WEB SITE FOR THIS DOCUMENT

http://www.usu.edu/remap/wat2000c/paper/wat20v1.html

CITATION

Goodwin, C. N., Rasmussen, G. A., and Dobrowolski, J. P., 2000.  Rangeland watersheds: integrated monitoring and modeling within a GIS.  Proceedings of the Watershed Management 2000 Conference, Vancouver, British Columbia, July 9-12, 2000.  Available on CD-ROM from the Water Environment Federation, Alexandria, Virginia. WEF homepage: http://www.wef.org.

ABSTRACT

Presented herein is a protocol for developing a monitoring program for assessing the overall hydrologic stability of rangelands at the watershed scale. Rangelands dominate the world land area (47 percent) and are greater than 70 percent of the area of the western United States. Erosion and sedimentation issues are the predominant land and water quality issues of rangeland areas. A variety of processes deliver non-point source pollutants to streams within rangeland environments, and human actions frequently modify these processes and increase the rate of soil loss. Increased rates of soil loss not only reduce long term sustainability of the upland ecosystem, but may also increase the rate of sediment delivery to the drainage network therein affecting both stream condition and water quality. Monitoring to determine trends in soil loss rates from rangelands is therefore a critical element in managing the sustainability of these ecosystems.

Recent emphasis on ecosystem management has forced natural resource managers to evaluate and manage rangeland ecosystems at much larger scales and for more diverse resources than has been undertaken in the past. Unfortunately, holistic monitoring approaches, designed to define environmental changes and the causes thereof across large areas and multiple resources, are few. The major product of the research is a watershed-based monitoring protocol that will help managers monitor and understand the cause of erosion- and sediment-related environmental changes in a watershed. The protocol uses GIS technology and conceptual and mathematical models for its implementation.

KEYWORDS

Monitoring, watersheds, rangelands, modeling, GIS, WEPP, TOPMODEL

INTRODUCTION

Recent emphasis on ecosystem management has forced natural resource managers to evaluate and manage natural systems at much larger scales and for more diverse resources than has been undertaken in the past. Monitoring, which is an essential component in providing information to direct natural resource management decisions, has traditionally be compartmentalized by the requirements of specific technical disciplines and has most typically been undertaken at the site scale. Therefore, most environmental monitoring methods have focused on single-objective measurements aimed at assessing a specific resource at a specific location. These simple monitoring schemes often lead to problems when an attempt is made to correlate or aggregate the monitoring information for use with "bigger picture" picture problems of ecosystem management. Additionally, although simple monitoring schemes may indicate a change in ecosystem condition, they often provide little evidence as to why the change occurred. Holistic monitoring approaches, designed to define environmental changes and the causes thereof across large areas and multiple resources, are few.

Rangelands dominate the world land area (47 percent) and are greater than 70 percent of the area of the western U.S. In the United States, these lands are managed for numerous societal values and products that include food, fibers, minerals, and recreation. These areas also provide and affect the quantity and quality of water utilized in most of the western U.S. The quality of water delivered from rangelands represents the cumulative effects of geologic, geomorphic, soil, vegetation, climatologic, and hydrologic conditions of these diverse, complex landscapes. Human activities can change the hydrologic stability and other conditions of rangelands, which, in turn, can severely affect water quality conditions. A variety of processes deliver non-point source (NPS) pollutants to streams within rangeland environments, and human actions frequently modify these processes and increase the rate of soil loss. Increased rates of soil loss not only reduce long term sustainability of the upland ecosystem, but may also increase the rate of sediment delivery to the drainage network therein affecting both stream condition and water quality. Monitoring to determine trends in soil loss rates from rangelands is therefore a critical element in managing the sustainability of these ecosystems.

The purpose of this investigation was to develop a protocol for monitoring and assessing the overall hydrologic stability of rangelands at the watershed scale. The watershed is rapidly becoming recognized as the fundamental natural landscape element for land and water analysis and management decision making. When water-caused erosion and sedimentation are the predominant issues of concern in an area, a watershed based approach for monitoring, assessment, and management is essential. The major product of the research is a watershed-based monitoring protocol that will help managers monitor to understand the cause of erosion- and sediment-related environmental changes in a watershed. The protocol uses GIS technology and conceptual and mathematical models for its implementation. Although some specific data collection methods are required to implement components of the protocol, generally, specific data collection methodologies are not part of the protocol. Instead, a flexible watershed-based framework is provided that can be adapted to non-sediment resource management issues.

MONITORING BASICS

Monitoring is defined as ‘to watch, observe, or check on for a specific purpose’ (Webster’s, 1983). In natural resource management, monitoring is done is to determine if the existing management strategy is meeting specific goals. For environmental monitoring and natural resource management, a more specific definition has been developed – the orderly collection, analysis, and interpretation of data to evaluate progress toward objectives (SRM, 1989). Determining whether management objectives are being met is the purpose for monitoring. The fundamental goal is to answer the question: do we understand what our actions are doing to a particular ecosystem?

Most environmental monitoring methods have concentrated on single-objective measurements (e.g. stream condition, upland, or riparian ecological conditions) to assess complex ecological or hydrological conditions. These simple schemes often lead to problems when other investigators or agencies try to correlate or aggregate monitoring information for other purposes or bigger picture problems associated with ecosystem management. For example, upland and riparian monitoring have typically focused on biological objectives. Examples are livestock impacts on plant composition, deer use on winter ranges, or recreation impacts on riparian areas. While biological objectives such as biodiversity are important, they do not always correlate to other ecosystem properties such as stream condition or water quality. Likewise, stream condition monitoring can determine conditions at a single point on the stream network. However, these measures are difficult to link to upland and riparian condition because no information on hydrologic stability of those areas was obtained. Thus, single index monitoring may indicate a change in ecosystem condition, but there is little chance to determine why the change has occurred.

Managers have not linked existing monitoring methods on upland and riparian areas with stream condition. However, upland and riparian monitoring techniques were developed to consider the ecological condition of a site from a biological perspective (i.e. production, seral status, biodiversity etc.). Generally, these methods do not evaluate the hydrologic stability of the site despite the fact that this stability is essential to understanding stream and water quality conditions. On rangelands with few perennial streams and associated riparian areas (<2% of watershed area) hydrologic stability is of paramount importance. Upland and riparian areas represent the cumulative impacts on 98% of the area, but little or no data exist regarding hydrologic stability of the upslope or upstream watershed.

As science has provided more information about the parts of the systems and their interconnectedness, the importance of looking at the entirety of the system is being realized. The current economics of declining budgets and reduced time that can be spent on monitoring makes it inevitable that single-objective monitoring will be harder to accomplish in the future. Monitoring protocols must become more strategic to ensure that any protocols can meet the basic needs of managers to deal with the multiple objectives found on a particular system as well as the integrity of the entire system. To develop a strategic approach to monitoring requires that more time to be spent understanding what questions need to be asked in order to meet the different management objectives on a single watershed. This is important because budgets will not allow repetitive sampling in a single watershed for different objectives.

These issues force managers to consider the most important questions in developing a monitoring program. Such questions will not always correlate to the managers’ most important interests. However, methods that have been designed for single objectives often miss what is happening in the other parts of the system. Using modeling within a GIS system provides a viable alternative to linking data on uplands, riparian areas, and streams together to answer larger scale questions as well as point managers to areas of specific problems. This may allow an economic alternative to deal with monitoring multiple scales.

Monitoring programs generally are established to identify short- or long-term trends in system conditions. However, the spatial organization and dynamics of a system must be considered in the design of appropriate monitoring strategies. Within a watershed, stream networks act to route water, sediment, and dissolved materials toward the basin outlet. This network represents an inherent spatial structure reflected by the basic topology of the stream system (Figure 1). In order to assess accurately the impacts of rangeland or riparian conditions on stream conditions, it is necessary to first recognize the spatial order created by the network of stream channels and other flow pathways.

What is measured depends directly on what questions are being asked. There have been single-objective measurements that are important, but when systems are evaluated, there is a need to select techniques that are capable of assessing linkages to the other parts of the system. Many upland monitoring programs often use measurements like plant species frequency to determine biological objectives. However, when evaluating whole systems, these measurements are difficult to link to other components. Measurements like frequency are often selected because on time constraints but in a strategic monitoring program using a measurement like cover would allow linkages to be understood within the system.

BASIS OF THE RANGELAND WATERSHED MONITORING (RWM) PROTOCOL

Our approach to the monitoring problem can be best described as developing a protocol that is top down or strategic in nature, in which an understanding of watershed processes is used to guide selection of the most appropriate monitoring methods and sites. Figure 2 illustrates the features of the top-down (versus bottom-up) approach that we have modified from Landers (1997). Bottom-up, tactical approaches are site specific, discipline focused, and methods driven. They are reductionist, not holistic. A strategic approach is holistic, and when issues of erosion and sedimentation are paramount, then a watershed-based strategy is essential. The critical connections between a strategic protocol design and monitoring site implementation are (Landers, 1997):

Scientific understanding of the process/function linkages of watershed condition,

Consideration of environmental change at multiple temporal and spatial scales, and

Knowledge to select robust indicators of environmental change.

Thus, our protocol development is not based upon the generation of specific, new monitoring methods, for which there are many adequate examples. Rather, it centers upon developing an overall strategy for holistically monitoring rangeland watersheds. For testing purposes, we are implementing several specific data collection methods, but these methods could be replaced with different methods when the monitoring issue is something other than hydrologic stability. Overall, the protocol strategy is based upon 1) understanding the physical processes of erosion and sedimentation (hydrologic stability), 2) knowledge of how these processes vary spatially and temporally within a watershed, and 3) use of conceptual and mathematical models to assess the spatial and temporal variability in erosional processes.

Hydrologic Stability

Hydrologic stability as we define it herein is the ability of a site to hold soil in place and is dependent upon geomorphic and vegetation conditions. Areas of poor hydrologic stability have high soil erosion and sediment yield rates and are usually poorly vegetated. In western rangeland landscapes where water-limited conditions keep vegetation cover below 100 percent levels, hydrologic stability is an important issue in devising sustainable land management plans and in improving water quality. The Society for Range Management (SRM, 1995) has stated that the "most important and basic physical resource" of rangeland sites is the soil, and that if too much soil is lost due to erosion, the potential of the site is reduced. In areas of the western U.S. where rangelands dominate, sediment problems produce the major water quality issues, and sediment-based TMDLs are being implemented to alleviate these problems and improve water quality. Addressing causes and sources of a water quality impairment are a necessary component of developing a sediment TMDL (USEPA, 1999).

Most typically, rangeland monitoring has been undertaken using methods that ascertain the ecological condition of a site from a biological perspective (Joyce, 1993). Generally, these methods do not evaluate the hydrologic stability of a site despite the fact that this stability is essential to understanding land and water quality conditions. At the other extreme, water quality sampling for sediment is infrequent, and usually occurs far downstream on high-order rivers. Such downstream sampling does not allow the determination of causes or locations of problem areas of hydrologic instability within a river basin.

Changes in hydrologic stability -- or more specifically changes in rates of soil and sediment erosion -- are the primary concern of the protocol. Soil and sediment erosion has been categorized as belonging to a specific group of monitoring measures termed geoindicators. Geoindicators "are measures of geological processes and phenomena occurring at or near the Earth’s surface and subject to changes that are significant in understanding environmental change over periods of 100 years or less" (Berger, 1996). Processes measured by geoindicators are those that may change over time due to either natural or human activities. The soil and sediment erosion geoindicator can potentially be used in any landscape, but is best applied to uplands and valley bottomlands mantled with soil or loose sediment (Berger and Iams, 1996). Measuring rates of watershed erosion are difficult. There are three general approaches to determining these rates:

Direct measurement methods. These methods attempt to directly measure erosion rates and include methods such as using erosion pins to determine soil loss/buildup, sediment tracers to detect soil movement, or channel cross section measurements to assess bank erosion/sedimentation. Measuring sediment buildup in a reservoir or measuring stream sediment load can provide indications of erosion rates, although these methods do not provide a location or cause of an erosion problem.

Indirect measurement methods. Indirect methods estimate erosion by measuring parameters that are correlated with erosion but are often more easily measurable. For example, erosion rates tend to increase with decreasing vegetation cover, so percentage vegetation cover may provide a surrogate measure of erosion potential.

Modeling methods. A potentially more accurate indirect method for determining erosion rates when compared with simple indirect methods are the various erosion models such as the Revised Universal Soil Loss Equation (RUSLE) or the Water Erosion Prediction Project (WEPP).

Indirect and modeling methods are the primary means for determining hydrologic stability within the protocol. However, nothing precludes using combinations of methods derived from the above three categories. Any methods that meet a monitoring program’s strategic goals and are robust indicators of environmental change can be used.

Watershed-Based Approach

A key aspect of the monitoring protocol is that it utilizes a watershed-based geographic structure for data collection and analysis. Watershed-based approaches have been recommended by many for ecosystem analysis (Frissell et al., 1986; Gregory et al., 1991; Allan and Johnson, 1997; Johnson and Gage, 1997). The adoption of a watershed-based approach for this monitoring protocol is founded upon the concept that sediment, solutes, and water link watershed headwater conditions to downstream water quality, stream, and riparian conditions. The physical linkages between headwaters and downstream areas are flow pathways, with gravity being the primary driving force. Thus, by knowing the spatial location of monitoring sites with respect to a watershed’s structure, it may be possible to gain additional insight into the causes of changes at a these sites. Walling (1983), for example, states that a detailed understanding of sediment delivery at a watershed’s outlet depends in part upon knowing the locations of sediment sources within the watershed. In particular, certain areas within a river basin may be more sensitive to environmental change or human impacts. Monitoring small watersheds may therefore have relevance to and be representative of the behavior of the larger river basins in which they are situated (Burt et al., 1993).

We conceptualize a watershed as consisting of a flow path or drainage network and of hillslopes that drain into the network. Flow pathways are non-hillslope areas of concentrated surface and subsurface flow associated with valley bottoms. Flow pathways occur as corridor elements in a watershed’s tree-like network structure. Stream channels and riparian areas are located within the flow path network, but the upstream elements of the network may be neither channelized nor contain a riparian area. Unchannelized hollows, however, may transmit groundwater flows and may become channelized under certain environmental conditions. Points along or reaches of the flow path network are selected for monitoring sites to assess changes in network (channel, riparian area) conditions.

A watershed’s uplands are comprised of a set of hillslopes that begin at drainage divides and terminate at a flow path corridor. Each hillslope consists of one or more overland flow elements (OFEs) where runoff occurs as either sheet or rill flow. An OFE is a hillslope area or strip of uniform soil, vegetation, geomorphic, topographic, and hydrologic properties (Figure 3). Runoff generation and sediment production within an OFE is dependent upon these properties. Ecological sites (SRM, 1995) are also areas of unique vegetation, soil, and erosion potential, and within the protocol, we consider ecological sites and OFEs to be interchangeable. For the RWM protocol, the OFE is the smallest spatial unit used for characterizing upland conditions.

At the largest spatial scale of the RWM protocol, a region is subdivided into 5th- to 6th-order river basins, having sizes ranging from about 500 to 1,000 km². Nested within a river basin are a variable number of lower-order watersheds. The geographic unit delineated for detailed study by the RWM protocol is the second-order watershed. Second-order watersheds are defined herein using Strahler’s (1957) modification of Horton’s (1945) stream ordering system. In this system, a stream with no tributaries is considered first order. A second-order stream begins at the confluence of two first order streams and ends at the confluence with another second or higher order stream. The order a stream segment is given is highly dependent upon the method used to delineate the stream channel network. A stream network and its stream ordering may be identified using topographic maps, digital elevation models, aerial photography, or field mapping. Field mapping provides the most accurate depiction of a complete channel network, although data collection expense usually limits use of this method (Gandolfi and Bischetti, 1997). Aerial photography may provide results nearly as good as field mapping. Stream networks are frequently defined using the "blue lines" on U.S. Geological Survey topographic maps due to the availability of such maps. Unfortunately, topographic map blue lines are cartographic generalizations that portray the stream network using the cartographer’s subjective judgement (Leopold, 1994); they may also underestimate "true" field derived stream order by several or more orders (Leopold et. al., 1964). As we discuss in detail in a following section, we derive flow path networks that approximate the 1:24,000 scale topographic map blue line stream networks but which are derived using objective criteria and a GIS.

The protocol’s sampling strategy employs watershed structure for selecting monitoring site locations. The following sequential steps are taken in determining sampling site locations using this watershed conceptualization:

• The flow network is delineated.
• The river basin of interest is delineated.
• All second-order watersheds within the river basin are delineated.
• A set of second-order monitoring watersheds is selected.
• For each second-order monitoring watershed:

• Hillslopes are delineated.
• OFEs are delineated.
• OFEs are classified in 3 to 5 categories based upon soil, vegetation, and geomorphic characteristics.
• OFEs are randomly selected from each of the classes. These OFEs become the hillslope sample sites.
• Flow path sample sites are located where a hillslope containing an OFE sample site drains into the flow network.

Typically, 5^th- to 6^th-order river basins will have from 50 to several hundred second-order watersheds, with larger basins containing larger numbers of watersheds. Second-order watersheds, defined and delineated as above, will generally have an average size of several square kilometers. Depending upon watershed size and characteristics, each second-order watershed will be comprised of 50 to 100 OFE’s that may be categorized into three to five classes. Even though defined as second-order watersheds, field determination of the flow path network will usually increase actual watershed order by one to three orders.

In concluding this section regarding the watershed-based approach for monitoring site selection, we note that this selection procedure is in dramatic contrast to that used in the U.S. EPA Environmental Monitoring and Assessment Program (EMAP) (e.g., Kapner et al., 1993) for selecting sampling points on a regional and national basis. In the EMAP program, a hexagonal grid sampling design based upon a triangular point network is used for spatially subdividing the landscape, with the largest spatial elements consisting of 635 km² hexagons packed into a honeycomb like structure (Figure 4). A smaller 40 km² hexagon is centered within each larger hexagon. From this data set, a probability sampling of points can be selected by several methods. The regularly spaced grid ensures a uniform spatial dispersion of sample points over spatially distributed ecological resources. This protocol’s sampling strategy differs in that it employs watershed structure for selecting monitoring site locations (Figure 4). Thus, site selection is not made in a completely random manner. Rather, the spatial ordering of the landscape via flow pathways is recognized as linking downstream flow path conditions with upstream hillslope (OFE) conditions. By taking advantage of flow-related spatial correlation within a watershed, the goal of this protocol is to extract more information about flow-linked causality than would be available from a spatially random sampling design.

Modeling Methods

Numerous conceptual and mathematical models have been developed to simulate watershed processes of runoff and erosion. As we stated above, modeling methods are potentially a more accurate means for determining erosion rates (and changes in these rates) than are other methods. In addition, once a model is established for an area, the potential for forecasting consequences of land use management decision or responses to climate change become possible (Burt, 1994). Models may also have utility in defining complex spatial patterns that could not otherwise be easily determined. These patterns may be useful for tasks including the selection of monitoring sites. Two modeling methods are being employed in development of the RWM protocol; however, other modeling methodologies could potentially be used in lieu of these two.

Hillslope erosion is assessed at the OFE scale using WEPP. WEPP is a process-based model having a theoretical framework and measurable model parameters used for determining runoff, sediment detachment, sediment deposition, and sediment yields from farmlands and rangelands (Flanagan et al., 1996). WEPP allows the description of unique combinations of soil types and vegetation as OFEs (Figure 3) to simulate runoff and sediment movement over non-uniform hillslopes. Process-based models like WEPP have utility in that 1) their output is based upon simulation of actual physical processes, and 2) model input parameters are measurable real-world parameters. WEPP vegetation input parameters, for example, can be modified to reflect actual OFE vegetation changes as determined from on-ground or other monitoring methods. Changes in erosion rates can then be calculated within WEPP. Thus, WEPP has the potential to simulate the effects of vegetation change on sediment production (Wilcox et al., 1992).

The other modeling concept being employed is a TOPMODEL-based delineation and characterization of the flow path network. TOPMODEL was conceptualized by Beven and Kirkby (1979) as a means for determining the spatial distribution of subsurface moisture within a watershed. Field observations by many scientists indicate that higher soil moisture or areas of surface saturation tend to occur with increasing specific catchment area, decreasing slope, thin soils, and an increased rate of watershed recharge (precipitation input minus losses). Beven and Kirkby (1979) implemented this idea into a simple relative wetness index. Hillslope areas are relatively dry but points farther downstream become wetter until saturation and overland flow occur. To delineate flow pathways, we are employing a wetness index using TOPMODEL concepts that has been devised by Pack and others (1998):

W = min [( R a / T sin(theta) ), 1]                  Eq. 1

where:

W = is the wetness index with an upper bound of 1 representing saturation,
R = steady-state recharge rate over the watershed (m/hr),
a = specific catchment area (m) obtained by dividing drainage area by flow path width,
T = soil transmissivity (m²/hr), and
sin(theta) = land slope


Figure 5 illustrates an example of the implementation of this index using DEM data and GIS methods on a second order watershed. We are currently in the process of extending the TOPMODEL wetness concept into a form for determining flow path sediment storage capacity.

IMPLEMENTING THE PROTOCOL WITHIN A GIS FRAMEWORK

Because of the spatial nature of watershed data, a geographic information system (GIS) provides a logical tool for mapping and presenting watershed monitoring information. Frequently, people tend to focus only upon GIS’s automated mapping capabilities. The real benefits of GIS come from its data storage, analysis, and manipulation capabilities (Somers, 2000). We recognize that GIS is a tool capable of substantially more than just map making. A GIS can provide an integrated framework for efficiently conducting the many tasks required for implementing a monitoring protocol. Within the RWM protocol, a GIS is utilized for the following tasks:
 

• Establishing and organizing a spatial working environment
• Locating sampling sites
• Determining parameters of watershed elements
• Storing, analyzing, and presenting information
• Modeling


Establishing and Organizing a Spatial Working Environment

An inherent advantage of employing a GIS in a monitoring protocol is that it places an up-front emphasis upon the importance of establishing a spatial context for the monitoring program’s data collection effort. Necessarily, a geographic reference system must be selected that can provide the geographic basis for locating all monitoring sites. In the field, a global positioning system (GPS) is used to locate a point within the geographic reference system. Within the GIS, data sets are geographically stored and mapped using the reference system. Geographic reference systems include latitude-longitude and various mapping projections such as the Universal Transverse Mercator (UTM) and State Plane coordinate systems. GIS software often allows conversion between coordinate systems, even to the point of allowing on-the-fly conversion for simultaneously displaying data sets stored in different coordinate systems. In our analyses, we utilize the UTM coordinate system as our geographic reference system. Use of the UTM system is advantageous in that it provides a uniform metric grid backdrop for spatially locating monitoring sites. More importantly, three fundamental data sets used in the monitoring protocol are available in digital format that is georeferenced by UTM coordinates: 1) digital raster graphics (DRG), 2) digital orthophoto quarter-quads (DOQQ), and 3) digital elevation models (DEM).

DRGs are scanned versions of U.S. Geological Survey (USGS) topographic maps. As with normal topographic maps, DRGs provide fundamental information regarding the physical and cultural features of an area, albeit generalized and simplified by cartographic interpretation. For the scale of problems being evaluated by the monitoring protocol, DRGs derived from 1:24,000 scale topographic maps (7½ minute quads) are most appropriate. A DRG based upon a 7½ minute quad has a ground pixel resolution of about 2.4 m. DRGs maintain the positional accuracy of their source maps, which is about 12 m for 90 percent of points, based upon National Map Accuracy Standard (NMAS) production requirements for these maps. The utility of DRGs for the monitoring program is the same as that of the paper maps they replace: they provide a readily understandable visual display of an area.

DOQQs are digital files of scanned (rasterized), rectified, and georeferenced aerial photographs with a resolution (1 m) that allows landscape details to approximately 2 m to be assessed. DOQQs are created from National Aerial Photography Program (NAPP) aerial imagery that is being retaken nationally on an approximately 5-year cycle. NAPP photographs are acquired at 6,100 m (20,000 feet) above average terrain elevation and may be either black-and-white or color-infrared images. Standard DOQQs derived from NAPP imagery cover one-quarter of a 7 ½ minute quadrangle and have similar accuracy standards to DRGs as described above. Complete DOQQ coverage of the U.S. is expected by 2004, with a 10-year update thereafter for most areas and a 5-year update in urban areas. The anticipated updating of DOQQs makes them an essential element in the monitoring protocol.

DEMs are grids of elevation points used for determining land slopes, aspects, elevations, and more complex topographic features such as watersheds. Most typically, DEMs are generated from digitized contours, but they may also be derived by surveying or photogrammetric methods. In the U.S., DEMs are produced by the USGS in 7½-minute by 7½-minute blocks providing the same coverage areas as 7½ minute DRGs and topographic maps. These DEMs have a 30 m by 30 m data point spacing with horizontal position located using UTM coordinates and vertical position presented as elevation above sea level. Data quality of DEMs is dependent upon a number of factors including original data source quality and method of extracting the DEM from the source. Within the protocol, the DEM data set is used for delineating watershed elements and for deriving several landscape parameters.

Locating Sampling Sites

A GIS allows ready implementation of random, systematic, or stratified sample site selection procedures. For the RWM protocol, we implement several levels of randomization and stratification in site selection. All site selection is undertaken in the office prior to any field data collection. Maps and air photos with plotted sample sites and lists of sample site georeference coordinates are produced to help locate sampling sites in the field. As described earlier, the monitoring protocol is based upon a watershed approach for geographically selecting sample sites, with sample sites located within a river basin, watershed, subwatershed, hillslope, OFE, and/or the drainage network.

For the protocol, we use a GIS-based approach to extract drainage networks and create watersheds from DEMs. Methods for extracting drainage features from DEMs have been devised by Jenson and Domingue (1988), Tarboton et al. (1991), and others and are available in many GIS software packages. Using this GIS-based approach, all second-order watersheds within a river basin are identified. Depending upon specific monitoring program goals, one or more of the second order watersheds can be selected either randomly or for meeting specific project requirements. Each selected second-order watershed is subdivided into its constituent elements -- a drainage network, subwatersheds, hillslopes, and OFEs. The drainage network and subwatersheds are delineated using automated GIS procedures similar to the procedures for delineation of second-order watersheds. At present, automated procedures have not been developed to delineate hillslopes or OFEs. However, "heads-up" delineation of these features is simplified by interactive use of the GIS environment. Figure 6 illustrates the delineation of a second-order watershed into OFEs. Depending upon watershed size and character, there will typically be 50 or more OFEs per second-order watershed.

Two categories of information are extracted from each watershed: that representing hillslope conditions and that representing conditions along the flow path network. A subset of OFEs is selected for assessing and monitoring hillslope conditions within the watershed. OFE’s are first grouped into one of several classes in each watershed, with classes derived using cluster analysis statistical procedures. These classes are created based upon similarity of OFE slope, vegetation, and soils. For each OFE class, one or more OFE’s are randomly selected, with economics and program goals determining the number of required OFEs. Point or site information required from within OFEs is randomly selected using a spatial random grid point generator available with the GIS (Figure 6). These points can be printed out overlaying the DRG map or DOQQ aerial photograph so they may be located in the field. Additionally, a table of geographic coordinates can be generated which can be used in conjunction with a GPS to locate these points in the field.

The second type of information required by the protocol is that collected along the flow path network within a watershed. Data are collected at points or reaches along the network at locations where the sampled OFE’s drain into the channel network and at the watershed outlet. This stationing assures that sites selected to represent network conditions have the best potential for linking to hillslope conditions. Additional random points can be selected if necessary for other requirements. As with the upland points, these points can be printed onto maps aerial photograph and tabulated into GPS coordinate lists.

Determining Parameters of Watershed Elements

Many required watershed parameters can be derived from the three geographic data sets used in establishing and organizing a monitoring program’s spatial working environment. The most important data set is the DEM. Various GIS procedures can use the DEM to 1) spatially define the watershed, subwatersheds, hillslopes, OFEs, and the flow path network and 2) derive parameters representing various aspects of these elements. For each grid cell (pixel), some of the parameters that can be derived include:
 

• Elevation
• Slope
• Aspect
• Watershed area
• Maximum flow path length to cell
• Total of flow path lengths to cell


Depending upon the requirements of specific analyses, these parameters may be aggregated up from the pixel level to average values (or distributions) for the various watershed elements. Although many of these data could be derived using non-GIS methods, GIS allows more rapid and perhaps more accurate determination of these parameters.

Storing, Analyzing, and Presenting Information

In any long-term monitoring program, handling and maintaining the tremendous volume of data that are collected can be overwhelming (Pickett, 1991). Because a GIS is a database, its database functionality can be used for organizing and storing data. The special advantage of a GIS database is that information may be selected and accessed using spatial criteria. If a GIS utilizes a standard database format for storage of non-spatial data, then these data may be accessed and manipulated with non-GIS software, although the capability for selecting information using spatial criteria may not be available.

Three general categories of information are associated with the RWM protocol. These categories are:
 

• Spatial/Temporal Information. Data of this type has a spatial location within the watershed and varies over time. An example of this type of data set is vegetation cover information collected along established transects (Figure 7).
• Non-Spatial/Temporal Information. Some types of information vary temporally but will not have a spatial component to them. For example, annual precipitation depth might be assumed to be spatially uniform over a watershed but be variable year to year. This information would not be associated with any specific watershed location.
• Spatial/Non-Temporal Information. Many of the landscape parameters derived using GIS methods (described in the previous section) are time invariant. Average OFE slope, for example, is not expected to vary over the short-term time scales of human observation.

 

 
 
 
 
 

A GIS may be capable of storing all three of these information types, though GISs are generally designed to be most efficient at storing spatial/non-temporal information. The key requirement to storing temporal information is the inclusion of a time/date attribute field into a tabular data set. As with any database software, development of user-friendly interfaces and subprograms will be necessary for most efficiently handling and maintaining the data set. The RWM protocol has no specific requirements for a particular GIS or database implementation. Once stored in the tabular database format of a GIS, watershed data sets may be accessed and manipulated for many types of analysis. Either through direct access to GIS database files or using exported data files, spreadsheets and statistical analysis software programs can be used for many types of analyses or graphical display (Figure 7).

Modeling

Perhaps the most underutilized use of GIS is for modeling. At present, there are two endpoint means for linking GIS and modeling, which are referred to as loose coupling and tight coupling.
 

• Loosely coupled models. Loosely coupled models use existing models with no changes made to model structure, input, or output. However, the GIS database and its associated programs are written to extract information and write it out into the input file structure(s) required of the model being used. After a model simulation is run, its output may be input into the GIS for spatial analysis of results.
• Tightly coupled models. In tightly-coupled models, the science behind the model is completely incorporated into the GIS. No additional software modeling programs are required, for the model is completely contained in the GIS.

 

 
 
 
 
 

To date, most efforts to link GIS and modeling have followed the loose coupling approach, as this approach does not require rewriting existing models. We are taking this approach to link the WEPP model to the GIS. For flow path modeling, we are in the process of incorporating a TOPMODEL concepts and sediment redistribution directly into a GIS.

SUMMARY

A rangeland watershed monitoring (RWM) protocol has been developed to allow holistic monitoring of watershed conditions. The protocol uses a top down or strategic approach to guide selection of the most appropriate monitoring methods and sites. The protocol’s strategy is based upon understanding the physical processes of erosion and sedimentation (hydrologic stability) and use of models to assess the spatial and temporal variability of these processes within a watershed. Because of the spatial nature of watershed data, a geographic information system is used to provide an integrated framework for efficiently conducting the many tasks required for implementing a monitoring protocol. These tasks include establishing and organizing a spatial working environment, locating sampling sites, determining parameters of watershed elements, storing, analyzing, and presenting information, and modeling. Modeling approaches utilizing both loose and tight coupling are being implemented in the protocol.

ACKNOWLEDGEMENTS

This research is being supported by a contract (Contract No. CR826454-01-0) with the U.S. Environmental Protection Agency. We thank Roger Dean, USEPA Region 8, Denver, and Anne Neal, USEPA Environmental Sciences Division, Las Vegas, for their support. The viewpoints presented in this paper do not necessarily represent those of the U.S. Environmental Protection Agency.

REFERENCES

Allan, J. D., and Johnson, L. B., 1997, Catchment-scale analysis of aquatic ecosystems: Freshwater Biology, v. 37, p. 107-111.

Berger, A. R., 1996, Introduction to geoindicator checklist, in Berger, A. R., and Iams, W. J., eds., Geoindicators; Assessing Rapid Environmental Change in Earth Systems: A. A. Balkema, Rotterdam, p. 383-394.

Berger, A. R., and Iams, W. J., 1996, Geoindicators; assessing rapid environmental change in earth systems: A. A. Balkema, Rotterdam, 466 p.

Beven, K., and Kirkby, M. J., 1979, A physically based, variable contributing area model of basin hydrology: Hydrological Science Bulletin, v. 24, p. 43-69.

Burt, T. P., 1994, Long-term study of the natural environment - perceptive science or mindless monitoring Progress in Physical Geography, v. 18, p. 475-496.

Burt, T. P., Heathwaite, A. L., and Trudgill, S. T., 1993, Catchment sensitivity to land use controls, in Thomas, D. S. G., and Allison, R. J., eds., Landscape Sensitivity: Chichester, John Wiley and Sons, p. 229-240.

Frissell, C. A., Liss, W. J., Warren, C. E., and Hurley, M. D., 1986, A hierarchical framework for stream habitat classification: viewing streams in a watershed context: Environmental Management, v. 10, p. 199-214.

Gandolfi, C., and Bischetti, G. B., 1997, Influence of the drainage network identification method on geomorphological properties and hydrological response: Hydrological Processes, v. 11, p. 353-375.

Gregory, S. V., Swanson, F. J., McKee, W. A., and Cummins, K. W., 1991, An ecosystem perspective of riparian zones: BioScience, v. 41, p. 540-551.

Horton, R. E., 1945, Erosional development of streams and their drainage basins, hydrophysical approach to quantitative morphology: Geological Society of America Bulletin, v. 56, p. 275-370.

Jensen, S. K., and Domingue, J. O., 1988, Extracting topographic structure from digital elevation data for geographic information system analysis: Photogrametric Engineering and Remote Sensing, v. 54, p. 1593-1600.

Johnson, L. B., and Gage, S. H., 1997, Landscape approaches to the analysis of aquatic ecosystems: Freshwater Biology, v. 37, p. 113-132.

Joyce, L., 1993, The life cycle of the range condition concept: Journal of Range Management, v. 46, p. 132-138.

Kapner, W. G., Kuehi, R.O., Breckenridge, R. P., Baker, J. R., O’Leary, D. O., Lancaster, J. M., Leonard, S. G., Mouat, D. A., Reinsch, T. G., Weltz, M. A., Sutton, D. W., Tidwell, R. L., Neale, A. C., Byers, G. E., and Slagel, R. L., Environmental monitoring and assessment program; Arid ecosystems 1993 implementation plan Colorado Plateau plot design pilot study: U.S. EPA Environmental Systems Monitoring Laboratory, Las Vegas Nevada, EPA/620/R-93/016.

Laflen, J. M., Elliot, W. J., Flanagan, D. C., Meyer, C. R., and Nearing, M. A., 1996, WEPP-predicting water erosion using a process based model: Journal of Soil and Water Conservation, v. 52, p. 96-102.

Landers, D., 1997, Riparian restoration: current status and the reach to the future: Restoration Ecology, v. 5, p. 113-121.

Leopold, L. B., Wolman, M. G., and Miller, J. P., 1964, Fluvial processes in Geomorphology: W. H. Freeman and Company, San Francisco, 522 p.

Leopold, L. B., 1994, A view of the river: Berkeley, CA, United States, University of California at Berkeley, 298 p.

Pack, R. T., Tarboton, D. G., and Goodwin, C. N., 1998, The SINMAP approach to terrain stability mapping: Paper Submitted to 8th Congress of the International Association of Engineering Geology, Vancouver, British Columbia, Canada 21-25 September 1998.

Pickett, S. T. A., 1991, Long-term studies: past experience and recommendations for the future, in Risser, P. G., ed., Long-term ecological research - an international perspective: Chichester, John Wiley and Sons, p. 71-88.

Society for Range Management Glossary Revision Special Committee, 1989, A glossary of terms used in range management: a definition of terms commonly used in range management. Society for Range Management, Denver, Colorado, 20 p.

Society for Range Management Task Group on Unity in Concepts and Terminology, 1995, New concepts for assessment of rangeland condition: Journal of Range Management, v. 48, p. 271-282.

Somers, R., 2000, GIS stalled; Getting past the mapping phase: GeoInfo Systems, v 10., no. 3, p. 20-22.

Strahler, A. N., 1957, Quantitative analysis of watershed geomorphology: American Geophysical Union Transactions, v. 38, p. 913-920.

Tarboton, D. G., Bras, R. L., and Rodriguez-Iturbe, I., 1991, On the extraction of channel networks from digital elevation data: Hydrological Processes, v. 5, p. 81-100.

U. S. Environmental Protection Agency, 1999, Protocol for developing sediment TMDLs: U.S. EPA Office of Water, Washington D.C., EPA 841-B-99-004.

Walling, D. E., 1983, The sediment delivery problem: Journal of Hydrology, v. 65, p. 209-237.

Webster’s, 1983, Webster's ninth new collegiate dictionary. Merriam-Webster, Springfield, Mass., 1563 p.

Wilcox, B. P., Sbaa, M., Blackburn, W. H., and Milligan, J. H., 1992, Runoff prediction from sagebrush rangelands using water erosion prediction project (WEPP) technology: Journal of Range Management, v. 45, p. 470-474.

---