Tải bản đầy đủ - 0 (trang)
Remote Sensing and Spatial Analysis of Watershed and Estuarine Processes for Conservation Planning in Elkhorn Slough, Monterey County, California

Remote Sensing and Spatial Analysis of Watershed and Estuarine Processes for Conservation Planning in Elkhorn Slough, Monterey County, California

Tải bản đầy đủ - 0trang

496



K.B. Byrd



Fig. 21.1 The Elkhorn Slough watershed and Moss Landing Harbor, located on the coast of Monterey Bay in Central California



extent only second to that of San Francisco Bay (Fig. 21.1). The slough is a seasonal estuary extending inland for 11.4 km from Monterey Bay and contains approximately 1,090 ha of estuarine habitat types including subtidal channels, tidal creeks,

mudflats, salt marshes, and tidal brackish marshes (Elkhorn Slough Tidal Wetland

Project Team 2007). As only about 3% of conterminous United States salt marsh



21



Remote Sensing and Spatial Analysis of Watershed and Estuarine Processes



497



acreage occurs along the Pacific Coast (Field et al. 1991), this represents a significant resource for wildlife, fisheries, research and education, and recreation. Despite

its proximity to the urbanized Silicon Valley (about 90 km), Elkhorn Slough and

its surrounding watershed have retained high biodiversity, attractive viewsheds, and

open space. Elkhorn Slough estuarine habitats support over 340 birds (135 aquatic

species), 550 marine invertebrates, and 102 fish species including 24 rare, threatened or endangered species (Caffrey et al. 2002). The watershed is drained by many

small seasonal streams and one main channel, Carneros Creek. Steep hills rise 30

to 100 m from the marsh; while many hills are cultivated, the uplands contain a

rich matrix of oak woodland, grassland and ridges covered with rare central maritime chaparral plant communities, including several endemic plant species. This

diverse, interdependent complex of upland and wetland habitat types supports a diversity of species that may be the highest in California for a watershed of its size,

182 km2 (Elkhorn Slough Foundation and Tom Scharffenberger Land Planning and

Design 2002).

Over a century of human activity has affected Elkhorn Slough’s natural environments in ways that have altered ecosystem processes required to sustain the rich

complex of habitats and species diversity both within the estuary and its watershed.

In 1872 a raised embankment for the Southern Pacific Railroad was constructed

through the marshlands, and during the early to mid 1900s tidal wetlands were converted to pastureland for dairy operations, salt evaporation ponds, and waterfowl

habitat (Van Dyke and Wasson 2005). In 1947 the U.S. Army Corps of Engineers

modified the Elkhorn Slough main channel and created a new opening to Monterey Bay to accommodate vessel traffic in the newly created Moss Landing Harbor.

Prior to this point, Elkhorn Slough joined the Salinas River, which meandered north

along the shoreline and entered the bay approximately 1 km from the present-day

harbor (Fig. 21.2). In less than 50 years the slough changed from a largely freshwater regime dominated by deposition from the Salinas River and surrounding watershed to a highly saline system characterized by stronger tidal flow and greater

tidal reach (Schwartz et al. 1986, Crampton 1994). Within the watershed, farming

on the plateau west and north of Elkhorn Slough began in the late 1880s with establishment of potatoes and sugar beets (ABA Consultants 1989) and intensified in

the 1970s with the expansion of strawberry farms in response to growing markets

(Dickert and Tuttle 1980, Caffrey et al. 2002). Between 1981 and 1993, crop area

in the watershed increased 29%, with strawberry acreage increasing 53% (USDASCS 1994). Currently 24% of the Elkhorn Slough watershed is intensively farmed

(Caffrey et al. 2002).

The ridge tops and south facing slopes of the eastern portion of the upper watershed, the Elkhorn Highlands, are highly desirable for farmers and residential buyers,

which threatens large areas of undisturbed central maritime chaparral. A majority of

the soils in the watershed uplands are derived from the Aromas sands formation, an

aeolian sandy parent material producing soils with high sand content that are highly

erodible when disturbed (Fig. 21.3). As the hills are highly susceptible to erosion,

the health of aquatic habitats is intertwined with preservation of central maritime

chaparral. With vegetation clearing, the sandy soils underneath can be washed away



498

Fig. 21.2 Aerial photographs

comparing the mouth of

Elkhorn Slough in 1931 (prior

to the 1947 construction of

Moss Landing Harbor) and

1949. Circles represent where

the Elkhorn Slough estuary

enters Monterey Bay. Source:

Elkhorn Slough Tidal Wetland

Project Team (2007)



K.B. Byrd



21



Remote Sensing and Spatial Analysis of Watershed and Estuarine Processes



499



Fig. 21.3 A mosaic of habitat types, including pickleweed-dominated diked salt marsh in the foreground, agricultural fields on the hillside, and coast live oak woodland and central maritime chaparral on the hilltops. Source: Tuxen (2003)



by winter runoff and greatly impact downstream wetlands (Elkhorn Slough Foundation and Tom Scharffenberger Land Planning and Design 2002).

The Elkhorn Slough watershed spans a complex patchwork of public and

privately-owned lands. The southeastern portion of the estuary contains a NOAA

National Estuarine Research Reserve (ESNERR), designated in 1979. The California Department of Fish and Game manages the 631 ha Research Reserve along with

583 ha of other wetlands including the Moss Landing Wildlife Management Area.

The Elkhorn Slough Foundation, a land trust, and the Nature Conservancy have

acquired through fee or easement 194 ha of wetland and 1227 ha of upland for conservation purposes. Together these land acquisitions and state designations account

for a quarter of the estuary’s habitats (Elkhorn Slough Tidal Wetland Project Team

2007). Cooperation among public and private owners, state and local agencies, and

academic institutions have facilitated a conservation and restoration planning process targeting protection of important biological resources.

Likewise through the involvement of multiple institutions, Elkhorn Slough has

become a nexus of remote sensing research. The ESNERR brings an inflow of

GIS and remote sensing resources from the NOAA Coastal Services Center. Proximity to Silicon Valley, the home of NASA Ames Research Center, and the high

concentration of research institutions in Monterey Bay, including the U.S. Naval



500



K.B. Byrd



Postgraduate School, the Monterey Bay Aquarium Research Institute, the University

of California, Santa Cruz and California State University, Monterey Bay also provide significant resources.

This chapter describes the influence of a conservation planning process in Elkhorn

Slough on research applying GIS and remote sensing technologies. The chapter

features innovations in geospatial research that address impacts to this estuarine

system. It further highlights how results from these studies have informed recent

planning strategies. Three areas of research relevant to Elkhorn Slough conservation and restoration are covered: historical ecology, connections between watershed

land use and estuarine response, and change detection of tidal marsh and slough

bathymetry.



21.2 Early Planning Initiatives

By 1999 State and federal land designations as well as several conservation programs were in place, and several individuals representing non-profit, local and

state government, and academic institutions came together to develop a Watershed

Conservation Plan that would guide future conservation activities. Central to this

planning process was the development of a GIS, which was used to assess spatial

distribution of biological resources and land use patterns in order to prioritize lands

for acquisition, management and restoration. The plan identified critical resources,

most significant threats to these resources, and strategies to protect these resources

over time (Scharffenberger et al. 1999). Critical resources, or conservation targets,

were defined as natural wetland and upland habitats and the rare and endangered

species they support, and productive agricultural areas vital to the local and county

economy. The report identified the most significant threats to Elkhorn Slough to

be: (1) sedimentation and contamination of marshes mostly due to uncontrolled

runoff from steep cultivated fields, (2) destruction and fragmentation of maritime

chaparral habitat from residential development, (3) depletion of groundwater and

accompanying seawater intrusion from excess pumping of wells for irrigation and

(4) loss of marsh habitat by tidal erosion and conversion from human manipulation

of marsh hydrology. These threats remain a high priority for resource managers, and

threats (1) and (4) have been the focus of remote sensing-based research over the

past eight years.

The location, area and distribution of conservation targets were mapped from

aerial photography, and a planning strategy was built around five conservation zones

delineated to represent major regions of the watershed that contained the most intact remaining conservation targets in the area. Generally, overarching conservation

strategies applied to each zone were to acquire unprotected parcels through fee title or conservation easement, create vegetated buffers between cultivated fields and

wetlands, and restore natural habitat where possible.

These conservation strategies were further refined in a follow-up planning document in 2002, called “Elkhorn Slough at the Crossroads” (Elkhorn Slough Foundation and Tom Scharffenberger Land Planning and Design 2002). One of its



21



Remote Sensing and Spatial Analysis of Watershed and Estuarine Processes



501



major contributions was a more complete understanding of the watershed’s sensitive habitats through detailed vegetation mapping. Authors of the report identified

a high diversity of unusual interdependent habitats and reinforced the need to properly manage land uses since aquatic resources were so heavily impacted by runoff

from cultivation and development. Upon completion of the report, the director of the

Elkhorn Slough Foundation, Mark Silberstein stated, “The process of developing the

GIS and the layers of information behind the maps has brought a new appreciation

for the importance, beauty and fragility of the slough. Accurate mapping provides

the basis for sound land use decisions and we supply this information in the spirit

of informed dialogue as the community works toward a vision of the future of this

part of the coast and toward a new General Plan for the County.”

The keystone of the report was a vegetation map created through hand digitizing aerial photography in order to update the 1999 Watershed Conservation Plan’s

biological and agricultural resource inventories. The map contained 18 classes including dominant vegetation types and land uses such as agriculture, greenhouses,

and developed areas. From these maps and coordinated efforts a vision was developed to create an intact and interconnected network of natural communities including over 1616 ha of coastal marsh within Elkhorn Slough and Moro Cojo Slough to

the south, enhanced freshwater wetlands of McClusky Slough to the north, restored

riparian forest along the lower Carneros Creek floodplain, and a series of upland

ridges with unfragmented maritime chaparral in the Elkhorn Highlands. These natural communities were to be surrounded by productive, habitat-compatible farmland,

scenic vistas and residences.



21.3 A Remote Sensing Approach to Historical Ecology

The implementation of conservation and restoration strategies in Elkhorn Slough required information about the area’s environmental baseline conditions prior to major

human modification of the watershed. Historical information has become increasingly important in setting sustainable management goals and has become central to

the field of restoration ecology (Swetnam et al. 1999). Historical ecology is characterized by the use of a long time sequence of measurements or observations to

gain information about changes in populations, ecosystem structures, disturbance

frequencies, process rates, trends, and periodicities (Swetnam et al. 1999). Applied

historical ecology involves the use of historical knowledge in ecosystem management and planning.

Recognizing the importance of historical information for informing conservation decisions, an historical ecology research program was developed at ESNERR.

This program was made feasible by the rich archive of historical maps and aerial

photographs of the region, due to the coastal location of the watershed and its proximity to the historical city of Monterey (Van Dyke and Wasson 2005). The direction of historical ecology research at ESNERR and elsewhere were influenced by

two anthropogenic impacts discussed in the early planning documents: long-term



502



K.B. Byrd



human modification of marsh hydrology and sedimentation into the slough from

eroding upland cultivated areas. As research progressed, innovative remote sensing

techniques were developed to discern historical environmental conditions from the

archival materials and to assess historical ecological changes. The use of historical geographical data, like aerial photographs, for map-making is a key component

of historical ecology research, as maps most clearly illustrate the relationship between physical processes and habitats, habitats and species, and habitats of different

kinds (Grossinger 2001). Maps also can help identify significant historical habitat

types and current habitat remnants that are an important source of regional species

diversity (Grossinger 2001).

Aerial photography, which began early in the 20th century, is an especially

important tool for monitoring long-term ecological changes. Not only can aerial

photographs be used to quantify ecosystem changes, but they can help explore explicit linkages between ecosystem change and human resource uses that drive those

changes. While several remote sensing data sources are available to map coastal areas and wetlands [e.g. Light Detection and Ranging (LiDAR) data, Landsat satellite

imagery, Airborne Data Acquisition and Registration (ADAR) imagery], only color

or color infrared (IR) aerial photography provides the combination of spatial and

spectral resolution and temporal scale required to conduct long-term historical time

series analysis of wetland change at fine spatial scales such as the plant community level.

Despite the advantageous spatial resolution provided by most aerial photography, the method of image capture creates several challenges when using the data.

Problems with color balance and geometric distortion exist due to camera tilt, relief displacement, lens distortion, and atmospheric refraction, which can cause scale

variation and horizontal error (Dickert and Tuttle 1980, Bolstad 1992, Barrette et al.

2000) and must be rectified. Contact printing and scanning introduce more distortion

(Moore 2000), and the process of matching ground control points during rectification is another source of error (Van Dyke and Wasson 2005). Ideally, the problem

being studied would determine the time between image dates chosen for historical

studies or the frequency of change detection. Ultimately however, the availability of

imagery influences the choice of image dates and what historical processes can be

identified (Klemas 2001).

At ESNERR a project was initiated to track trends in tidal wetland habitat

changes due to human modification of the Elkhorn Slough system. Developing an

efficient method for georeferencing and mosaicking multiple color and color IR

contact prints was a key step to extracting information from historical aerial photographs. Van Dyke and Wasson (2005) scanned, georectified and mosaicked 26

historical maps and charts dating from 1853 to 1925 and 13 aerial photograph

flights taken between 1931 and 2003, which together comprised more than 300

photos. This combination of black and white, true color and color IR aerial photos were scanned to produce a resolution of 0.6 m/pixel after rectification. Ground

control features were identified in 0.6 m/pixel digital orthophotographs. Because

image distortion increases with increasing distance from the image center, the team

constructed the mosaic using the central portion of each overlapping image. This



21



Remote Sensing and Spatial Analysis of Watershed and Estuarine Processes



503



central “effective area” was identified with a script that applied a proximity function

and created a honeycomb polygon shapefile that served as a mask for assembling the

mosaic. Control points were located at well spaced locations near the perimeter of

the effective areas. Because camera tilt was likely the greatest source of distortion,

the images were resampled with a four-point plane projective model.

Van Dyke and Wasson (2005) conducted a marsh and tidal creek time series analysis on 196 fixed 100 m2 quadrats and 196 tidal creek cross sections in undiked areas

from 12 dates ranging between 1931 and 2003. Within each quadrat and date they

determined the percentages of vegetated salt marsh and unvegetated areas. Typical

of Northern California salt marshes, the dominant vegetation at Elkhorn Slough is

pickleweed (Sarcocornia pacifica), a low-growing succulent plant that forms dense

monospecific stands across the marsh plain.

Black and white historical aerial photographs provide limited capacity to differentiate plant community types, especially through automated classification methods.

To address this issue image classification was accomplished using a custom interactive application to perform semi-automated image interpretation. Starting with

the first quadrat, the user selects a threshold grayscale value that defines the isoline boundaries between vegetated and bare portions, then the tool automatically

produces the corresponding set of polygons (vegetated and unvegetated) within a

shapefile. Moving across the aerial photo, the initial grayscale selection seeds the selection for the following quadrat, and the tool allows the user to adjust the grayscale

value to account for contrast variations among quadrats. All analysis was performed

on grayscale imagery; for color IR photos the researchers converted red-green-blue

to hue-saturation-intensity then interpreted the intensity component.

Interpretation of these aerial photos demonstrated that major changes to Elkhorn

Slough’s wetland habitats occurred over 150 years. Since 1870 more than two thirds

of the slough’s salt marsh has either degraded or been converted to other habitat

types. From 1870 to 1956 the construction of more than 60 km of levees and embankments reduced the range of unobstructed tidal influence by 59%. During this

same period of extensive diking, the extent of intact salt marsh (veg cover >75%)

decreased by 66%. By 2000 the extent of high-quality marsh was 23% of its coverage 100 years before (Fig. 21.4) (Van Dyke and Wasson 2005).

From the quadrat analysis it became clear that much of the upper slough that was

once densely vegetated is now mudflat and open water. The mean percentage of salt

marsh vegetation cover in undiked areas decreased from 89.6% to 46.4% in 2003.

Tidal creeks have also become wider in undiked areas, from an average of 2.5 m

in 1931 to 12.4 m in 2003. The extent of marsh loss and changes in creek channel

width increased with increasing distance from Monterey Bay (Fig. 21.5).

The changes to Elkhorn Slough’s marshlands over 100 years were attributed to

two major alterations in the slough’s hydrology: restrictions to the range of tidal flow

occurring earlier in the study period and expansion of tidal range, amplitude, and velocity that occurred since the opening of Moss Landing Harbor in 1947 (Van Dyke

and Wasson 2005). The newly created, deeper, wider channel entrance to Monterey

Bay increased the velocity and amplitude of tidal exchange within the slough, transforming it into a highly erosional system. The marsh quadrat and tidal creek analysis



504



K.B. Byrd



Fig. 21.4 An example of

evolution of salt marsh to mud

flat. Dark areas are salt marsh,

light areas are unvegetated.

(a) Tidal creek network and

growing interior panes, 1980

aerial photo. (b) Deteriorated

marsh largely converted to

open mud flat, 2001 aerial

photo. Source: Van Dyke

and Wasson (2005), Fig. 8,

p. 185; c 2005 Estuarine

Research Federation, with

kind permission of Springer

Science and Business Media



showed that conversion of salt marsh habitat to mudflat and widening of tidal creeks

accelerated only after 1949, timing that coincides with artificial channel opening

(Van Dyke and Wasson 2005).

Moving up into the watershed, an earlier study (Dickert and Tuttle 1980) examined historical land use change between 1931 and 1980 and calculated erosion rates

associated with different land uses, soil types, and slopes. The authors identified

sediment fans resulting from this erosion that had formed in the marsh at the base of

slopes draining into the slough. Between 1931 and 1980 there was a 5-fold increase

in the number of sediment fans and a doubling of their acreage in the pickleweed

marsh as agriculture in the watershed increased by 282%. By 1980 at least 30 sediment fans had formed in the salt marsh, ponds, and freshwater marsh of Elkhorn

Slough.

Another historical ecology project based out of the University of California,

Berkeley, addressed this problem of watershed erosion and examined how sedimentation from eroding cultivated areas buried salt marsh vegetation at the upland

margin of the slough. Byrd et al. (2004) analyzed the same archive of historical

aerial photos used by Van Dyke and Wasson to produce a time series of vegetation

change on sediment fans over four decades. A combination of black and white and

color IR aerial photographs and orthophotos from May 1971, April 1980, May 1992,



21



Remote Sensing and Spatial Analysis of Watershed and Estuarine Processes



505



Fig. 21.5 Annualized change, vegetation cover and tidal creek width. 1931–1956: (a) High to very

high marsh loss at lower slough; moderate loss at mid and upper slough. (b) Low to moderate

overall tidal creek width increase. 1956–1980: (c) Little change or marsh recovery at lower and

mid slough; very high loss at upper slough. (d) High to very high creek width increase at upper

slough and southern part of lower slough; low to moderate increase elsewhere. 1980–2003: (e)

Moderate to high marsh loss at lower and mid slough; very high loss at upper slough. (f) Moderate

to very high creek width increase at lower and mid slough; very high increase at upper slough.

Source: Van Dyke and Wasson (2005), Fig. 6, p. 183; c 2005 Estuarine Research Federation,

with kind permission of Springer Science and Business Media



and May and June 2001 were chosen for analysis, and May 1931 aerial photographs

provided reference conditions. Again all photos were scanned to generate digital

images with a resolution of approximately 0.6 m/pixel. This sub-meter resolution

was required to differentiate vegetation types in the salt marsh. Fifteen sediment

fans present in salt marsh, both diked and undiked, were chosen as study areas.



506



K.B. Byrd



The color IR images (1980, 1992, and 2001) were classified based on a plant

community level scheme derived from “A Manual of California Vegetation” (Sawyer

and Keeler-Wolf 1995), and classes included bare soil (including mudflat), pickleweed, saltgrass/jaumea (Distichlis spicata/Jaumea carnosa), bulrush/cattail

(Schoenoplectus spp./Typha spp.), arroyo willow/coast live oak (Salix lasiolepis/

Quercus agrifolia), coyote brush scrubland (Baccharis pilularis), and California annual grassland. In the black and white images (1931 and 1971), only pickleweed and

mudflat areas were classified.

Overall, a hierarchical supervised classification of color IR aerial photographs

was successful in discriminating several plant communities. The study area had

little heterogeneity within plant communities that were for the most part monotypic, which lent to a greater success rate. The high spatial resolution (0.6 m/pixel)

enhanced separability among classes and likely reduced the number of spectrally

mixed pixels. This resolution allowed for the analysis of vegetation change in relatively small areas. Together, all 15 study sites comprised 37 ha. Additional bands

generated from the original IR, red and green bands also proved necessary for a successful classification. Use of NDVI was instrumental in separating vegetation from

bare soil, and the IR and green variance texture bands contributed to the separation

of forest from grasses and shrubs.

A time series analysis from 1980 to 2001 identified the relative change in area

among pickleweed, bare soil, saltgrass/jaumea, and willows. The results demonstrated a process of succession that was typical on five sediment fans, all located on

the western side of the slough where tidal action was still present. Between 1980 and

1992 as sedimentation occurred, pickleweed and mudflat were replaced by saltgrass

and jaumea on fans below upland drainages. Between 1992 and 2001 sedimentation

likely continued, especially during the strong El Ni˜no winter of 1997–1998, and

arroyo willow took the place of saltgrass and jaumea and continued to extend into

the marsh plain. On the eastern side of the slough the pattern of successional change

lacked the intermediate stage of saltgrass and jaumea, but the final outcome was still

an expansion of willow cover and movement of willows into the salt marsh. Overall willow expansion occurred on 11 of 15 sediment fans encompassing an area of

4.75 ha (Fig. 21.6). Of this expansion 57% occurred as willows moved into grassland and 43% as willows extended into the marsh.

Though not a main objective of this project, it became evident that the postclassification change detection methods applied here separated effects of tidal

erosion and sedimentation, the two contrasting processes of salt marsh loss both

occurring within the study areas. A from-to matrix generated by the change detection from 1980 to 2001 revealed that pickleweed within the study area declined by

3.64 ha. Eighteen percent of the pickleweed was converted to mudflat, which can be

interpreted as an impact of tidal erosion, the likely cause of pickleweed loss to mudflat (Van Dyke and Wasson 2005). Fourteen percent of pickleweed was converted to

another vegetation class, primarily arroyo willow, and represents the establishment

of new species where sedimentation occurred.

Despite the challenges associated with historical aerial photographs, such as image quality, scale, and image dates, they enable the study of historical ecological



Tài liệu bạn tìm kiếm đã sẵn sàng tải về

Remote Sensing and Spatial Analysis of Watershed and Estuarine Processes for Conservation Planning in Elkhorn Slough, Monterey County, California

Tải bản đầy đủ ngay(0 tr)

×