United States
Environmental Protection
Agency
Office of Research and
Development
Washington, D.C. 20460
EPA/600/R-01/069
September 2001
xvEPA
Savannah River Basin
Landscape Analysis
North Carolina
South Carolina
Georgia
dam
1-95
252LEB01 RPT * 9/18/01
-------
EPA/600/R-01/069
September 2001
Savannah River Basin
Landscape Analysis
By
Deborah J. Chaloud, Curtis M. Edmonds, and Daniel T. Heggem
U.S. Environmental Protection Agency
Office of Research and Development
National Exposure Research Laboratory
Environmental Sciences Division
Landscape Ecology Branch
Las Vegas, Nevada
-------
Notice
The U.S. Environmental Protection Agency (EPA), through its Office of Research and Development
(ORD), funded and performed the research described here. It has been peer reviewed by the EPA and
approved for publication. Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
-------
Acknowledgments
Data processing and analysis support was provided by Lockheed-Martin under Contract 68-C5-0065 to
the U.S. Environmental Protection Agency. In particular, the contributions of Lee Bice, Karen Lee, and
Dick Dulaney of Lockheed-Martin were essential in the preparation of analyses used in this report.
-------
Table of Contents
Paee#
Acknowledgments iii
i
Section 1 - Introduction 1
Section 2 - Methods 3
2.1 Data Sets Used 3
Section 3 - Results 4
3.1 Sampling Site Ranking, Selection, and Drainage Area Creation 4
Section 4 - Landscape Assessment 15
4.1 HUC Metrics 15
4.2 Subbasin Metrics 20
4.3 Sampling Site Drainage Landscape Metrics 26
4.4 Landscape Change 37
IV
-------
List of Tables
Table page #
1 Physical Dimension Statistics for 8-digit HUCs 15
2 Forest Cover Types, Percent Cover by HUC 16
3 Agricultural Land Cover Types, Percent Cover by HUC 16
4 Urban Land Cover Types, Percent Cover by HUC 17
5 Other Land Cover Types, Percent Cover by HUC 17
6 Riparian Corridor Land Cover Types, Percent by HUC 18
7 Agriculture on Slopes and Erodible Soils, Percent by HUC 19
8 Agriculture-Related Metrics in Riparian Corridors, Percent Buffered Area 19
9 Roads Crossing Streams and Impoundments 20
10 Physical Dimension Statistics for Selected Subbasins 21
11 Land Cover Types for Selected Subbasins, Percent Area 22
12 Land Cover Types for Selected Subbasin Riparian Corridors, Percent Area 23
13 Agriculture-Related Metrics for Selected Subbasins and Riparian Corridors 24
14 Roads Crossing Streams and Impoundments for Selected Subbasins 24
15 Physical Dimension Statistics for Sampling Site Drainages, Percent Area 26
16 Aggregated Land Cover Types for Sampling Site Drainages, Percent Area 27
17 Aggregated Land Cover Types for Sampling Site Riparian Corridors, Percent Area 27
18 Agriculture-Related Metrics for Sampling Site Drainages and Riparian Corridors,
Percent Area 28
-------
List of Tables, Continued
Table Page #
19 Roads Crossing Streams and Impoundments for Sampling Site Drainages 29
20 Correlation of Aquatic and Landscape Metrics for Sample Site Drainages and Riparian
Corridors 36
21 Landscape Change for Selected Subbasins 41
i
22 Landscape Change for Sampling Site Drainages 42
VI
-------
List of Figures
Figure Page #
1 Savannah River Basin 1
2 Ecoregions of the Savannah River Basin 2
3 Sampling Site Locations 4
4 HUC 3060101 Sampling Site Results 5
5 HUC 3060102 Sampling Site Results 6
6 HUC 3060103 Sampling Site Results 7
7 HUC 3060104 Sampling Site Results 8
8 HUC 3060105 Sampling Site Results 9
9 HUC 3060106 Sampling Site Results 10
10 HUC 3060107 Sampling Site Results 11
11 HUC 3060108 Sampling Site Results 12
12 HUC 3060109 Sampling Site Results 13
13 HUCs and Subbasins 14
14 Roads Crossing Streams and Dams in Selected Subbasins 25
15 Sampling Site S68 Drainage 30
16 Sampling Site SI 13 Drainage 31
17 Sampling Site S195 Drainage 32
18 Sampling Site S22 Drainage 33
VII
-------
List of Figures, Continued
Figure Page #
19 Sampling Site S80 Drainage 34
20 Sampling Site S149 Drainage 35
21 Mosaic of Circa 1970 NALC Images 38
i 22 Mosaic of Circa 1992 NALC Images 39
23 Gains (green) and Losses (red) in Vegetation, 1970s to 1990s 40
24 Lake Russell was Created between the 1970s and 1990s 41
25 Land Surface Loss to Lake Russell in Subbasin #26 41
VIII
-------
Section 1
Introduction
Scientists from the U.S. Environmental Protection Agency (EPA), Region 4, Science and Ecosystem
Support Division, enlisted the assistance of the landscape ecology group of U.S. EPA, Office of
Research and Development (ORD), National Exposure Research Laboratory, Environmental Sciences
Division (ESD), in conducting a landscape assessment of the Savannah River Basin as part of their
ongoing Regional Environmental Monitoring and Assessment Program (REMAP) demonstration project.
In the Scope of Work provided by Region 4, the goal was stated as "provide technical/scientific
assistance ...to EPA Region 4 in assessing current wadeable stream conditions in the Savannah River
Basin with landscape factors that may be contributing to these conditions or gradients." Three specific
objectives were presented in the form of questions. These were:
• Are both the proportions of land uses and the
spatial pattern of land uses important for
characterizing and modeling stream condition in
watersheds/ecoregions of different areas?
• Can land uses near the streams better account for
the variability in ecological condition than land use
for the entire watershed/ecoregion?
• Does the size of the watershed/ecoregion influence
statistical relationships between landscape
characteristics and ecological condition?
In addition, an assessment of landscape change was to
be conducted as part of continuing ESD research in
application of change detection techniques.
The data analysis plan developed to address the
objectives given above called for calculation of a
specific suite of landscape metrics for all nine United
States Geological Survey (USGS) 8-digit hydrological
unit codes (HUC; USGS, 1982), a selected subset of
the 94 Georgia and South Carolina subbasins, and the
riparian corridors in the HUCs and selected subbasins.
The subbasins, shown in Figure I, are generally
equivalent in area to USGS 11-digit HUCs. The
riparian corridor was defined as 100 meters on either
North Carolina
I-85
South Carolina
I-20
Georgia
I 95
Figure 1. Savannah River Basin.
-------
Blue Ridge Mountains
side of stream arcs; this size was selected from a review of state laws and literature available on the
Internet (e.g., Santa Cruz County, 1998; U.S. EPA, 1998; South Carolina Department of Natural
Resources, 1998). The suite of metrics
included: landcover types, u-index,' agriculture
on slopes greater than 3 percent, agriculture on
highly erodible soils, agriculture on moderately
erodible soils, agriculture on highly erodible
soils with slopes greater than 3 percent, number
of occurrences of roads crossing streams, and
number of impoundments. Landscape metric
statistics were also computed for the drainage
areas and associated riparian corridors of a
selected set of sites sampled by Region 4 using
REMAP protocols. Region 4 provided an
ARC/INFO coverage of the sampling locations
and QuatroPro spreadsheets of the water
quality and biotic measurements.
The Savannah River Basin is arrowhead-
shaped, trending generally northwest to
southeast. The basin is comprised of nine
USGS 8-digit HUCs (numbered 3060101
through 3060109, hereafter referred simply by
the last digit), spanning three ecoregions: Blue
Ridge, Piedmont and Coastal Plains. As shown
in Figure 2, HUCs 1 and 2 are primarily in the
Blue Ridge ecoregion, HUCs 3, 4, 5, and 7 lie
in the Piedmont, and the majority of HUCs 6, 8,
and 9 are in the Coastal Plain.
Piedmont
Coastal Plain
Atlantic Coastal Plain
Figure 2. Ecoregions of the Savannah River Basin.
1 A measure of anthropogenic influence, comprised primarily of agricultural and urban
landcover classes.
-------
Section 2
Methods
The selected landscape metrics are identical to, or based on, metrics used in the mid-Atlantic atlas (Jones
et al, 1997). In the atlas, the metrics were calculated only for 8-digit HUCs; in this study, metrics are
additionally calculated for smaller spatial units. The basic methodology is the same, however. In
general, calculation of the landscape metrics involves ARC/INFO techniques of extracting or "cookie
cutting" the desired area from a spatial data set. The data are formatted in an ARC/INFO grid of uniform
cell size. In this study, a 30-m cell size is used for all grids. For metrics which are produced from more
than one data set (e.g., roads crossing streams), ARC/INFO overlay and intersection techniques are used.
A few metrics, used only on the drainage areas of the individual sampling sites, are produced from an in-
house custom statistics program. These are metrics of fragmentation, i.e., the degree to which landcover
types are present in patches rather than in continuous, homogenous blocks. The landscape change metric
is produced from comparison of satellite imagery from two dates. This is the only metric which does not
use ARC/INFO as the primary data analysis software. Landscape change assessment employs ENVI, an
image processing software package available for PC or Unix systems.
2.1 Data Sets Used
The spatial data sets used are obtained from a variety of sources. The primary data sets used in this
landscape assessment include: Multi Resolution Land Characteristics (MRLC) Interagency Consortium
land cover/land use (Bara, 1994), State Soil Geographic data base(STATSGO)soils (Natural Resources
Conservation Service, 1996), RF3 streams (U.S. EPA, 1997), USGS 8-digit HUCs, Georgia and South
Carolina subbasins, Region 4 sampling site locational and sampling data, 30-m and 100-m digital
elevation models (DEM; USGS, 1990), digital line graph (DLG) roads (USGS, 1989), and National
Inventory of Dams impoundments (U.S. Army Corps of Engineers, 1997). Landscape change assessment
used North American Land Characterization (NALC) imagery from the 1970s and 1990s (U.S. EPA,
1993). Data sets were subset to the area of interest using the basin boundary coverage.
-------
Section 3
Results
3.1 Sampling Site Ranking, Selection, and Drainage Area Creation
A simple, unweighted scoring system was used to rank the sampling sites
(n = 94), shown in Figure 3, by their results. Water quality variables (pH,
dissolved oxygen, conductivity) and biota [algal growth potential test
(AGPT); Ephemeroptera, Plecoptera, and Trichoptera index (EPT); fish
index of biological integrity (fish_ibi), microinvertebrate habitat,
and microinvertebrate richness] were scored separately. The
frequency distributions for each variable was examined. Most
indicated a bimodal distribution, with reduced frequencies near the
lower and upper ends of the variable's range. Measurement
values corresponding to the inflection points of the curve
were selected to divide the range into three classes. A
score value was ascribed to the measurement value, 1
for bad, 2 for fair, 3 for good, and 0 for missing data.
Although these are labeled as good, fair, and bad,
these terms apply to the measurement value
compared to the range of measurement values, not
to any applicable water quality standards or
other measurement system. The scores were
summed and recorded. The number of
measurements used in the summation was
also recorded; this was necessary because of
the large number of sites missing results for
one or more variables. The measurement
data and scoring data were then
associated with the site location
coverage. Map compositions were
prepared for each HUC, presented here
as Figures 4 through 12. The figures
were useful in characterizing relative
conditions across the basin and
making preliminary decisions
about areas for further
investigation.
Figure 3. Sampling Site Locations.
-------
I WntrQuilily
* dnJ
Fair
• Sri
• Note
A Dm
Figure 4. HUC 3060101 Sampling Site Results.
-------
HA
ay
Biology Scott
WwQualily
Good
fm
lit
No Dm
Stem
bnut
US Higk«,
Figure 5. HUC 3060102 Sampling Site Results.
-------
ffiAN
Biology Scott
**Quil«>
Good
U
U
Noftu
ta
SuraJii
IwmUK
lSlli.h*iv
Riilnad
Figure 6. HUC 3060103 Sampling Site Results.
-------
HABERSI
WmrQuilky
Cool
Fair
Bid
No Dm
Dm
SH»
Illmult
US Highly
Figure 7. HUC 3060104 Sampling Site Results.
-------
V
Biota) fait
Watt QM|«,
fad
M
U
talljo
DM
Figure 8. HUC 3060105 Sampling Site Results.
-------
Cool
Fa
Bid
Note
Da
Sunn
Innae
ISM.iy
kill. .1,1
Figure 9. HUC 3060106 Sampling Site Results.
10
-------
•
A
Bute S,-,«t
WntrQwIiy
Ond
fm
lu
No Dm
Stan
USHijk»iv
Figure 10. HUC 3060107 Sampling Site Results.
11
-------
H>i«", Score
»'» Quality
Goal
Nt
B«l
total
ta
Figure 11. HUC 3060108 Sampling Site Results.
The sampling locations were selected by the Region using the EMAP site selection protocol. Several
discussions and correspondences were conducted with a lead EMAP Statistician, Dr. Tony Olson, about
the spatial area represented by the sampling sites. It was determined that it would be necessary to
develop the specific drainage area of each sampling location and to treat the water quality and biota
information as point data. Accurate drainage area computation requires OEMs of 30-m intervals or
better; at the time of analysis, these were available for only portions of the Savannah River Basin,
primarily the north end and part of the central area.
The process used to delineate the drainage areas employs hydrological analyses tools contained in the
Grid module of ARC/INFO. First sinks in the OEMs are identified and filled. Flow direction is
computed as the direction from each 30-m cell towards its steepest downslope neighbor. From the flow
direction grid, a flow accumulation grid is created by calculation of the number of cells which flow into
each downslope cell; this grid resembles the existing stream network. The sampling station locations are
input as pour points. In some cases, the sampling point coordinates did not fall directly on a flow
accumulation path; in these instances, the pour point was placed on the flow accumulation in the cell
nearest to the given station coordinates.
12
-------
In the selection of the subset of sites for landscape metric assessment, efforts were made to select sites
that met the following criteria: 1. Full suite of stream measurement variables, 2. Located in the areas
indicated to be of greatest interest to the Region, 3. 30-m DEM data available to use in drainage area
determination, 4. Representation of the full range of condition gradient measurement values, and 5.
Representation of first through third stream order classes. Using these criteria, sixteen sites were
selected.
•
•
•
A
Good
M
NoDn
Itm
Figure 12. HUC 3060109 Sampling Site Results.
13
-------
3060102
3060101
3060103
3060104
The selection of subbasins for presentation of landscape
metrics was made after selection of the sampling sites.
The selected subbasins are all in HUC 3060103
and each includes one or more of the sampling
site subset. This provides the nested hierarchy
of spatial units in the assessment. An arbitrary
number was assigned to each subbasin after
merging the separate Georgia and
South Carolina coverages. The
subbasins are shown in Figure 13.
3060107
3060105
3060106
3060109
Figure 13. HUCs and Subbasins.
14
-------
Section 4
Landscape Assessment
4.1 HUC Metrics
The landscape assessment was begun by an analyses of the 8-digit HUCs. As shown in Table I, the size
of the HUCs varies from 200,987.55 ha (HUC 7) to 488,842.20 ha (HUC 6). Associated riparian areas
vary from 31,324.14 ha (HUC 7) to 88,651.85 ha (HUC 3), based on a 100-m corridor on either side of
all RF3 stream arcs.
Table 1. Physical Dimension Statistics for 8-digit HUCs
HUC
3060101
3060102
3060103
3060104
3060105
3060106
3060107
3060108
3060109
Total Area
(ha)
272,812.23
258,218.91
483,189.03
398,298.06
204,446.97
488,842.20
200,987.55
220,108.41
248,158.71
Riparian
Corridor
(ha)
55,585.95
54,114.18
88,651.85
65,842.94
32,453.26
83,668.92
31,324.14
37,124.59
47,316.49
Stream
Length
(km)
3066.68
2994.61
4803.99
3463.28
1636.33
1765.63
4771.76
2044.32
2679.38
Stream
Density
(m/ha)
11.24
11.60
9.94
8.70
8.00
3.61
23.74
9.29
10.80
Landcover types are derived from MRLC data, nominal base year 1992. Differences among the three
ecoregions are evident in the forest landcover statistics for the HUCs, Table 2. Deciduous and evergreen
forests predominant in HUCs 1 through 5 and 7, the HUCs comprising the Blue Ridge and Piedmont
ecoregions; all forest types account for 64.7 to 83.49% of the total land cover. Forest landcover accounts
for 37.20 to 53.35% of the landcover in the Coastal Plain HUCs, with evergreen forests the predominant
forest type. Wetland landcover types are found primarily in the Coastal Plain HUCs, accounting for
11.10 to 35.93% of the total landcover, most of it in woody wetlands. Wetlands comprise less than one
percent of the landcover in the HUCs outside the Coastal Plains.
15
-------
Table 2. Forest Cover Types, Percent Cover by HUC
HUC
3060101
3060102
3060103
3060104
3060105
3060106
3060107
3060108
3060109
Evergreen
23.36
25.66
28.35
23.72
39.95
33.39
50.21
24.17
25.24
Mixed
10.45
12.15
11.07
9.65
8.85
7.22
9.72
7.50
4.63
Deciduous
37.92
45.68
25.28
38.66
28.57
12.74
18.69
15.38
7.33
Woody
Wetlands
0.62
0.27
0.55
0.36
0.69
11.54
0.74
10.86
31.46
Agricultural landcover types, Table 3, comprise 9.91 to 32.47% of the total landcover in each HUC.
Pasture/hay is the dominant agricultural land use in the upper part of the basin, while row crops are the
largest agricultural land use in the lower basin. Urban landcover types, Table 4, account for between
0.85 to 5.33% of the total land use in all HUCs. There is no ecoregion-related pattern to the distribution
of urban landcover. Barren landcover types, Table 5, comprise less than one percent of the total
landcover in HUCs 1 and 2, and approximately 2 to 10 percent of the landcover of the Piedmont and
Coastal Plains HUCs.
Table 3. Agricultural Land Cover Types, Percent Cover by HUC
HUC
3060101
3060102
3060103
3060104
3060105
3060106
3060107
3060108
3060109
Pasture/Hay
10.18
6.76
13.21
15.51
4.01
1.60
3.32
2.76
1.78
Row Crops
5.05
3.46
9.08
7.59
6.90
14.60
6.59
29.71
14.15
Other Grasses
0.76
0.28
0.48
0.35
0.12
0.46
0.09
0.06
0.38
16
-------
Table 4. Urban Land Cover Types, Percent Cover by HUC
HUC
3060101
3060102
3060103
3060104
3060105
3060106
3060107
3060108
3060109
Low
Intensity
Residential
3.35
1.03
1.74
0.88
0.60
2.72
0.71
0.49
1.03
High
Intensity
Residential
0.29
0.06
0.25
0.06
0.08
1.01
0.10
0.10
0.64
High Intensity
Commercial/
Industrial
1.16
0.42
0.60
0.49
0.32
1.60
0.30
0.26
1.18
Table 5. Other Land Cover Types, Percent Cover by HUC
HUC
3060101
3060102
3060103
3060104
3060105
3060106
3060107
3060108
3060109
Water
6.41
3.61
6.60
0.57
4.87
1.46
0.39
0.47
3.10
Emergent
Wetlands
0.04
0.07
0.03
0.02
0.03
0.48
0.03
0.24
4.47
Barren:
Quarries/Strip
Mines
0.19
0.04
0.14
0.14
0.19
0.60
0.07
0.54
0.14
Barren: Bare
Rock/Sand
<0.01
<0.01
<0.01
<0.01
<0.01
0.01
<0.01
<0.01
0.05
Barren:
Transitional
0.22
0.49
2.63
1.98
4.81
10.57
9.04
7.45
4.40
17
-------
The patterns of landcover/land use within the riparian corridors, Table 6, are not substantially different
than those for the HUCs overall, with the exception that water is an appreciable percentage of the
landcover within riparian corridors in most HUCs. Agricultural land use within the riparian corridor
ranges from 4.63% to 12.78% and urban land use ranges from 0.33% to 3.51%. Barren landcover ranges
from less than 1% to a little more than 6%. The predominant landcover types in the HUC riparian
corridors are forest and wetlands in the Coastal Plains and forest in the other ecoregions.
Table 6. Riparian Corridor Land Cover Types, Percent by HUC
HUC
3060101
3060102
3060103
3060104
3060105
3060106
3060107
3060108
3060109
Forest
71.96
79.06
70.63
83.20
76.83
48.79
86.00
48.53
24.13
Agriculture
8.16
7.03
10.10
12.16
4.84
6.54
4.63
12.78
6.38
Urban
3.19
1.31
1.38
0.76
0.57
3.51
0.40
0.33
1.34
Wetland
1.92
1.09
1.77
1.21
2.23
28.90
1.99
32.43
57.23
Barren
0.44
0.33
1.26
0.64
4.17
6.23
5.51
4.20
2.48
Water
14.51
11.20
14.86
2.04
11.33
6.06
1.50
1.77
8.44
While there is some variation in landcover types among the three ecoregions, overall the HUCs are
relatively homogeneous in landcover/land use pattern. In all HUCs, natural landcover types comprise
greater than 50% of the total landcover. Urban land uses account for only a small percent of the total
landcover and agricultural uses account for 1/10 to approximately 1/3 of the total land cover/land use.
These results contrast greatly with the results obtained for 8-digit HUCs in the mid-Atlantic region (Jones
et al, 1997), where large differences were evident at this scale. The broad-scale patterns evident in the
mid-Atlantic (e.g., intensive urbanization of the Coastal Plains, concentrated agricultural land uses in
valleys, and isolation of forests to highland areas) are not in evidence in the Savannah River Basin.
The proportion of agriculture on slopes greater than 3% grade has been developed as a landscape metric
because the potential for erosion increases significantly at this grade. Similarly, agriculture practiced on
highly or moderately erodible soils has a higher potential for erosion. These metrics are developed from
overlays of OEMs, MRLC land cover/land use, and credibility factors contained in the STATSGO soils
data base. Results for all of these metrics are generally low, as shown in Table 7. Only HUCs 3 and 4
showed greater than 20% total land area for any of the agriculture-soil-slope metrics, that being
agriculture on moderately erodible soils, most of it in pasture/hay. Results for these metrics within the
riparian corridors are lower, ranging from nonexistent to less than 12% agriculture on moderately
erodible soil in HUC 4, as shown in Table 8.
18
-------
Table 7. Agriculture on Slopes and Erodible Soils, Percent by HUG
HUC
3060101
3060102
3060103
3060104
3060105
3060106
3060107
3060108
3060109
Pasture/Hay
on Slopes
>3%
2.24
1.33
1.25
2.25
0.22
0.13
0.10
0.03
<0.01
Row Crops
on Slopes
>3%
0.98
0.67
0.75
0.95
0.42
0.72
0.26
0.47
0.02
Pasture/Hay
on
Moderately
Erodible
Soils
10.03
6.54
12.51
15.24
2.22
0.22
4.20
0.48
0.41
Row Crops
on
Moderately
Erodible
Soils
4.96
3.36
8.41
7.45
3.84
1.67
3.06
3.23
2.51
Pasture/Hay
on Highly
Erodible
Soils
—
—
0.61
0.26
1.76
<0.01
—
—
—
Row Crops
on Highly
Erodible
Soils
—
—
0.58
0.14
2.71
0.02
—
—
—
Table 8. Agriculture-Related Metrics in Riparian Corridors, Percent Buffered Area
HUC
3060101
3060102
3060103
3060104
3060105
3060106
3060107
3060108
3060109
Pasture/Hay
on Slopes
>3%
1.04
0.95
0.51
1.00
0.11
0.05
0.03
0.01
<0.01
Row Crops
on Slopes
>3%
0.55
0.48
0.33
0.47
0.16
0.23
0.12
0.10
<0.01
Pasture/Hay
on
Moderately
Erodible
Soils
4.66
4.08
5.68
7.98
0.73
0.21
1.88
0.25
0.21
Row Crops
on
Moderately
Erodible
Soils
2.50
1.94
3.53
3.89
1.46
1.17
1.78
1.35
1.30
Pasture/Hay
on Highly
Erodible
Soils
—
—
0.20
0.08
0.98
<0.01
—
—
—
Row Crops
on Highly
Erodible
Soils
—
—
0.21
0.07
1.56
0.01
—
—
—
19
-------
Roads frequently cause increased runoff to streams and contribute pollutants washed off the road
surfaces. This phenomenon is represented by the roads-crossing-streams metric, computed from
intersecting digital line graph roads with RF3 stream arcs. As shown in Table 9, values for this metric
range from 362 in HUC 5 to 1,914 in HUC 6. Normalizing these values to the number of road crossings
per stream kilometer, as shown in Table 9, shows the greatest frequency of roads crossings per stream
kilometer is in HUC 6 with more than one road crossing per kilometer of stream length. The lowest
frequency is in HUC 7 with approximately one road crossing for every 8 kilometers of stream length.
The remaining HUCs have frequencies in the range of one road crossing for every 2.5 to 5 kilometers of
stream length.
Table 9. Roads Crossing Streams and Impoundments
HUC
3060101
3060102
3060103
3060104
3060105
3060106
3060107
3060108
3060109
Road
Crossings
1235
964
1487
1227
362
1914
637
842
723
No.
Crossings/Stream
km
0.40
0.32
0.31
0.35
0.22
1.08
0.13
0.41
0.27
Dams
117
58
98
102
35
191
60
52
31
No.
Dams/Stream
km
0.038
0.019
0.020
0.029
0.021
0.108
0.013
0.025
0.012
Information for dams was obtained from the National Inventory of Dams which tracks all dams greater
than 6 feet in height for inspection purposes. As shown in Table 9, the fewest number of dams in any
HUC is 31 in HUC 9 while the greatest number is 191 in HUC 6. Normalizing by the total stream length
within each HUC shows the greatest frequency of dams is also in HUC 6, with one dam for every 9
kilometers of stream length. The lowest frequencies of dams are in HUC 9 and HUC 7, with roughly one
dam for every 80 kilometers of stream length. The locations of dams are depicted in Figure 1.
4.2 Subbasin Metrics
As discussed above, the landscape metrics at the HUC level show some variation among HUCs
attributable to natural landcover variation at the ecoregion level. However, the patterns of land use are
generally consistent across ecoregions and among HUCs. This section focuses on the next scale, the
subbasin. Landscape metrics are presented for several subbasins of HUC 3. These particular subbasins
were selected because they each contain one or more of the sampling sites selected for analysis. The
landscape metrics produced for the subbasins are the same as those produced for the HUCs.
20
-------
Physical dimensions of the selected subbasins are shown in Table 10. The total land area in each
subbasin ranges from 17,195.76 ha in #32 to 68,295.33 ha in #53. The associated riparian corridors
range from 4,311.00 to 12,800.34 ha.
Table 10. Physical Dimension Statistics for Selected Subbasins
Subbasin
20
26
32
36
53
Total Area
(ha)
55,797.39
53,225.73
17,195.76
61,462.62
68,295.33
Riparian
Corridor
(ha)
9,089.19
10,089.00
4,311.00
9,704.07
12,800.34
Stream
Length
(km)
486.79
528.63
255.27
499.79
695.35
Stream
Density
(m/ha)
8.72
9.93
14.84
8.13
10.18
The landcover statistics for HLJC 3 overall are 64.70% forest (approximately 28% evergreen, 25%
deciduous and 11% mixed forest), 22.29% agriculture (approximately 13% pasture/hay and 9% row
crops), 2.59% urban, 6.60% water, approximately 3% barren, and less than one percent wetlands.
Among the subbasins, the forest landcover classes vary from 40.26% in #32 to 73.66% in #53. As shown
in Table 11, evergreen forests are the largest forest class in #26, #36, and #53; deciduous is the largest
class in #20 and #32. Agricultural land use in #26 is about the same as in the HUC overall (23.65% of
which approximately 16% is in pasture/hay). Greater agricultural land use is evident in #20 (34.07%
with about 20% in pasture/hay) and #32 (32.45% of which almost 18% is pasture/hay). Less landcover is
in agricultural land uses in #36 (15.16%, with more than 8% pasture/hay) and #53 (11.58%, with row
crops slightly exceeding pasture/hay). Urban land use is lowest in #53 at less than one percent and
highest in #20 at 8.05%. The remaining three subbasins have urban land use in slightly higher
percentages than for the HUC overall, ranging 3.05% in #36 to 4.81% in #32.
21
-------
Table 11. Land Cover Types for Selected Subbasins, Percent Area
Land Cover Type
Water
Low Intensity Residential
High Intensity Residential
High Intensity Commercial/Industrial
Pasture/Hay
Row Crops
Other Grasses
Evergreen Forest
Mixed Forest
Deciduous Forest
Woody Wetlands
Emergent Wetlands
Barren: Quarries/Strip Mines
Barren: Transitional
Subbasin
20
1.30
5.60
0.83
1.62
20.23
13.84
1.53
16.09
9.55
28.30
0.57
0.03
0.26
0.24
26
8.11
2.87
0.57
1.23
16.33
7.32
0.95
25.32
11.00
21.12
0.75
0.02
0.14
4.27
32
20.59
3.34
0.43
1.04
17.54
14.91
0.75
10.53
5.96
23.77
0.82
0.11
0.21
<0.01
36
0.57
2.10
0.28
0.67
8.43
6.73
0.36
37.08
13.12
24.61
0.30
0.02
0.07
5.66
53
9.36
0.45
0.03
0.15
5.58
6.00
0.14
36.70
10.98
25.98
0.63
0.03
0.09
3.88
In the riparian corridors, forest comprises 53.02 to 85.52% of the total cover, with deciduous the most
dominant forest cover type, as shown in Table 12. Agricultural land use within the riparian corridor
ranges from 4.41% in #53 to 15.92% in #32 and urban land use comprises from 0.39% to 4.99% of the
total riparian land cover. Wetlands account for approximately 3% or less of the riparian land cover
types.
22
-------
Table 12. Land Cover Types for Selected Subbasin Riparian Corridors, Percent Area
Land Cover Type
Water
Low Intensity Residential
High Intensity Residential
High Intensity Commercial/Industrial
Pasture/Hay
Row Crops
Other Grasses
Evergreen Forest
Mixed Forest
Deciduous Forest
Woody Wetlands
Emergent Wetlands
Barren: Quarries/Strip Mines
Barren: Transitional
Subbasin
20
5.89
3.97
0.26
0.76
8.84
5.54
0.36
15.78
12.76
43.15
2.11
0.09
0.34
0.16
26
22.38
1.68
0.19
0.47
8.68
3.40
0.24
18.59
12.30
28.94
1.69
0.05
0.06
1.32
32
25.31
1.82
0.07
0.39
7.60
8.32
0.21
14.30
8.49
30.23
2.73
0.37
0.15
<0.01
36
2.25
1.36
0.08
0.20
3.84
2.84
0.05
25.59
15.37
44.56
0.79
0.04
0.03
2.99
53
18.58
0.31
<0.01
0.08
1.94
2.47
0.02
29.40
10.54
33.20
2.27
0.10
0.02
1.07
The agriculture-soil-slope metric results for HUC 3 are 2% agriculture on slopes greater than 3%,
approximately 21% agriculture on moderately erodible soils, approximately 1% agriculture on highly
erodible soils, and less than 0.1% agriculture on slopes greater than 3% in highly erodible soils. Among
the subbasins, #20, #26, and #32 have more agriculture on slopes greater than 3% and more agriculture
on moderately erodible soil than for the HUC overall; the remaining two subbasins are substantially
lower than the HUC overall for both these metrics, as shown in Table 13. Only #53 has any agriculture
on highly erodible soil (about 7%) and agriculture on slopes greater than 3% and highly erodible soils
(0.35%). Results for these metrics are lower for the riparian corridors, with only subbasins #20, #26, and
#32 having more than 10% riparian land cover in agriculture on moderately erodible soils.
Table 14 provides results for the number and frequency of roads crossing streams and dams (Figure 14).
Roads crossing streams ranges from 56 in #32 to 299 in #20. There are no dams in #32, but 19 dams in
#20. The frequency of roads crossing streams is highest in #20 with approximately one road crossing for
every 1.6 kilometers of stream length; the lowest frequency is in #53 with one road crossing per
approximately 9 kilometers of stream length. The frequency of roads crossing streams for the HUC
overall is approximately one crossing per 3 kilometers of stream length. The frequency of impoundments
for the HUC overall is approximately one dam for every 50 stream kilometers. The frequency of dams is
lower than for the HUC overall in #32 with no dams and in #53 with approximately one dam for every
167 kilometers of stream length. The greatest frequency of dams among the subbasins is in #20 with one
dam per approximately 25 stream kilometers.
23
-------
Table 13. Agriculture-Related Metrics for Selected Subbasins and Riparian Corridors
Subbasin
20
26
32
36
53
Percent of Subbasin Total Area
Agriculture on Slopes >3%
Agriculture on Moderately Erodible Soils
Agriculture on Highly Erodible Soils
3.54
34.03
—
2.71
23.53
—
4.29
31.64
—
1.17
15.15
—
0.60
4.39
7.11
Percent of Subbasin Riparian Corridor
Agriculture on Slopes >3%
Agriculture on Moderately Erodible Soils
Agriculture on Highly Erodible Soils
1.31
14.13
—
1.34
11.99
—
1.24
13.02
—
0.39
6.67
—
0.27
1.43
2.61
Table 14. Roads Crossing Streams and Impoundments for Selected Subbasins
Subbasin
20
26
32
36
53
Total Area
(ha)
299
227
56
170
82
Riparian Corridor
(ha)
0.61
0.43
0.22
0.24
0.11
Stream Length
(km)
19
15
0
13
4
Stream Density
(m/ha)
0.039
0.028
—
0.026
0.006
At this scale, patterns which may impact water quality begin to be evident. In Figure 6, the sampling
stations in #53 are indicated as fair to good (as compared to the overall data range). This subbasin has
the highest proportion of landcover in forest among the subbasins, the lowest proportion of agriculture
and urban land uses, and a low proportion of agriculture on slopes greater than 3%. Among the selected
subbasins, it has the lowest frequency of roads crossing streams. Although is the largest of the subbasins
in total area, this subbasin has only 4 dams. However, #53 is the only subbasin among those examined
with agriculture on highly erodible soils and agriculture on slopes greater than 3% and highly erodible
soils.
In contrast, the sampling sites in #32 and #20 rank as fair to bad compared to the overall data ranges.
These two subbasins contain the greatest proportion of agriculture among the subbasins, 28 to 33%
agriculture on moderately erodible soils, and 3 to 4% agriculture on slopes greater than 3%. In addition,
#20 has the highest proportion of urban land use, the highest normalized roads crossing streams value,
and the greatest frequency of dams among the selected subbasins.
24
-------
Figure 14 Roads Crossing Streams and Dams in Selected Subbasins.
25
-------
4.3 Sampling Site Drainage Landscape Metrics
One of the objectives of this project is to try to establish relationships among landscape metrics and
water quality/aquatic biota metrics. The water quality data were collected at specific sampling sites. To
investigate relationships with landscape metrics, it is necessary to delineate the drainage area to the
individual sampling site. This was done for a subset of 16 sampling sites. The selection process was
described earlier, as was the methodology for delineating the drainage areas.
The drainage areas for the sampling sites range from 122.58 to 10,665.18 ha, as shown in Table 15. In
delineating the drainage areas, the locations for the sampling sites frequently did not lie on a stream arc,
necessitating a best guess, based on the indicated stream order and proximity to stream arc, as to the
point on the arc to use as the pourpoint. In addition to the landscape metrics calculated for the HUCs and
subbasins, metrics of fragmentation were generated using a custom, in-house software program. For the
fragmentation metrics, the 15 landcover/land use classes of the MRLC data were aggregated to six
classes: water, urban, forest, agriculture, wetlands, and barren, as shown in Table 16 for the overall
drainage area and in Table 17 for the riparian corridors. In these aggregated land cover types, other
grasses are included in agriculture and woody wetlands are included in the wetlands cover type.
Table 15. Physical Dimension Statistics for Sampling Site Drainages, Percent Area
Site
S22
S27
S68
S80
S81
S95
S103
S113
S130
S149
S151
S155
S195
S197
S200
S216
Total Area
(ha)
973.98
4,950.90
468.09
6,499.71
6,612.21
10,665.18
572.76
747.00
1,169.73
776.52
1,076.22
2,556.72
4,279.41
122.58
1,798.47
551.16
Riparian Corridor
(ha)
149.85
939.78
88.65
884.16
908.01
1,727.73
69.30
83.79
163.80
139.50
191.88
381.06
860.94
43.56
377.37
116.19
Stream Length
(km)
7.34
47.57
4.34
44.90
45.85
89.30
3.32
4.51
8.68
6.87
9.53
18.98
46.05
2.06
19.11
5.69
Stream Density
(m/ha)
7.54
9.61
9.28
6.91
6.93
8.38
5.80
6.03
7.42
8.84
8.85
7.42
10.76
16.78
10.62
10.33
26
-------
Table 16. Aggregated Land Cover Types for Sampling Site Drainages, Percent Area
Site
S22
S27
S68
S80
S81
S95
S103
S113
S130
S149
S151
S155
S195
S197
S200
S216
Water
0.11
0.34
<0.01
0.56
0.55
0.62
<0.01
0.59
1.02
0.03
0.09
0.11
0.85
O.01
0.19
0.10
Urban
<0.01
0.34
<0.01
7.75
7.62
2.84
<0.01
2.80
<0.01
16.02
13.01
3.21
0.44
5.50
6.77
0.07
Agriculture
59.40
18.76
0.27
18.28
17.97
8.40
0.04
23.70
62.27
44.33
40.59
17.77
3.45
42.59
47.64
4.90
Forests
40.21
80.06
96.01
72.71
73.17
81.52
89.92
61.78
36.28
39.38
46.00
73.21
94.49
51.62
42.37
87.62
Wetlands
0.24
0.37
<0.01
0.45
0.44
0.87
0.08
4.56
0.40
0.10
0.19
0.56
0.04
0.29
0.23
0.13
Barren
0.04
0.11
3.73
0.26
0.25
5.75
9.96
. 6.55
0.03
0.13
0.10
5.14
0.73
<0.01
0.11
7.18
Table 17. Aggregated Land Cover Types for Sampling Site Riparian Corridors, Percent Area
Site
S22
S27
S68
S80
S81
S95
S103
S113
S130
S149
S151
S155
S195
S197
S200
S216
Water
<0.01
0.48
O.01
2.28
2.21
2.97
<0.01
4.19
6.21
<0.01
<0.01
0.07
4.21
<0.01
0.45
0.08
Urban
<0.01
0.05
O.01
6.17
6.02
2.06
<0.01
<0.01
<0.01
11.35
9.52
1.49
0.08
9.09
6.13
0.15
Agriculture
29.91
8.19
0.41
11.81
11.63
2.83
<0.01
4.72
25.60
20.70
17.82
6.55
7.07
10.75
26.18
2.47
Forests
69.37
90.41
99.19
78.70
79.14
88.22
95.84
60.36
66.60
67.93
72.47
89.14
87.39
80.17
66.71
92.56
Wetlands
0.72
0.82
<0.01
0.92
0.88
1.79
0.13
29.53
1.59
<0.01
0.19
0.64
0.09
<0.01
0.45
0.23
Barren
<0.01
0.06
0.41
0.11
0.11
2.14
4.03
1.18
<0.01
<0.01
<0.01
2.13
1.15
<0.01
0.07
4.49
27
-------
Results for agriculture-related metrics over the entire drainage area and the riparian corridor are
presented in Table 18. The number of road crossing streams and dams are shown in Table 19. Ten of the
16 sampling site drainages contain no dams; however, where dams are present, they are generally greater
in frequency than in the HUC or subbasins overall. The frequency of roads crossing streams ranges from
approximately one road crossing per 5.5 kilometers of stream length to a maximum of one road crossing
for every stream kilometer.
Table 18. Agriculture-Related Metrics for Sampling Site Drainages and Riparian Corridors, Percent
Area
Site
S22
S27
S68
S80
S81
S95
S103
S113
S130
S149
S151
S155
S195
S197
S200
S216
Total Area
Agriculture
on Slopes
>3%
3.91
0.26
0.19
1.57
1.54
0.28
—
2.14
9.59
1.85
2.35
1.07
0.64
9.92
4.25
0.09
Agriculture
on Moderately
Erodible Soil
59.40
18.69
0.27
17.72
17.42
5.87
0.04
—
62.27
41.40
38.36
17.33
3.37
42.59
45.51
4.90
Agriculture
on Highly
Erodible Soils
—
—
—
—
—
—
0.04
—
—
—
—
17.33
—
—
—
—
Riparian Corridor
Agriculture
on Slopes
>3%
2.22
0.03
0.30
0.66
0.65
0.02
—
2.04
1.70
0.32
0.28
0.02
0.96
0.21
2.43
0.15
Agriculture
on Moderately
Erodible Soils
29.91
8.19
0.41
11.64
11.43
2.30
—
—
25.60
18.83
16.46
6.38
7.07
10.75
25.42
2.47
Agriculture
on Highly
Erodible
Soils
—
—
—
—
—
—
—
—
—
—
—
6.38
—
—
—
—
Figures 15 through 20 depict six of the sampling station drainage areas. Sites S68, SI 13, and SI 95 are
ranked as good data sites, based on the relative rankings of the data measurements. Site S68 is a small
forested drainage located in HUC 2. Site 113 is also relatively small and is located in HUC 6; although
agriculture and urban areas are evident within the drainage, they are fragmented as compared to the forest
landcover; much of the riparian corridor is wetlands. Site SI 95 is a larger drainage and higher order
stream located in HUC 2. All of the landcover types are present, as are a number of roads and a few
dams. The predominant landcover, however, is unfragmented forest.
The remaining three figures are indicative of sites with fair to bad relative rankings. Site S22, located in
HUC 3, subbasin # 39 has extensive agriculture, much of it in large blocks while the forest landcover
types are fragmented. Site S80 is a large drainage area located in HUC 3 subbasin # 36; the sampling site
is located in an area of unfragmented forest, but the upper reaches of the drainage, including the
28
-------
headwaters of most of the streams are dominated by urban and agricultural landcovers and extensive road
networks. Site SI49 is a fairly small drainage located in HUC 3, subbasin #20. There is extensive
agriculture and urban land use; the forest landcover is highly fragmented. The headwaters of one of the
two streams in the drainage is found in an area of high intensity commercial/industrial land use.
Table 19. Roads Crossing Streams and Impoundments for Sampling Site Drainages
Site
S22
S27
S68
S80
S81
S95
S103
S113
S130
S149
S151
S155
S195
S197
S200
S216
Roads
Crossings
3
21
2
30
30
37
1
1
4
5
8
4
27
1
19
1
No. Crossings/
Stream km
0.41
0.44
0.46
0.67
0.65
0.41
0.30
0.22
0.46
0.73
0.84
0.21
0.59
0.49
0.99
0.18
Dams
0
1
0
6
6
4
0
0
1
0
0
0
4
0
0
0
No. Dams/
Stream km
—
0.021
—
0.134
0.131
0.045
—
—
0.115
—
—
—
0.087
—
—
—
Results for each landscape metrics were encoded into ARC/INFO Grids. A Grid stack was generated and
used to develop a correlation matrix. A separate Grid stack was generated for the riparian corridors
contained in the drainage areas for the sixteen sampling sites. With an n of 16, the correlation
coefficients are significant at values greater than 0.666 for a = 0.005, at values greater than 0.601 for a =
0.01, at values greater than 0.507 for a = 0.025, and at values greater than 0.425 for a = 0.05. Using
these values, a number of significant correlations between water quality/aquatic biology metrics and
landscape metrics were indicated, as shown in Table 20. In general, correlations were the same or less
for the riparian corridor than for landscape metrics over the whole drainage area. The primary exception
is dissolved oxygen, which exhibited significant correlation only with total anthropogenic cover (U-
index, comprised of an aggregation of urban and agriculture land cover types) in the riparian corridor. It
should be noted that this analysis is preliminary and is based only on the nonrandomly selected subset of
sixteen sampling locations. The data set size was insufficient to perform a cluster analysis. The
strongest correlations were between landscape metrics; this is not surprising as several of the landscape
metrics contain similar information. The redundancy is needed at this point in the research until the
strongest and most sensitive relationships with aquatic metrics can be established.
29
-------
Savannah River
Sampling Site S68 Drainage
Barren
Hay
Row Crops
Evergreen Forest
Mixed Forest
Deciduous Forest
/%,/ Roads
A 'RiverReach
Dam Locations
Figure 15. Sampling Site S68 Drainage.
30
-------
Savannah River
Sampling Site SI 13
Water
Low Residential
High Residential
Commercial Urban
Hay
0.6 Mile*
I
Row Crops
Other Grasses
Evergreen Forest
Mixed Forest
Wetlands
Barren
/V' Roads
V\/ River Reach
• Dam Locations
Deciduous Forest
Figure 16. Sampling Site S113 Drainage.
31
-------
Savannah River
Sampling Site S195 Drainage
Ulle*
Water
Low Residential
Commercial Urban
Hay
Other Grasses
Evergreen Forest
Mixed Forest
Deciduous Forest
Row Crops •• Wetlands
Figure 17. Sampling Site S195 Drainage.
Barren
Roads
River Reach
Dams
32
-------
Savannah River
Sampling Site S22 Drainage
0.7
0.7
1.4 Miles
r
Water
Hay
Row Crops
Deciduous Forest
Wetlands
Barren
Evergreen Forest /\/ Roads
/\/ River Reach
Mixed Forest
Figure 18. Sampling Site S22 Drainage.
33
-------
r
Savannah River
Sampling Site S80 Drainage
2 Miles
Water
Low Residential
High Residential
Commercial Urban
Hay
Row Crops
-
Other Grasses
Evergreen Forest
Mixed Forest
I Deciduous Forest
Wetlands
Barren
/\/Roods
/Y/ River Reach
• Dams
Figure 19. Sampling Site S80 Drainage.
34
-------
Savannah River
Sampling Site S149 Drainage
N
A
0.7
Water
Low Residential
Hfgh Residential
Commercial Urban
0,7 Miles
Row Crops
Other Grasses
Evergreen Forest
Wetlands
Barren
/\/Road8
Mixed Forest Rlver Reach
Hay Deciduous Forest
Figure 20. Sampling Site S149 Drainage.
Dams
35
-------
Table 20. Correlation of Aquatic and Landscape Metrics for Sample Site
Drainages and Riparian Corridors
Aquatic Metric
AGPT
EPT
Richness
Fishjbi
pH
Dissolved Oxygen
Habitat
Conductivity
Landscape
Metric
%Forest Landcover
forest edge
U-index
ag_edge
avg ag patch
avg forest patch
avg forest patch
%forest cover
U-index
avg ag patch
forest edge
ag edge
ag on slopes > 3%
% Forest cover
forest edge
U-index
avg ag patch
avg forest patch
ag edge
ag on slopes > 3%
avg forest patch
forest edge
%forest cover
U-index
roads/streams
%forest cover
forest edge
U-index
ag on slopes >3%
U-index
avg forest patch
ag edge
%forest cover
forest edge
U-index
%forest cover
ag on slopes >3%
Direction of
Correlation
negative
negative
positive
positive
positive
negative
positive
positive
negative
negative
positive
negative
negative
positive
positive
negative
negative
positive
negative
negative
positive
positive
positive
negative
positive
positive
positive
negative
negative
negative
positive
negative
positive
positive
negative
negative
positive
Significance
Level (a = )
0.005/0.01 R
0.005/0.025R
0.005*
0. 025/0. 05R
0.025
0.025/0.01 R
0.005/0.025R
0.01
0.01
0.01
0.025
0.025
0.05
0.025
0.025
0.025*
0.05
0.05
0.05
0.05
0.025*
0.05*
0.05
0.05
0.025
0.05
0.05
0.05
0.05
-/0.05R
0.025*
0.025
0.05
0.05
0.05/0.01 R
0.005
0.05
36
-------
4.4 Landscape Change
Two mosaics were developed from the NALC data base for the 1970s (Figure 21) and the 1990s
(Figure 22) Savannah River Basin study area. The mosaics were matched to provide analysis across
similar areas of the two mosaics. Both mosaics were processed into normalized difference vegetation
index (NDVI) images and the values in the 70s mosaic was subtracted from the 90s. Positive numbers
indicate gains in vegetation and negative numbers equate to losses in vegetation; in Figure 23 vegetation
gains are shown in green while vegetation losses are shown in red. A standard deviation was calculated
using n-1, for the entire change NDVI image. The Arc/Info grid coverages depicting the various areas of
interest were then converted to image files (hereafter referred to as masks), and the UTM coordinates for
each were recorded. The resolution for each mask was converted to 60 meters to match the resolution of
the change NDVI image. The change NDVI image was repeatedly sub-sampled to select the matching
areias of each mask. Each sub-sampled change NDVI image and its corresponding mask were then used
as inputs to a custom in-house software program which calculates the amount of cells (pixels) that are
inside the mask and groups them into 4 categories. They are: cells which are greater than or equal to 4
standard deviations of loss in vegetation, those cells which are greater than or equal to 2 standard
deviations of loss in vegetation, and the corresponding numbers of cells for gains in vegetation. In the
following tables the losses and gains have been grouped together and shown as either a negative number
for percent of loss or a positive number for percent of gain.
An additional column is used to represent the cells removed from the study area, which contain negative
NDVI indices in either the 70s or 90s NDVI image. Negative NDVI indices are generated by clouds,
water and other non-vegetation. This also helps to remove erratic NDVI values caused by differences in
solar illumination. However, sometimes these values are meaningful, as in the case where an
impoundment may have been installed after the 70s image and before the 90s. An example of this is
shown in Figure 24.
37
-------
Figure 21. Mosaic of Circa 1970 NALC Images.
38
-------
Figure 22. Mosaic of Circa 1992 NALC Images.
39
-------
• V, ••
'. AT •• i..
'•••;••• .•••"!•. ;/ .
•'•**•-"'' . • • '•. '•.'-•
"•'•A - , •, ^ •''•
• ^,.;~ -
• •
Savannah River 90s - 70s NDVI Change
2.0 Std. Dev.
Figure 23. Gains (green) and Losses (red) in Vegetation, 1970s to 1990s.
40
-------
1970s 1990s
Savannah River Basin
Figure 24. Lake Russell was Created between the 1970s and 1990s
Table 21 depicts change in selected subbasins of HUC 3.
Subhasin #26 reflects greater than 3% negative change,
because an impoundment was installed between the 70s and
the 90s image. Subbasin #26 is the white-shaded area shown
in Figure 25.
Table 21. Landscape Change for Selected Subbasins
Subbasin
20
26
32
36
52
Percent
NDVI Change
-9.803
-9.993
0661
-2.695
-0.366
Percent NDVI Change,
Negative Numbers
Removed
-9.670
-6.838
-6.384
-2.826
-1.873
•
Figure 25. Land Surface Loss to Lake
Russell in Subbasin #26.
41
-------
Subbasin # 32, differences in the water surface (solar glare) produced a positive change in vegetation
which offset the loss in that area. When the water areas (negative NDVI numbers) were removed the
overall sub-watershed had a loss of greater than 6%.
Table 22 shows the NDVI change in the drainage areas of the selected sixteen sampling sites.
Table 22. Landscape Change for Sampling Site Drainages
Site
Percent NDVI
Change
Percent NDVI Change,
Negative Numbers Removed
Class "Good"
S155
S68
S195
S113
-0.873
-0.613
-0.976
-4.193
-0.859
-0.613
-1.245
-4.000
Class "Bad"
S80
S197
S149
S22
-2.975
-8.235
-8.994
-18.404
-2.748
-8.235
-7.325
-17.591
Class "Fair"
S81
S216
S103
S27
-2.930
-1.235
-0.626
-4.717
-2.707
-1.235
-1.627
-4.674
Class "Other"
S151
S200
S130
S95
-6.992
-4.528
-13.372
-2.932
-5.721
-4.328
-11.149
-2.871
42
-------
Section 5
Summary
The three questions posed as objectives by the Region can now be addressed:
Are both the proportions of land uses and the spatial pattern of land uses important for
characterizing and modeling stream condition in watersheds/ecoregions of different
areas?
As shown in this landscape assessment, both the proportion and the patterns of land use are important in
assessing impacts on streams. In the correlation analysis conducted on the sampling site drainages, both
total landcover types (%forest, U-index) and pattern metrics (fragmentation metrics including average
patch size, forest and agriculture edges) were found to correlate with aquatic metrics. A third important
element is the scale at which analysis is done. In the analysis of selected subbasins, patterns of land use
began to emerge; this scale may be sufficient to provide a generalized characterization of the basin.
Can land uses near the streams better account for the variability in ecological condition
than land use for the entire watershed/ecoregion?
In this particular assessment, landscape metrics for the riparian corridors did not provide stronger
correlation with aquatic metrics, with the exception of dissolved oxygen. It should be remembered,
though, that this is one analysis of a small spatial area in one region with a particular suite of metrics. In
other situations the riparian corridor may be of greater importance than the overall watershed. Even in
this region, the southern portion of the basin has riparian corridors dominated by wetlands. Only one site
from this area was used in the analysis and the entire sampling data set contains only a few sites in this
ecoregion. A separate analysis of wetlands-dominated systems is probably worthwhile.
Does the size of the watershed/ecoregion influence statistical relationships between
landscape characteristics and ecological condition?
There was no indication in this analysis of any relationship with the spatial extent of the drainage areas.
This includes the landscape metrics developed for the HUCs and subbasins. In the sampling site
analysis, one of the selection criteria was to include streams of varying order; by doing so, both small and
large drainage areas were included. Drainage area was included in the correlation analysis; no
correlation was shown with any of the aquatic metrics.
43
-------
References
Bara, T.J., ed. 1994. Multi-Resolution Land Characteristics Consortium Documentation Notebook.
U.S. Environmental Protection Agency, Environmental Monitoring and Assessment Program,
Research TYiangle Park, NC.
Jones, K.B., K.H. Riitters, J.D. Wickham, R.D. Tankerslcy, Jr., R.V. O'Neill, D.J. Chaloud, E.R. Smith,
and A.C. Neale. 1997. An Ecological Assessment of the United States Mid-Atlantic Region: A
Landscape Atlas. U.S. Environmental Protection Agency, Office of Research and Development,
Washington, D.C.
Natural Resource Conservation Service. 1996. State Soil Survey Geographic Data Base (STATSGO)
Metadata. U.S. Department of Agriculture, Natural Resource Conservation Service, Washington,
D.C.
Santa Cruz County. 1998. Riparian Corridors http://gate.cruzio.com/~countysc/pln/riparian.htm
South Carolina Department of Natural Resources. 1998. South Carolina Rivers Program: Recommended
Best Management Practices for River-bordering Lands
http://wvvvv.dnr.state.sc.us/vvater/envaff/river/bnips.html
U.S. Army Corps of Engineers. 1997. The National Inventory of Dams http://corpsgeo I .usace.army.mil/
U.S. Environmental Protection Agency. 1998. Ecological Restoration
http://www.epa.gov/owovvwtrl/NPS/Ecology/chapl.html
U.S. Environmental Protection Agency. 1997. History of the U.S. EPA 's River Reach Eilc. U.S.
Environmental Protection Agency, Office of Water, Washington, D.C.
U.S. Environmental Protection Agency. 1993. North American Landscape Characterization (NALC)
Research Plan. U.S. Environmental Protection Agency, Office of Research and Development,
Washington, D.C.
U.S. Geological Survey. 1990. Digital Elevation Models, National Mapping Program Technical
Instructions, Data Users Guide 5. U.S. Geological Survey, Second Printing (Revised), Reston,
Virginia, 1990.
U.S. Geological Survey. 1989. Digital Line Graphs from 1: lOO.OOD-Scalc Maps Data Users Guide 2.
U.S. Geological Survey, Reston, Virginia.
44
-------
U.S. Geological Survey. 1982. Codes for identification ofhydrologic units in the United States and the
Caribbean outlying areas. U.S. Geological Survey, Circular 878-A, Reston, VA.
45
------- |