digital data, ArcHydro (an add-on data modeling application
for
GIS
) has the potential to integrate stormwater infrastructure
with terrain models and other geospatial layers, which will
boost analytic power of urban watershed evaluation, planning,
and management (Maidment, 2004; Nelson, 2009).
This study presents an analysis of an urban drainage
system to illustrate the impact of urban development upon
conventional delineation involving natural drainage systems.
The approach addresses urban watershed delineation and
integrates stormwater network systems and aerial photos with
geospatial techniques, using both vector and raster analysis.
We hypothesize that using these additional techniques, for
a complete evaluation in urban settings, would produce a
watershed area and boundary that varied significantly from
that defined by applying techniques accepted for delineation
of natural watersheds. Thus, our study illustrates specific ex-
amples of how application of standard geospatial techniques
developed, for identification of watershed and water flow in
natural areas, is insufficient in urban environments.
Study Area
A case study for a small urban watershed in Fairfax County,
Virginia illustrates the value of applying different methods to
delineate the watershed and flow network, and forms the basis
for comparative analysis. The Flatlick Branch of the Cub Run
watershed in Fairfax County, Virginia is our study site. This
stream and watershed are located in the extensively urbanized
northern Virginia area, just outside of Washington, D.C. Spe-
cifically, the stream is in northwestern Fairfax County, very
close to Washington Dulles International Airport (Figure 1).
Figure 2 depicts the spatial distribution of general land use
categories within the watershed. Land use in the watershed is
generally residential (64 percent) represented by single family
homes with a mixture of low and medium density. The water-
shed’s second major land use is categorized as open space (23
percent), however, a golf course (center of the image) repre-
sents almost half of the open space category. The watershed
has smaller areas of commercial and industrial land uses;
dispersed forest patches are present in all land use categories.
Extensive data are available for Fairfax County watersheds,
as the US Geological Survey (
USGS
) has numerous streams
gauged in this area for water quantity and quality assess-
ments. The
USGS
office in Richmond, Virginia provided the
data used in this analysis:
• 2009 natural color aerial photos at 15 cm resolu-
tion, acquired by the Virginia Base Mapping Program
through a contract with the Sanborn Map Company.
Virginia acquires orthoimagery across different loca-
tions over the Commonwealth on a four-year rotating
basis. The 2009 flyover included the northern Virginia
metropolitan area and the imagery was acquired using
North American 1983
HARN
Datum (Virginia Geograph-
ic Information Network 2011).
• Raster dataset for an elevation model generated from
lidar with one-meter2 resolution. The lidar acquisi-
tion was accomplished in 2009 and delivered to Fairfax
County as several preprocessed datasets including the
elevation model used in this analysis.
• Line shapefile for field documented above-ground loca-
tions of the stream channel(s), completed by Fairfax
County in 2008.
• Polygon shapefile, for the watershed, digitized by ana-
lysts at the
USGS
from their hand delineation completed
from a topo map.
• Point shapefile for the location of the
USGS
-Flatlick
Branch stream gauge; the gauge was installed in 2007
by the
USGS
.
• Shapefiles for all known or surveyed stormwater pipes
(lines), inlets - manholes and curbside (points), and re-
tention ponds (polygons) for Fairfax County, completed
by Fairfax County in 2005.
• General land use in a polygon shapefile, created by
Fairfax County in 2007.
Methods
First, comparing the lidar-watershed delineation to the
USGS
-
derived watershed was considered necessary to determine
which watershed model would be the best to use for our
analysis. The
Spatial Analyst Hydrology
toolset in Esri’s Arc-
Map
®
was used to process the lidar-elevation model into wa-
tershed and flow accumulation raster datasets. Figure 3 shows
the process, each tool used, and the resulting raster layer.
The
USGS
-stream gauge (point) shapefile was used as the pour
point input in the final step of the watershed delineation.
Several flow accumulation datasets, with varying thresholds,
were created with
Raster Calculator
to help evaluate the flow
of water across the watershed landscape. A contour line file
was also created using the
Spatial Analyst/Contour
tool, with
contour spacing of one meter.
In
GIS
, the point, line, and polygon stormwater network
shapefiles were overlaid on the lidar-derived watershed, along
with the lidar-contour lines, stream channel shapefile, and
lidar-derived flow accumulation raster datasets, to evaluate
how the stormwater networks would influence and impact
water flow within the watershed. We were specifically look-
ing for three situations:
1. stormwater networks that do not connect to surface
drainage ways, thus are not part of the “naturally” de-
lineated watershed and act as isolated catchment areas;
2. stormwater pipes that discharge outside the watershed
(de facto decrease in drainage area); and
3. stormwater pipes that drain into the watershed (de
facto addition to drainage area).
The aerial photos were added to our map as the bottom
layer of the ArcMap
®
window, and layers were turned on and
off as needed to assist with the visual analysis.
First, the stormwater facilities were evaluated under situ-
ation 1 (above), i.e., for detention versus retention ponds
(detention ponds are designed to only slow the flow of water
into natural stream channels, whereas retention ponds hold
water and prevent the flow going into natural channels). For
any locations determined as detention ponds, no changes
were made to the original lidar-watershed delineation. Any
areas identified as retention ponds, thus acting as isolated
catchment areas, were selected and exported into a new poly-
gon shapefile. Since the pour point (the ArcGIS term for the
T
able
1. R
esearch
including
S
tormwater
N
etwork
S
ystems
in
T
heir
I
dentification
of
U
rban
W
atershed
B
oundaries
Author Year
Method
Sample
et al
., 2001 Gathered vector based data from an on-the-ground analysis for a 43 acre neighborhood.
Rodriguez
et al
.,
2003
Used a land-based survey to delineate three specific catchment areas (between 18 and 180 hectares) and then added
stormwater sewers to analyze water flow.
Lhomme
et al
., 2004 Established a DEM delineated flow path, overlaid the stormwater drainage system and calculated the change in flow path
Amaguchi
et al
.,
2012
Vector-based urban landscape delineation and divided study area into city blocks to evaluate water flow within each
block. Also split water flow into above surface and below surface, eventually all flowing into the river channel.
366
May 2015
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
