METHODS FOR MULTI-SPATIAL SCALE CHARACTERIZATION OF RIPARIAN CORRIDORS
Thomas J. Moser,1 Dale R. Lindeman,1 P.J. Wigington, Jr.,2 Michael J. Schuft,3 and John Van Sickle2
ABSTRACT: This paper describes the application of aerial photography and GIS technology to develop flexible and
transferable methods for multi-spatial scale characterization and analysis of riparian corridors. Relationships between
structural attributes of riparian corridors and indicators of stream ecological condition are not well established. As part of a
research project focused on assessing riparian-stream systems in agricultural landscapes of Oregon's Willamette Valley, GIS
land cover/land use databases were created from 1997 aerial photography and digital orthophotography for 23 predominantly
agricultural watersheds, including detailed land cover/land use coverages within 150 m of perennial and intermittent streams.
GIS functions were used to partition the stream networks at various lateral-longitudinal scales to quantify land cover/land use
and to generate functionally relevant metrics of riparian vegetation, such as its composition, width, and continuity. The
methods developed provide considerable flexibility for generating metrics characterizing attributes of riparian corridors and
for exploring relationships among these metrics and indicators of stream ecological condition.
KEY TERMS: Willamette Valley; agricultural landscapes; riparian corridor; watershed; aerial photography; GIS
INTRODUCTION
Riparian plant communities are widely considered critical for maintaining stream ecological condition (Gregory et
al. 1991). Aquatic functions of riparian vegetation include: 1) providing stream shading, 2) contributing large woody debris
and fine organic matter, 3) regulating the flux of upland-derived sediments, nutrients, and other chemicals, and 4) stabilizing
stream banks (Brinson et al. 1981, Malanson 1993). The principal structural attributes of riparian corridors affecting the
above functions include the composition, width, and continuity of natural vegetation adjacent to and along the stream
network (Malanson, 1993, Forman 1997). The relationships between riparian structural attributes and stream ecological
condition are not well established. Because these relationships are likely to be scale-dependent, effective methods are
necessary for addressing the effects of spatial scale on the strength of associations between riparian and watershed structural
attributes and biotic, chemical, and physical indicators of stream condition.
As part of the U.S. Environmental Protection Agency's Pacific Northwest Research Program (Baker et al. 1995),
research is being conducted to determine the effect of riparian areas on the ecological condition of small, perennial streams in
agricultural landscapes of the Willamette Valley (Moser et al. 1997). The overall objective of this research is to quantify
relationships between remotely-sensed riparian and watershed structural attributes at varying spatial scales and field-based
indicators of stream ecological condition. Stream ecological condition indicators were derived from field measurements of
biotic assemblages, water chemistry, and physical habitat in a single stream reach at the base of each study watershed. A
stream reach was defined as 40 times its mean wetted width during summer, low flow conditions; and, for our study, stream
reaches ranged from 150 to 320 m in length. Currently, statistical relationships between reach-level indicators of stream
ecological condition and riparian and watershed attributes upstream of the sampling reach are being investigated (see
Wigington et al. this volume). This paper extends the methods reported by Schuft et al. (1999) from a sampling strategy to a
full multi-scale characterization of riparian areas along the entire stream network; and, provides specific applications of these
methods beyond the objective of the current research.
METHODS
The study was conducted on 23 watersheds (about 15 to 87 km2) distributed throughout Oregon's Willamette Valley
(Figure 1A). The Willamette Valley is predominantly an agricultural landscape of approximately 13,165 km2 lying between
the Coast Range on the west and the Cascade Range on the east; and, represents one of the largest concentrations of
1 Dynamac Corporation, 200 SW 35th Street, Corvallis, Oregon 97333, USA
2 U.S. Environmental Protection Agency, 200 SW 35th Street, Corvallis, Oregon 97333, USA
3 Oregon Department of Forestry, 2600 State Street, Salem, Oregon 97310, USA
-------
I I Agriculture
CH Grass/Forb
IB Shrub/Scrub
Forest
Built Up
Water
Total Band Number Total
Band Forest Area in of Gap
Band Area (ha) Area (ha) Forest (%) Gaps Length
_..J3fi
UL
Forest Gap
• Forest Buffer
' Stream
Figure 1, Location of the 23 study watersheds within the Willamette Valley (A); distribution of six LCLU classes in the Bear Branch watershed and within the 150 m band on
both sides of the stream network (B); segment of Bear Branch demonstrating the use of 10 m incremental banding for calculating percent LCLU (e.g., forest) and the number
and length (m) of gaps between forest patches as a function of distance from the stream (C); and, the same segment of Bear Branch demonstrating the use of arcs perpendicular
to the stream for estimating the width (m) of riparian vegetation, such as forest (D).
-------
diversified agriculture in the Pacific Northwest, with substantial areas in grass seed cropping systems. Because the focus of
the study is to determine the influence of riparian areas on the ecological condition of small, perennial streams in agricultural
landscapes, the selection of the study streams and associated riparian systems was restricted to streams draining watersheds
lying within the Valley's Prairie Terraces and Foothill ecoregions (Pater et al. 1998) Additional selection criteria included
watersheds having a large proportion of their area in agriculture and having no urban development.
Color-infrared aerial photography (1:31,680) of the study watersheds was flown in mid July 1997. Digital
orthophotos with a minimum resolvable unit of about one meter were created from the aerial photos. The watershed for each
of the sampled stream reaches was determined from 1:24,000 USGS topographic maps and subsequently delineated on the
digital orthophotos. Perennial and intermittent streams were delineated directly on the digital orthophotos to create a
geographic information system (GIS) coverage for each stream network. A detailed classification system, modified after
Anderson et al. (1976), was used to interpret and characterize land cover/land use (LCLU) of the riparian corridor, defined in
this study as a 150 m banded area adjacent to and along each side of the stream network (Figure IB). The riparian corridor
classification system incorporated 49 LCLU classes, including: 1) forest (pastured and non-pastured coniferous, deciduous,
and mixed forest, each with three canopy closure types, for a total of 18 subclasses), 2) clear-cut, 3) tree plantation, 4)
pastured and non-pastured shrub/scrub, 5) pastured and non-pastured grass/forb, 6) agriculture (13 subclasses); 7) built-up
(five subclasses), 8) barren, 9) water (three subclasses), and 10) other. The riparian corridor polygon coverage was digitized
with a minimum mapping unit of 0.1 ha. It was produced with the intention of being able to capture narrow patches of
riparian vegetation along the stream network. Watershed LCLU was characterized using a more generalized classification
system which incorporated 18 LCLU classes, including: 1) forest, 2) clear-cut, 3) shrub/scrub, 4) grass/forb, 5) agriculture (9
subclasses), 6) built-up (two subclasses), 7) barren, 8) water, and 9) other. The watershed coverage (Figure IB) was created
using a grid composed of 0.25 ha square polygons in which the dominant LULC class was assigned to each polygon.
Prior to the interpretation of riparian LCLU, extensive field reconnaissance was conducted during the summer of
1997 to record agriculture land use adjacent to and along both sides of the stream networks in the 23 watersheds. During the
photo-interpretation process, additional field reconnaissance was conducted to classify problematic polygons. Upon
completion of the GIS coverages, ground-truthing was conducted in each watershed to estimate the classification accuracy of
the photo-interpreted LCLU within the riparian corridor. Based on 1,889 ground observations, the overall classification
accuracy was 84.1%. At the highest level in the classification hierarchy, user's accuracy was over 90% for forest, agriculture,
built-up, barren, and water; and over 80% for shrub/scrub and grass/forb.
We are developing approaches and methods for multi-spatial scale characterization of riparian structural attributes
across and along the corridor's lateral and longitudinal dimensions (i.e., potential areas of influence). Data from this multi-
spatial scale approach are being used to investigate empirical relationships between metrics characterizing riparian structural
attributes and indicators of stream ecological condition (see Wigington et al. this volume). GIS functions and programs are
being used to band and partition the riparian corridor into various lateral-longitudinal combinations. The lateral dimension
captures the cross-sectional structure of the riparian corridor, while the longitudinal dimension captures the linear structure
along the riparian corridor. As shown in Figure 1C, the composition of riparian LCLU is being characterized using GIS
clipping functions to incrementally band the riparian corridor. In this example, the proportional area of forest as a function of
distance from the stream was calculated in 10-m incremental bands out to 50 m for both sides of a short segment of Bear
Branch. Figure 1C also illustrates the use of banding to address the continuity of natural riparian vegetation (e.g., forest,
shrub/scrub, and grass/forb), where the number and length of gaps between forest patches were calculated as a function of
distance from the stream. As shown in Figure ID, the width of the riparian buffer is being estimated from arcs aligned
perpendicular to both sides of the stream at 50-m intervals. In this example, the width of the forest cover immediately
adjacent to the stream was measured to the outer edge of contiguous forest cover.
Investigations of the effects of riparian structural attributes on stream ecological condition can be enhanced with the
development of methods for partitioning the riparian corridor along its longitudinal dimension. We used GIS clipping
functions to partition the 150-m laterally banded riparian corridor into several longitudinal sections (i.e., the reach, and
absolute distances along the perennial and intermittent streams of 500, 1,000, 2,500, 5,000, 7,500, 10,000 m above the reach,
as well as the entire network above the reach). This approach allows for the characterization of riparian structural attributes at
multiple lateral and longitudinal scales along the riparian corridor, such as LCLU composition within a 10-m lateral distance
from the stream over a 10,000-m longitudinal distance above the stream reach. Figure 2 illustrates this approach for riparian
LCLU composition, where the proportional area of three selected LCLU classes are plotted as a function of distance from the
stream at four different longitudinal scales of Bear Branch, including the 150-m stream reach, and absolute distances along
the primary channel and tributaries of 1,000 m, 10,000 m, and entire network above the reach. Width and gap measurements
of defined riparian buffers can also be calculated for multiple lateral and longitudinal scales.
-------
CD Agriculture
E3 Grass/Forb
1^ Forest
^1 Water
Reach
loo-
's 60-
TO
£ 40-
o
* 20-
Network above the Reach
30 60 90
Distance from Stream (m)
120
100'
«
I 80'
1 60
I40'
£ 20-
10,000 m above the Reach
0
30
60
90
120
1001
«
I 80
1 60-
I
£ 20-
Distance from Stream (m)
1,000 m above the Reach
30 60 90
Distance from Stream (m)
120
100-
(fl
I 80'
1 60-
C3
5 40'
£ 20'
Reach
30 60 90 120
Distance from Stream (m)
Forest -^—Grass/Fort) —o—Agriculture
Figure 2, Comparison of Bear Branch LCLU as a function of distance from the stream over four longitudinal scales of the stream network.
RESULTS AND DISCUSSION
An important consideration in designing a study to predict stream ecological condition through the use of spatially
explicit LCLU data is the ability to explore the effect of spatial scale on the strength of associations between a suite of
explanatory landscape variables and various response variables derived from stream measurements of biotic assemblages,
water chemistry, and physical habitat. Researchers conducting watershed-scale studies have used LCLU maps or GIS data to
demonstrate scale-dependent relationships between land cover composition and/or pattern to indicators of stream condition
(Barton, 1988, Lammert and Allan 1999, Roth et al. 1996, Steedman 1988). The methods for characterizing riparian corridors
that are discussed in this paper build upon the work of others, particularly Roth et al, (1996) and Schuft et al. (1999).
Compositional metrics, such as the proportion of an individual LCLU class or a combination of LCLU classes
relative to a unit area, are easy to generate and are often used to demonstrate statistical relationships with stream condition
indicators (Lammert and Allan 1999, Osborne and Wiley 1988, Roth et al. 1996, Steedman 1988). These metrics are also
useful in the initial, more descriptive phases of LCLU data analyses, where comparisons among study areas and between
different landscape units within the same study area can reveal overall trends and lead to questions about the distribution of
LCLU within the watershed and along the stream network. For example, Table 1 lists the percent composition of selected
LCLU classes for five of our 23 watersheds at two different spatial scales, watershed and riparian corridor. A considerable
range in percent areal cover for forest, grass/forb, and agriculture at both the watershed and riparian corridor spatial scales
can be seen. Because of their potential contributions to riparian buffer functions, it is interesting to note the greater
proportion of area in forest, shrub/scrub, and/or grass/forb at the riparian corridor scale compared to the watershed scale.
With the exception of Butte Creek, the proportional area in agriculture was less at the riparian corridor scale than at the
watershed scale.
-------
Table 1. Areal cover of selected LCLU classes for five of the 23 Willamette Valley watersheds and their associated riparian corridors at the
30-m lateral and network longitudinal scale (including the stream reach).
Percent of Watershed Area
Watershed Name
Howell Prairie Creek
Spoon Creek
Case Creek
Bear Branch
Butte Creek
Forest
3.7
4.6
9.8
16.4
46.8
Shrub/
Scrub
0.6
0.4
1.8
2.8
2.6
Grass/
Forb
3.6
12.3
12.5
28.5
12.5
Ag
89.0
81.6
71.1
50.5
28.9
Built
2.9
0.7
3.8
1.6
0.7
Percent of Riparian Corridor Area
Forest
23.3
7.4
45.5
42.4
45.6
Shrub/
Scrub
12.9
2.6
20.2
10.4
6.2
Grass/
Forb
20.6
17.5
16.3
26.7
15.3
Ag
38.0
70.2
4.7
17.0
29.5
Built
1.1
1.9
1.0
1.3
1.4
Because of the spatial heterogeneity within landscapes (see Figure 1), an analysis of LCLU data for an entire
watershed or for an entire riparian corridor (as in Table 1) may fail to discover finer spatial scale relationships between
riparian structural attributes and stream ecological condition. The investigation of the composition of LCLU classes at
varying lateral and longitudinal scales can provide important insights on the distribution and pattern of LCLU along a stream
network. Figure 2 shows there are clear distinctions in the distribution of the three LCLU classes plotted for Bear Branch as
the scale increases in both a lateral and longitudinal direction. Although forest declined and agriculture increased as a
function of distance from the stream at the four longitudinal scales shown, the rate of change in both LCLU classes was much
more pronounced at the reach scale and 1,000 m above the reach scale than at the two larger scales. Grass/forb increased as a
function of distance from the stream at longitudinal scales above the stream reach. Thus, characterization of riparian LCLU
only adjacent to an in-stream sampling reach may give a misleading picture of the entire riparian corridor. The incremental
banding of riparian corridors is an effective approach allowing for the visualization of spatial variability among riparian
attributes; and, for conducting explicit examinations of the affects of riparian LCLU over a range of lateral-longitudinal
combinations on stream ecological condition.
The width and continuity of riparian vegetation along stream networks are important structural attributes affecting
stream ecological condition (Barton et al. 1985, Castille et al. 1994, Forman 1997, Weller et al. 1998). However, metrics
addressing these attributes are more difficult to quantify than compositional metrics and require additional definition and
linkages to riparian functions of interest. Width and gap calculations were made using the methods illustrated in Figures 1C
and ID and frequency distributions of riparian vegetation for the Bear Branch network were plotted for two buffer types.
Figure 3 shows distinct differences in width and gap lengths between the two buffer types. The forest buffer was narrower
with a greater number of long gaps than the forest-shrub/scrub-grass/forb buffer, illustrating that width and continuity of a
riparian buffer are dependent upon its compositional components. About 17% of the network streambank had a forest buffer
width of more than 30 m, compared to about 44% for the forest-shrub/scrub-grass/forb buffer (Figure 3A). From inspection
of the y-intercept in Figure 3A, a forest buffer was absent on about 37% of the streambank, compared to about 8% for the
forest-shrub/scrub-grass/forb buffer. Continuity metrics provide information addressing the number and length of gaps
between riparian vegetation patches along the stream network. For example, the forest and forest-shrub/scrub-grass/forb
buffers had a total of 296 gaps (ranging from 1 to 887 m in length) and 63 gaps (ranging from 1 to 600 m in length),
respectively. About 50% of the gaps for the forest buffer were greater than 30 m, while 15% of the gaps for the forest-
shrub/scrub-grass/forb buffer were greater than 30 m (Figure 3B). Additional gap metrics can be generated from this basic
information, such as mean gap length and variation and the number of gaps per unit length of streambank or lateral band.
Forest
Forest-Shrub/Scrub-Grass/Forb
60
Buffer Width (m)
Forest
Forest-Shrub/Scrub-Grass/Forb
0 100 200 300 400 500 600 700 800 900
Buffer Gap Length (m)
Figure 3. Relative cumulative frequency distributions of width (A) and gap length (B) for two riparian buffer types for the Bear Branch
network. Width was measured at 50-m intervals as shown in Figure ID. The number and length of gaps between buffer vegetation on the
network streambank was determined as shown in Figure 1C.
-------
The methods described provide considerable flexibility and are broadly applicable for multi-scale characterization
and analysis of riparian and watershed GIS LCLU coverages. Partitioning the stream network into longitudinal segments
provides a method for investigating the influence of riparian structural attributes, such as LCLU composition and width and
gap metrics, on temperature or water quality data obtained from a series of in-stream reaches along a longitudinal gradient.
Because the contribution of forest vegetation in moderating stream temperature and providing large woody debris diminishes
with distance from the stream, the incremental banding approach provides the ability to generate data that "drive" riparian
process models of shading or woody debris input at the entire stream network scale. Mapping and graphically displaying the
riparian LCLU data at various lateral-longitudinal combinations provides the ability to identify sites for field studies, such as
process-level investigations of nutrient and sediment regulation. Finally, GIS LCLU coverages created from aerial
photography or high resolution, multi-spectral satellite imagery provide continuous data along a stream network and offer
alternatives to field-based inventory in conducting riparian assessments.
ACKNOWLEDGMENTS
We thank Lynne McAllister and Stephanie Gwin for their help in conducting initial field reconnaissance of
agricultural land use, and Denis White and Peter Lattin for critical reviews of the document. The research described in this
document has been funded by the U.S. Environmental Protection Agency, Western Ecology Division in Corvallis, Oregon
through Contract 68-C6-0005 to Dynamac Corporation. It has been subjected to the Agency's peer and administrative review
and approved for publication. Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
REFERENCES
Anderson, J.R., E.E. Hardy, J.T. Roach, and R.E. Witmer. 1976. A Land Use and Land Cover Classification System for Use
with Remote Sensor Data. Geological Survey Professional Paper 964. Washington D.C., U.S. Government Printing
Office for the U.S. Department of the Interior, Geological Survey.
Baker, J.P., D.H, Landers, H. Lee, P.L. Ringold, R.R, Sumner, P.J, Wigington, Jr., R.S. Bennett, E.M. Preston, W.E. Frick,
A.C. Sigleo, D.T. Specht, and D.R. Young. 1995. Ecosystem Management Research in the Pacific Northwest: Five-
Year Research Strategy, EPA/600/R-95/069, U.S. EPA, Corvallis, Oregon.
Barton, D.R., W.D. Taylor, and R.M. Biette. 1985. Dimensions of Riparian Buffer Strips Required to Maintain Trout Habitat
in Southern Ontario Streams. N. Am. J. Fish. Manage. 5:364-378.
Brinson, M.M., B.L. Swift, R.C. Plantico, and J.S. Barclay. 1981. Riparian Ecosystems: Their Ecology and Status.
FWS/OBS-81/17. U.S. Department of the Interior, Fish and Wildlife Service, Kearneysville, WV.
Castelle, A.J., A.W. Johnson, and C. Conolly. 1994. Wetland and Stream Buffer Size Requirements: A Review. J. Environ.
Qua]. 23:878-882.
Forman, R.T. 1997. Landscape Mosaics: The Ecology of Landscapes and Regions. Cambridge University Press, Cambridge,
United Kingdom.
Gregory, S.V., F.J. Swanson, W.A. McKee, and K.W, Cummins. 1991. An Ecosystem Perspective of Riparian Zones.
BioScience 41:540-551.
Lammert, M. and J.D. Allan. 1999. Assessing Biotic Integrity of Streams: Effects of Scale in Measuring the Influence of
Land Use/Cover and Habitat Structure on Fish and Macroinvertebrates. Environmental Management 23:257-270.
Malanson, G.P. 1993. Riparian Landscapes. Cambridge University Press, United Kingdom.
Moser, T.J., PJ. Wigington, Jr., M.J. Schuft, P.R. Kaufmann, A.T. Herlihy, J. Van Sickle, and L.S. McAllister. 1997. The
Effect of Riparian Areas on the Ecological Condition of Small, Perennial Streams in Agriculture Landscapes of the
Willamette Valley: Research Plan. EPA/600/R-97/074. U.S. Environmental Protection Agency, National Health and
Environmental Effects Laboratory, Western Ecology Division, Corvallis, Oregon.
Osborne, L.L. and M.J. Wiley. 1988. Empirical Relationships Between Land Use/Cover and Stream Water Quality in an
Agricultural Watershed. J. Envir. Mgmt. 26:9-27.
Pater, D.E., S.A. Bryce, T.D. Thorson, J. Kagan, C. Chappe, J.M. Omernik, S.H. Azevedo, A.J. Woods. 1998. Ecoregions of
Western Washington and Oregon. U.S. Geological Survey, Reston, Virginia.
Roth, N.E., JD. Allen, and D.L. Erickson. 1996. Landscape Influences on Stream Biotic Integrity Assessed at Multiple
Spatial Scales. Landscape Ecology 11:141-156.
Schuft, M.J., TJ. Moser, P.J. Wigington Jr., D.L. Stevens, Jr., L.S. McAllister, S.S. Chapman, and T.L. Ernst. 1999.
Development of Landscape Metrics for Characterizing Riparian-Stream Networks. Photogrammetric Engineering &
Remote Sensing, 65(10): 1157-1167.
Steedman, R.J. 1988. Modification and Assessment of an Index of Biotic Integrity to Quantify Stream Quality in Southern
Ontario. Can. J. Fish. Aquat. Sci. 45:492-501.
Weller, D.E., T.E. Jordan, and D.L. Correll. 1998. Heuristic Models for Material Discharge from Landscapes with Riparian
Buffers. Ecological Applications, 8:1156-1169.
-------
TECHNICAL REPORT DATA
(Please read instructions on the reverse before completing)
1 . REPORT NO.
EPA/600/A-00/021
2.
4. TITLE AND SUBTITLE Methods for multi-spatial scale characterization of riparian
corridors.
7. AUTHOR(S) Thomas J, Moser1, Dale R. Lindeman1, P.J. Wigington2, Michael J,
Schuft3,John Van Sickle2
9. PERFORMING ORGANIZATION NAME AND ADDRESS
'Dynmac Corporation z US EPA NHEERL WED 3 Oregon Department of Forestry
US EPA NHEERL WED 200 SW 35th Street 2600 State Street
200 SW 35lh Street Corvallis, OR 97333
Corvallis, OR 97333
Salem, OR 97310
12. SPONSORING AGENCY NAME AND ADDRESS
US EPA ENVIRONMENTAL RESEARCH LABORATORY
200 SW 35th Street
Corvallis, OR 97333
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
6. PERFORMING ORGANIZATION
CODE
8. PERFORMING ORGANIZATION REPORT
NO.
10. PROGRAM ELEMENT NO.
1 1 . CONTRACT/GRANT NO.
13. TYPE OF REPORT AND PERIOD
COVERED
14. SPONSORING AGENCY CODE
EPA/600/02
15. SUPPLEMENTARY NOTES:
1 6. ABSTRACT: This paper describes the application of aerial photography and GIS technology to develop flexible and transferable methods for
multi-spatial scale characterization and analySsis of riparian corridors. Relationships between structural attributes of riparian corridors and
indicators of stream ecological condition are not well established. As part of a research project focused on assessing riparian-stream systems in
agricultural landscapes of Oregon's Willamette Valley, GIS land cover/land use databases were created from 1 997 aerial photography and
digital orthophotography for 23 predominately agricultural watersheds, including detailed land cover/land use coverages within 150 m of
perennial and intermittent streams. GIS functions were used to partition the stream networks at various lateral-longitudinal scales to quantify
land cover/land use and to generate functionally relevant metrics of riparian vegetation, such as its composition, width, and continuity. The
methods developed provide considerable flexibility for generating metrics characterizing attributes of riparian corridors and for exploring
relationships among these metrics and indicators of stream ecological condition.
17.
a. DESCRIPTORS
Willamette Valley, agricultural landscapes,
riparian corridor, watershed, aerial
photography, GIS
18. DISTRIBUTION STATEMENT
KEY WORDS AND DOCUMENT ANALYSIS
b. IDENTIFIERS/OPEN ENDED
TERMS
1 9. SECURITY CLASS (This Report)
20. SECURITY CLASS (This page)
c. COSATI Field/Group
21. NO. OF PAGES; 6
22. PRICE
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDITION IS OBSOLETE
------- |