Edith Read, Jennifer Gough
BEYOND MAPPING: USING GIS FOR
NATURAL RESOURCE ASSESSMENT
AND ANALYSIS
ABSTRACT
Many agencies and private companies now use GIS as a tool for producing
maps of wildlife habitats and other natural resources. The next step in
application of GIS is data analysis to address issues of specific concern
to resource managers. We present three diverse examples which show how GIS
can be applied to relationships between hydrological and biological resources.
In the first example, we show cation-anion characteristics of spring water
in the San Bernardino Mountains of California. Plots of the data in the
form of Stiff diagrams clearly illustrate differences between calcareous,
hydrothermal waters along a fault line and artesian waters that typify most
of the sample sites. Further study is needed to determine if such differences
help explain the patchy distribution of certain wetland plant species within
the San Bernardino Mountains. In the second example, we show application
of GIS to vegetation transect and hydrological data obtained as part of
a long-term riparian monitoring program in the Sierra Nevada of California.
Box-and-whisker plots of the data using S-Plus (StatSci Division, Mathsoft,
Inc., Seattle, Washington, USA prove especially useful in showing the distribution
of wetland indicators in relation to distance from stream. The third example
shows habitat analysis for the Santa Ana River woolly star, an endangered
plant species whose life cycle is linked to flooding frequency and diversity
in successional stage of alluvial terraces within the floodplain.
EXAMPLE 1. VARIABILITY IN CATION-ANION BALANCE OF SPRING WATERS IN THE
SAN BERNARDINO MOUNTAINS OF CALIFORNIA
Background
Biologists often view springs from the perspective of variability in plant
composition and aquatic habitat, but in many cases, water quality differences
between springs which may account for this variability are unknown. This
example highlights the first phase of an investigation of water quality
- habitat relationships. This phase consists of a large - scale water sampling
program to identify areas that might be suitable for more intensive habitat
analysis.
Methods
Water samples were collected in the summer and fall of 1995 from 15 locations
in the southwestern foothills of the San Bernardino Mountains, north of
Highland. The resource locations were determined using the Real Time Kinematic
Global Positioning System (RTKGPS) to survey the sites as a part of the
fieldwork in mountainous terrain. The water samples were analyzed by a laboratory.
The resource positions were used to build a resource location ArcInfo point
coverage. One reference data table was created by merging point identification
number and location from the coverage data and laboratory results from Excel
spreadsheets.
To compare sample sites, cation and anion data were incorporated into a
series of diamond-shaped polygons known as Stiff diagrams (also known as
shape diagrams), named after H.A.Stiff who developed this method of pattern
analysis for tracing waters of similar formation over large areas (Stiff,
1951). A compiled 'C' program created vectors from the laboratory cation
and anion values in the reference table and output a generate file from
which the Stiff polygons could be created. In order to scale the sides of
the Stiff polygons for a good fit on the final plot, a scale factor was
input into the program.
Results and Discussion
The Stiff diagrams show clear differences in the cation-anion balance of
different springs in relation to the geography of the area:
Please refer to the graphic entitled "Variation in Cation-Anion
Balance in Spring Waters of the San Bernardino Mountains" presented
at the conference. Due to file size limitations, we regret that we are unable
to include the graphic here.
The horizontal axis indicates concentration (in equivalent parts per million)
of cations (left side) and anions (right side). Each major cation or anion
is shown on a horizontal line within the diagram. The polygon is formed
when data points for each element are connected. The distinctive shapes
of the polygons provide some indication of common or differing recharge
sources. Minor geochemical differences can be attributed to specific lithologic
units through which the source of groundwater recharge flows before reaching
the sample site.
In the above plot, the San Andreas fault follows an northwest- southeast
path along the base of the mountains. Associated trace faults are located
in the larger canyon to the north. These faults are not apparent on surface
topography at the scale of this plot, but the water chemistry data show
their signature: a predominance of calcium and bicarbonate elements in waters
of hydrothermal origin near the fault traces.
This type of baseline data and method of comparison between samples should
prove useful to resource managers who may be concerned with seasonal and
long-term changes in water quality. Among other applications, this type
of analysis should be useful in selecting a subset of sites for intensive
study of vegetation and aquatic habitat in relation to water quality.
EXAMPLE 2. DISTRIBUTION OF WETLAND INDICATOR SPECIES IN RELATION TO
GROUNDWATER DEPTH AND DISTANCE FROM STREAM
Background
Montane riparian vegetation in the eastern Sierra Nevada of California is
impacted by disturbances that range from natural flooding to human disturbances
from campers, grazing, and water diversion for hydroelectric power generation.
A monitoring program was established to evaluate a number of vegetation
parameters over time in relation to changes in stream flow expected when
certain hydroelectric facilities begin releasing more water into the streams
in accordance with regulatory requirements. This example highlights some
of the results of the monitoring program for two out of 13 sites monitored,
one on Mill Creek in Lundy Canyon (Mono County) and the other on Bishop
Creek (Inyo County). We chose these two sites as examples because they exhibit
contrasting features which should be of interest to those who are faced
with such diversity in other riparian systems.
Methods
Streamflow and groundwater data were recorded continuously by dataloggers
at the monitoring sites in 1992 and 1993 by Southern California Edison.
The energy source for the dataloggers came from RV-type batteries connected
to a solar panel. Streamflow was measured with a calibrated transducer located
in the center of the stream channel at each site. Groundwater data were
collected from a single monitoring well at each site. Groundwater profiles
were generated from the well data and stream stage (height) data. While
it would be desirable to have more than one monitoring well at each site,
considerations of economics and additional impacts to habitat from well
construction restricted the number of wells. However, due to the small size
of the sites (0.1-0.5 hectare), variation in subsurface geomorphology that
would cause significant variation in groundwater conditions is expected
to be small -- thus we believe the groundwater profiles generated from the
hydrological data provide reasonable views of subsurface conditions.
We collected vegetation data from permanent belt transects during the same
period in which the hydrological data were collected. Palmtop computers
were used to collect the vegetation data, which were downloaded daily to
a PC and checked for errors. Data files were then converted to comma-delimited
format for use in GIS.
At this point, the task for GIS at the request of the client was to produce
plots that would show the temporal and spatial variability of the vegetation
data. These plots would replace numerous bar charts and conventional graphics
that had been used previously in an attempt to show as much of the data
as possible.
Plan and profile views were created for each site using a combination of
'C' programs and ArcInfo generated tin routines. Three parallel rows of
meter squares were defined along each transect (see plot image below) and
each square was assigned a unique identification number indicating transect
number, distance from the transect endpoint and year of data collection.
Tabular data were imported into pre-defined Info tables. Using the meter
value, the year and the transect number from the tabular data for all records
were given numbers which corresponded with the meter square numbering system.
Tabular data were then processed into sub-tables and related to the appropriate
meter square. Class commands were used to analyze the final data and to
shade the planar views.
A sample plot(plan view) is shown below:
Please refer to the graphic entitled "Absolute Cover of Trees and
Shrubs on the Mill Creek Monitoring Site, 1992 - 1994" presented at
the conference. Due to file size limitations, we regret that we are unable
to include the graphic here.
Data for each belt transect for each of the three monitoring years are shown
as strips, plotted side-by-side, to illustrate variability between years.
The resolution of the data is one meter, represented by colored rectangles
of various patterns and colors to illustrate variation in canopy cover.
The large arrow shows the direction of stream flow. We also record the distribution
of alluvial terraces (fluvial surfaces), which are distinguished by different
codes in the database (e.g., F1L means the first fluvial surface adjacent
to the stream channel on the left bank; S1R indicates a slope between two
fluvial surfaces on the right bank).
We related the vegetation and hydrology data by comparing the vegetation
transect data for each meter interval with depth-to-groundwater values generated
from the hydrology data for the same meter intervals. Statistical analyses
and plots were conducted in S-Plus, exported to QuattroPro (v.6.0) for final
text editing, then saved in GIF format for presentation here.
Results and Discussion
We have a substantial amount of data, only a portion of which can be presented
here. The Mill Creek and Bishop Creek sites are very different hydrologically
and therefore can be expected to have significantly different capacities
to support wetland vegetation. These hydrological differences need to be
recognized for accurate assessment of human impacts and potential for habitat
restoration.
The Mill Creek site is in a glacial valley, 2268 meters elevation in a "gaining"
reach of Mill Creek (groundwater feeds surface flow). Creek flow is perennial,
but depth to groundwater increases significantly with distance from the
channel because of steep topography. In contrast, the Bishop Creek site
is in an alluvial fan, 1398 meters elevation in a "losing" reach
of Bishop Creek (surface flow feeds groundwater). The creek follows a narrow
channel within a relatively broad series of alluvial terraces.
The following plot shows contrasting relationships between groundwater depth
and distance from stream for the two sites, taken from data collected in
1992 (a low runoff year) and 1993 (a high runoff year):
;
Differences in slope gradient between the two sites account for the differing
groundwater profiles. From an ecological perspective, the most interesting
aspect of the data shown by these plots is response of the groundwater to
changes in surface runoff. Depth to groundwater at the glacial valley site
was not significantly affected by a small increase in average annual stream
discharge of 0.3 cubic feet per second (cfs). In contrast, average groundwater
depth at the alluvial fan site was reduced in 1993 in association with an
increase in average annual runoff from 0.4 to 12.6 cfs (the fitted curve
through the data points helps visualize this change). The reduction in depth
to groundwater was most evident at distances of ten meters or more from
the stream, where groundwater level rose about one meter in 1993.
The glacial valley and alluvial fan sites exhibit contrasting patterns of
wetland species distribution:
;
;
These box-and-whisker plots show medians and outlying values for each wetland
indicator class (as designated in Reed, 1988) and lifeform (trees, shrubs,
herbs). We find that the median is more useful than the mean as a measure
of central tendency because the median is unbiased by extreme values (outliers)
which could lead to misleading interpretations of the data.
At the glacial valley site, obligate wetland species such as willow (Salix
boothii) and facultative wetland species (Cornus stolonifera)
are mostly confined within a zone of about 15 meters from the stream channel.
The dominant tree on the site, quaking aspen (Populus tremuloides)
is classified as a "FAC" in Reed (1988) but to our knowledge this
species is confined to wetland situations in this region of the Sierra Nevada.
Our data for the glacial valley site show that quaking aspen rarely occurs
more than about 15 meters from the stream channel, or about six to eight
meters from groundwater.
At the alluvial fan site, obligate and facultative wetland species mostly
occur within a narrow band of about 10 meters from the stream channel. Most
obligate and facultative wetland woody species (trees and shrubs, including
western cottonwood and willows) occur within an even narrower region of
five meters or less from the stream. Interestingly, this obligate/facultative
wetland zone for woody species appears to be in the most hydrologically
stable area in terms of yearly changes in groundwater depth -- beyond this
zone, depth to groundwater can become significantly reduced when streamflow
declines.
We are working on further analysis of the data for all monitoring sites
and expect to devote an entire paper to the results at the 1997 Esri meeting.
EXAMPLE 3. HABITAT ANALYSIS OF THE ENDANGERED SANTA ANA RIVER WOOLLY
STAR
This last example of applying GIS to analysis of natural resources examines
the Santa Ana River Woolly Star (Eriastrum densifolium ssp. sanctorum),
a federally listed endangered plant species that is restricted to the Santa
Ana River wash in San Bernardino County, California. The plants are short-lived
perennials in the phlox family (Polemoniaceae). The bright blue flowers
of the plants attract a variety of insect and hummingbird visitors, but
only certain visitors (such as digger bees and the giant flower-loving fly)
are effective pollinators.
Biological studies of the species by Drs. Gene Jones and Jack Burk at California
State University, Fullerton have shown that the life cycle of the woolly
star is dependent on flooding frequency -- woolly stars successfully grow
and reproduce as long as their habitat has not aged to a point where competing
shrubs and annuals exclude young seedlings. With construction of a large
dam near the headwaters of the Santa Ana River (the Seven Oaks Dam), there
was concern that the dam would reduce flooding frequency and magnitude,
and thereby significantly alter the habitat of the Santa Ana River Woolly
Star. Consequently, 760 acres of woolly star habitat was set aside as a
preserve, to be managed in perpetuity to ensure survival of the woolly star.
In conjunction with Cal State Fullerton, we are using GIS to analyze existing
habitat distribution and establish a baseline set of maps that can be used
determine if woolly star habitat changes over time. This work is being conducted
in conjunction with habitat renewal experiments to establish the best method
of enhancing woolly star habitat. The project as a whole is part of implementation
of a Management Plan for the Santa Ana River Woolly Star prepared for the
U.S. Army Corps of Engineers.
The following plot shows the habitat surfaces which were digitized into
ArcInfo following visual analysis and subsequent ground truthing of aerial
photography. Two aerial photographs, which were taken in March 1995, were
scanned and merged to form the background image. The vector and raster data
were combined using ArcPress to produce the final plot.
Please refer to the graphic entitled "Alluvial Terraces in the Santa
Ana River Wash" presented at the conference. Due to file size limitations,
we regret that we are unable to include the graphic here.
This plot shows alluvial terraces classified into four categories: early,
intermediate, late, and disturbed. The classification is based on successional
stage of the plant community. A portion of the Santa Ana River Woolly Star
Preserve boundary is also shown -- due to file size limitations, only a
portion of the original plot can be shown here.
Early surfaces include the active river channel and adjacent areas dominated
by "pioneer" shrub species such as scale broom (Lepidospartum
squamatum)and California buckwheat (Eriogonum californicum).
Early surfaces away from the active channel provide favorable woolly star
habitat. Intermediate surfaces are occupied by "pioneer" species
but also by California juniper (Juniperus californica), yerba santa
(Eriodictyon trichocalyx), and cholla (Opuntia spp.). Woolly
stars are found on intermediate surfaces but in decreasing numbers as annual
herbs become more frequent. Late surfaces support sugar bush (Rhus ovata),
holly-leaved cherry (Prunus ilicifolia) and chamise (Adenostoma
fasciculatum). Woolly stars are found on these late surfaces only where
minor stream channels have disturbed the sand or where animals have moved
sand to the surface. Disturbed surfaces are those that have been subject
to substantial human impacts from mining and vehicles.
By understanding the ecology of the Santa Ana River Woolly Star in relation
to alluvial surfaces within the wash, we expect to greatly improve chances
for successful management of the habitat. With use of GIS, we expect to
document early indications of habitat change (e.g., increases in area of
old or disturbed habitat), leading to prompt management action to prevent
irreversible loss of woolly star populations.
SUMMARY
We have shown three diverse examples of how GIS programs can be customized
for three types of natural resource applications: groundwater studies, wetland
studies, and endangered species management. Clear identification of project
goals prior to development of the GIS database enabled efficient integration
of biological and hydrological data. However several features demonstrated
in these examples, such as the Stiff diagrams for water quality and meter-level
display of the riparian vegetation data, required substantial programming
effort because these types of displays had not been attempted before. The
time required for such customized GIS applications should be an important
consideration for others who are interested in GIS for natural resource
applications.
ACKNOWLEDGEMENTS
We would like to thank Robert Johnson and Scott D. White for assisting in
field data collection, and Duane Haselfield for his patient dedication to
GIS. We also thank John Irwin and Gene Hawkins of Southern California Edison,
and Terry Hicks of the Inyo National Forest, for their comments on draft
versions of the riparian vegetation plots.
REFERENCES
Reed, P.B., 1988. National List of Plant Species That Occur in Wetlands:
Intermountain - Region 8. U.S. National Ecology Research Center, Fort Collins,
Colorado.
Stiff, H.A., 1951. The interpretation of chemical water analyses by means
of patterns, Journal of Petroleum Technology, v. 3, n. 10, secs. 1-3.
AUTHOR INFORMATION
Edith A. Read, Ph.D.
Manager of Biological Resources
Psomas and Associates
3187 Redhill Ave., Suite 250
Costa Mesa, CA 92626
phone: 714.751.7373
fax: 714.545.8883
email: eread@psomas.com
Jennifer Gough, L.S.I.T.
GIS Project Manager
Psomas and Associates
3187 Redhill Ave., Suite 250
Costa Mesa, CA 92626
phone: 714.751.7373
fax: 714.545.8883
email: jgough@psomas.com