channel (after Goetz and
Fiske
2008
). For each watershed in the study area, a distance-weighted surface was
developed, allowing us to use the distance from the stream channel and catchment
“pour point” independent of adjacent watersheds. The first component of the
