The Wildlife Society
Bethesda, Maryland
1994
By
Gregory T. Koeln, Lewis M. Cowardin, and Laurence L. Strong
Theodore A. Bookhout, Editor
U.S. Fish and Wildlife Service
Ohio Cooperative Fish and Wildlife Research Unit
The Ohio State University
Columbus, Ohio
Koeln, Gregory T., Lewis M. Cowardin, and Laurence L. Strong. 1994. Geographic Information Systems. Pages 540-566 in T. A. Bookhout, ed. Research and management techniques for wildlife. Wildl. Soc., Bethesda, MD. 740 pp.This resource should be cited as:
Koeln, Gregory T., Lewis M. Cowardin, and Laurence L. Strong. 1994. Geographic Information Systems. Pages 540-566 in T. A. Bookhout, ed. Research and management techniques for wildlife. Wildl. Soc., Bethesda, MD. 740 pp. Jamestown ND: Northern Prairie Wildlife Research Center Online. http://www.npwrc.usgs.gov/resource/habitat/research/index.htm (Version 15JUN99).
A study of 61 fish and wildlife agencies in the U.S. revealed that GIS use is widespread and growing rapidly (Rodcay 1991). Thirty-nine percent of the agencies used GISs regularly in their programs, 30.2% used GISs rarely, and 30.5% were not using GISs. Seventy-three percent of those agencies not using GISs plan to use them in the near future. For those agencies using GISs, habitat mapping was the most common use. Other applications included land use inventory, vegetation mapping, species distribution estimation, preferred habitat definition, and land development planning.
Wildlife managers have used GISs for monitoring wetlands for waterfowl habitat (Barnard et al. 1981, Koeln et al. 1988), for mapping Florida scrub jay habitat (Breininger et al. 1991), for evaluating grizzly bear (Craighead et al. 1986, Agee et al. 1989), lesser prairie-chicken (Cannon et al. 1982), and elk habitat (Leckenby et al. 1985), for preserving biological diversity (Davis et al. 1990), for monitoring wood stork foraging habitat (Hodgson et al. 1988), for analyzing radiotelemetry data (Koeln and Cook 1984, Young et al. 1987), for characterizing the spatial structure of habitats (Heinem and Mead 1984, Ripple et al. 1991), for characterizing ecotones (Johnston and Bonde 1989), for predicting wildlife densities (Palmerim 1988, Broschart et al. 1989), for modeling the spatial distribution of species (Palmerim 1987, Walker 1990), for designing reserve systems (Saxon and Dudzinski 1984, Murphy and Noon 1991), for examining the cumulative impacts of habitat loss (Johnston et al. 1988, Gosselink and Lee 1989), for quantifying beaver pond creation (Johnston and Naiman 1990a,b), and in many other ways (de Steiguer and Giles 1981, Steenhof 1982, Lyon 1983, Mayer 1984, Peterson and Matney 1986, Ormsby and Lunetta 1987, Scepan et al. 1987, Stenback et al, 1987, Miller and Conroy 1990, Shaw and Atkinson 1990).
The intent of this chapter is to provide wildlife managers and wildlife management students an overview of the technology of GISs. Many universities offer graduate and undergraduate courses in GISs. Various books, journals, and other publications are listed in this chapter for those wanting to further explore the use of GISs in wildlife management and research.
![]() |
Fig. 1. -- Geographically referenced data (spatial data) are any data that can be represented as a point (A), line (B), or area (C) (Aronoff 1989). |
Dueker and Kjerne (1989:8-9) defined a GIS as "a system of hardware, software, data, people, organizations, and institutional arrangements for collecting, storing, analyzing, and disseminating information about areas of the earth." GISs differ from general database management systems (DBMS). DBMS do not handle adequately the spatial data requirements of GISs. Spatial data have two components: a geographic reference, and an attribute. A road in a GIS may be represented by a line. The geographic reference of that line would be the coordinates describing its location. The location of the road (line) can be recorded with Universal Transverse Mercator (UTM) coordinates, state plane coordinates, latitude and longitude, or other (including arbitrary) coordinate systems. The attribute component of the spatial data describing the road may include the type of road (gravel), the route number (Interstate 44), or other attributes of the road, such as the average number of cars that use the road in a year. DBMSs often can manage well the attribute component of spatial data, but poorly manage the geographic reference.
A GIS frequently is described in terms of hardware and software, but it should be thought of as a general system with inputs, processes, outputs, and a context. The input component is the most expensive. Capturing, registering, interpreting, and converting spatial data frequently comprise 60-90% of the expense of operating a GIS. Processes of the GIS include efficient and effective means of storing and retrieving both the attribute and geographic reference of spatial data and creating new information derived from spatial data stored in the system. This new information includes such things as the distance to the nearest stream or the size of continuous blocks of forestland.
The outputs of the GIS include hard-copy maps, graphic displays on color or monochrome monitors, and tabular information. Technological advances are vastly improving output capabilities of GISs.
The context of the system includes the organizational and institutional components of the GIS, e.g., staff, funds, and administrative support. Administering the organizational and institutional components of the GIS is frequently much more difficult than selecting, learning, and using the GIS software and hardware (Lauer et al. 1991).
Many types of systems frequently are confused with GISs (Korte 1991). Computer-aided mapping (CAM) systems automate the design, creation, and maintenance of maps. These systems usually are enhancements to computer-aided drafting (CAD) software and provide powerful tools for making and updating maps. CAM/CAD systems handle well the geographic reference of spatial data, but often poorly handle the attribute components of spatial data. The analytical capabilities of CAM/CAD systems are not the same as those of a GIS. GISs usually do not perform as well as CAM/CAD for purely cartographic application, but the cartographic capabilities of modem GISs are improving. In addition, many GISs now can use and manipulate data created from CAM/CAD systems.
Automated mapping and facilities management (AM/FM) is the use of CAM or GISs for public works and utility information (Ducker and Kjerne 1989, Vonderohe et al. 1991). Information on telephone lines, electrical lines, water lines, sewers, and other utilities often are managed with AM/FM systems. The AM/FM allows linking attribute data to spatial data, but, like CAM, spatial relationships are not defined and spatial analysis is slow and cumbersome at best.
Cadastral systems are used to manage quantity, value, and ownership of real estate. Multipurpose cadastral systems are parcel-based land information systems (Dueker and Kjerne 1989, Vonderohe et al. 1991). These parcels could be sections of the public land survey, counties, or wildlife management units, for example.
Often, the application of GISs is termed Land Information Systems (LISs). Dueker and Kjerne (1989) described LISs and GISs as containing data primarily describing land records. Vonderohe et al. (1991) described the process of maintaining records on the land as an LIS, which does not necessarily require the use of computers. But once this process is automated by computer, it is a GIS.
The first question is one of the simplest functions of a GIS. The location can be described in many ways and can be defined (1) as a point, line, or area (polygon), (2) by place name (i.e., street address, city, county, or wildlife management area) or post or ZIP code, or (3) by geographic coordinates, such as UTM, state plane, or latitude and longitude. A wildlife manager may use this capability to describe the habitat occurring in a wildlife management area, a study area, or a county. A wildlife researcher, using radiotelemetry techniques, may use a GIS to determine various kinds of information about sites used by the species being studied. For each radiotelemetry location, the habitat type occurring at the location can be ascertained. Many other habitat parameters also can be obtained, such as the distance to the nearest road, stream, or forest edge; the size of a continuous block of habitat being used; the elevation, slope, and aspect of the location; or the area of various habitat types located within a determined distance from the location of the studied animal.
The second question that a GIS can answer is the converse of the first question. Instead of asking what occurs at a particular site, this question asks where do certain situations or conditions occur. The wildlife managers may want to know the location of all red-cockaded woodpecker nesting colonies occurring on private lands, which county sells the most duck stamps, or which county has the most area enrolled in the U.S. Department of Agriculture (USDA) Conservation Reserve Programs (CRP).
The third question addresses changes in time. A waterfowl manager may want to know which wetlands are typically dry during the summer and thus have little value for waterfowl brood habitat. A big-game biologist may want to know which counties have had the greatest reduction in forest cover. The GIS uses two or more inventories acquired at different times to address these types of questions.
The fourth question addresses the social, economic, environmental, or combined impact of an existing change in land use. Ascertaining the benefits of CRP to waterfowl populations requires information on the location of the CRP enrollment areas, the land use that previously occurred in the area, the success of establishment of permanent cover on the CRP areas, and the availability of wetlands within and near the CRP sites.
Walker and Miller (1990) believed that the fifth question may be the most important or highest use of a GIS. Using a GIS, a wildlife manager can answer the "What if?" questions. What will be the impact to waterfowl populations if temporary or seasonal wetlands no longer are protected by federal regulations? Which wetlands are most vulnerable to wetland drainage given that drainage rates are a function of distance to nearest road, wetland type, size of continuous wetland, and surrounding land use?
A GIS is not a computerized system for making maps, even though maps are an important output product of GIS and many GISs now have excellent mapping capabilities. A GIS is not a tool for storing maps or pictures (although many modem GISs can store on one CD-ROM disk images of all topographic maps or aerial photography that are found in stacks of map cases or in rolls in obscure corners of offices). Although maps are an essential source of information for a GIS, the information maintained in the database of the GIS is the central concept, not the maps.
A GIS is an approximate model of the real world that uses computer systems to abstract three key pieces of information about features of the land required for management decisions. For every land feature, the GIS must know (1) what it is, (2) where it is, and (3) how it relates to other features (Walker and Miller 1990). GISs provide a mechanism for maintaining information about the land. Gathering information is the first, and most important, step in developing a GIS, followed closely by updating and maintaining information as features of the land change.
The spatial components of geographic data can be represented by three data types: points, lines, and areas (Fig. 1). The spatial data types are referenced to a location by a standard system of coordinates, such as UTM (see Box 1), or by a local coordinate system. Local coordinate systems can be created simply by assigning the southwest corner of a map the X and Y coordinates of 0.0 and 0.0 and measuring the horizontal (X) and vertical (Y) distances from the southwest corner of the map to the feature. The coordinates for the location of a feature, such as a bird's nest, could be ascertained by marking the location of the nest on a map and assigning the X coordinate as the number of centimeters from the west edge of the map to the location marked on the map, and assigning the Y coordinate as the number of centimeters from the south edge of the map to the location marked on the map. Of course, using standard coordinates, such as UTM, state plane, or latitude and longitude, ensures that anyone using the data in the future will know the precise location of the bird's nest. In addition to the coordinates, a label describing which bird's nest is located at the given coordinates will be stored with the coordinates. The attribute record for the bird's nest will be referenced by the label and might include various attributes for the nest, including species of nesting bird, height of the nest, number of eggs laid, and number of eggs hatched. The label links the spatial data with the appropriate attribute record.
Spatial data are represented in GISs in two very different ways. Figure 2 shows the two different ways that a stream could be represented in a GIS. Spatial data can be represented as either rasters or vectors. In raster format, a grid is used to represent the study area. The location of features in the study area is depicted by the values in the cells overlaying the feature. Vector data represent geographic features by coordinates of points, lines, and polygons. Points represent small features such as wells, towers, or nest locations. Linear features such as roads and streams are represented by lines. Areas such as cities, forests, wetlands, and soil units are represented by polygons. Polygons are bounded on all sides by a series of straightline segments.
![]() |
Fig. 2. -- A stream can be represented in a GIS either by a raster (a) or vector (b) format. |
Raster data are stored in the computer as a matrix. The cells are referenced by lines and elements (Fig. 3). In the simplest form, each line is a computer record. Each record will contain the values for all elements in the line. Any cell not containing a feature would have the value of "0". In the simplest raster system, the value stored for each cell is the attribute component of the geographic data. In Fig. 4, cells with value of "1" are forests, cells with value of "2" are croplands, and cells with value of "3" are rangelands.
In more sophisticated raster systems, the cell value is a label that will link to records as an attribute file. In the above example, cells labeled as "1" could have many attributes, such as species composition, age of forest stand, and estimated volume of marketable timber.
![]() |
Fig. 4. -- Land cover as represented in a simple raster system. Cells with a "1" are forests, cells with a "2" are croplands, and cells with a "3" are rangelands. |
Because the raster system is strictly a two-dimensional matrix, various types of geographical data are stored as different layers or overlays in the GIS (Fig. 5). One layer may contain land use/land cover, another layer may contain wetland data, and another layer may contain information on the transportation system.
![]() |
Fig. 5. -- Various types of geographic data may be stored as different layers or overlays in a GIS. |
The user of a raster system must determine the size of the cells to be used. This size is referred to as spatial resolution (see Box 2 for various meanings for the term resolution). The cell size can vary tremendously depending upon the size of the study area and the objectives for the GIS. Cell sizes as large as 20 ha for state or regional planning may be adequate. For a wildlife management area, a cell size of 0.05 ha or smaller might be required, depending upon the application of the GIS and the size of the wildlife management area. Storage requirements increase drastically as the cell size is reduced. Reducing the cell size by one-half will increase the data storage requirements by a factor of four. Conversely, as cell size increases, the precision of the representation of the land feature is reduced. Choosing the appropriate cell size for a particular GIS application is a compromise between cost of data storage and computer time and reliability of the representation of the land feature.
Vector data provide for high precision in representing the location of features. Aronoff (1989) described how vector data can be used to define the location of a point, a line, and an area. A point is represented by a simple pair of coordinates. The line is represented by an ordered list of pairs of coordinates. The area is represented as a polygon with ordered pairs of coordinates that close the polygon (the first and last pair being the same).
The coordinates can be any arbitrary units but usually are stored as UTM, state plane, or latitude and longitude coordinates. The first vector system used simple techniques to store the X and Y coordinates for polygons. In this simple system the coordinates for the common boundary between two areas were stored twice, once for the first area and again for the adjacent area. These duplicate storage techniques simplified computations and plotting but wasted storage space and, more importantly, provided no information as to adjacency or connectivity of geographic features (topology). Most vector systems now use topological models (Aronoff 1989) for representing the location of areas.
In topological models (Fig. 6), a polygon is defined by a series of arcs. Arcs begin and end at nodes, which occur wherever two or more arcs meet. Each arc is defined by a series of coordinates, starting with the coordinates for the beginning node and ending with the coordinates for the ending node. Topological relationships are stored in three tables. The polygon topology table describes the arcs that bound each polygon, the node topology table describes the arcs that end at each of the nodes, and the arc topology table describes which end points (nodes) occur on each arc and which polygons are to the left and right of each arc. These three topology tables provide the tools required to efficiently determine the positional relationships of one feature to other features. A coordinate table defining the coordinates for each arc also is used in topological models. In addition to these topological databases, the attributes for the features are stored in an attribute database.
Early GISs were either raster or vector systems. Table 1 lists various advantages and disadvantages of vector and raster data systems (Burrough 1986, Aronoff 1989). Both approaches are equally valid ways of representing spatial data. The advantages and disadvantages of raster and vector systems have been heavily debated. Most modern GISs handle both raster and vector data but usually are designed primarily for one data type. The complete integration of raster and vector data capabilities will be common in GISs of the future (Faust et al. 1991). These new GISs will quickly and efficiently convert among rasters, vectors, and other data structures most appropriate for the application being performed (McKeown 1987, Ripple and Wang 1989, Piwowar and LeDrew 1990). GISs must function equally well with both raster and vector data.
Table 1. Comparison advantages and disadvantages of vector and raster methods as revised from Burrough (1986) and Aronoff (1989). |
|
Obtaining the data for a GIS is the major bottleneck in implementing a GIS (Aronoff 1989). The creation of an accurate and well-documented database is essential. Information generated from the GIS and resulting decisions made with that information can be accurate only if the initial data are accurate. Accuracy and reliability of all data layers should be documented. Documentation must include such information as date the information was collected, positional accuracy, classification accuracy, completeness, and procedures used to collect and encode the data. The accuracy of spatial databases is a complex issue and was the subject of a recent book (Goodchild and Gopal 1989).
The data to be entered into the GIS include the spatial data (location of the features) and the attribute information (data describing the features). Some data can be captured more readily in vector formats, whereas other data sets are more efficiently extracted from sources by using raster processing techniques.
Because of the expense of data for GISs, all data requirements must be documented before a GIS is initiated. Fortunately, the data required for a wildlife application may be the same data required for a land use planner, soils scientist, geographer, geologist, hydrologist, forester, or direct-marketing expert. Consequently, there is a growing source of existing data in digital format, and sharing or purchasing existing data is much cheaper than digitizing from existing maps or photos or extracting information from satellite data. It is critically important that all available sources of digital data are reviewed and evaluated before new data are acquired.
Various techniques are used to enter data into a GIS. Data can be entered manually from the computer keyboard or a digitizing table. Manual digitizing can be slow and expensive, but at times it may be the most efficient and accurate means of data entry. Scanning or scan digitizing is more automated than manual digitizing; Recent advances in scanning hardware and improvements in software for extracting information from scanned images are making scan digitizing a more attractive option to manual digitizing. For large areas, existing satellite technology and modem digital image-processing techniques can be a cost-effective means of capturing data for a GIS. The most effective and efficient method of capturing data for a GIS is to purchase existing digital data sources. Many federal agencies and a growing number of state and local agencies have digital data available. Many of these agencies are willing to share their data or will provide them at a minimum costs.
Often keyboard data entry is used in various digitizing techniques to enter attribute data for a specific feature. Location or the geographic component of features at times can be entered efficiently from the keyboard. This is particularly true for infrequent and widely distributed point data such as the location of cave entrances, nests, or animals that are radio-tracked. The location of points in the field now can be obtained with global positioning systems, termed GPSs (Fig. 7). Hand-held GPSs now can be obtained for <$3,000; they provide locational information in latitude and longitude, UTM, or other coordinates when used in the field. These field locations can be entered manually into the GIS with the computer keyboard, or the coordinates can be stored in the GPS and later downloaded to the GIS. For further information on GPSs, see Box 3.
![]() |
Fig. 7 -- A hand-held global positioning system can be used for obtaining accurate locations for features on the ground (photo provided by Trimble Navigation, Sunnyvale, Calif.). |
In manual-digitizing techniques, a map or aerial photograph is placed on a digitizing table (Fig. 8) and a pointing device (called a cursor, puck, or mouse) is used to record coordinates of features to be extracted from the map. The digitizing table electronically encodes the position of the cursor. Tracing the map features with the cursor can be time consuming and error prone.
![]() |
Fig. 8 -- A digitizing table can be used to record coordinates of features shown on maps or aerial photographs (photo by Altek Corp., Silver Spring, Md.). |
The attribute information about the feature also must be recorded. This frequently is done by labeling each feature with a unique number and building a list of attributes for each uniquely labeled feature. The efficiency of manual digitizing depends on the quality of the digitizing software, the skill of the operator, and the complexity of the map to be digitized. Editing the digitized data and assigning the feature labels or other attributes of the feature may take more time than initially digitizing the map.
Small digitizing tablets (0.3 X 0.3 m) can be purchased for <$100. Large digitizing tables (1.3 X 2 m) that can hold large maps range in cost from $3,000 to $20,000.
Recent advances in scanning hardware and software have made scanning a feasible alternative to manual digitizing for some applications. Continued advancements in this technology are coming and eventually it may replace manual digitizing.
Three types of scanners are available. Flat-bed scanners have a flat scanning surface on which a map or a photograph is placed. Small flat-bed scanners (20 X 30 cm.) cost <$2,000 and have scanning resolutions of 100-150 dots per centimeter (DPC). The flat-bed scanner that scans a 25 X 25-cm map at 100 DPC will produce a raster data file of 6,250,000 cells (a matrix of 2,500 lines by 2,500 elements). The scanned cells can contain intensity values ranging from 0 for a black object to 255 for a white object (When scanning is done in panchromatic mode with 8-bit data). When scanning is done in color mode, each cell contains the intensity of red, green, and blue light being reflected from the map. These intensities usually are measured in a range from 0 to 255.
Normally when resolutions of >150 DPC are required, or large maps are used, drum scanners are required. The map is mounted on a cylindrical drum, which spins as a detector is moved horizontally across the drum. Black and white intensities are recorded in panchromatic mode, or red, green, and blue intensities are recorded in color mode. The area viewed by the detector is termed the spot size (Aronoff 1989) and can be as small as 20 microns. Scanning a large map at 20 microns will create a large raster file.
For some applications, a video scanner can be used. A video camera is mounted on a copy stand and the map is placed beneath the video camera, which is raised or lowered to include a larger or smaller portion of the map. Video scanning typically produces a raster file with <512 elements and 512 lines. The spatial resolution of the cell depends upon the scale of the map and the distance between the map and the video camera.
Scanning by flat-bed, drum, or video scanners produces raster files. Maps that have been especially prepared for scanning show only the lines between features, and coordinates for these features are extracted readily. Extracted coordinates for features from scanned maps or aerial photographs may be complex and will rely on sophisticated, line-following algorithms or feature classification and extraction algorithms to obtain the desired information from the map or photo. As in manual digitizing, much time will be spent editing scanned maps.
GISs are more than simply a warehouse for map information or storage for maps. However, many GISs can effectively store images of maps and aerial photographs obtained from flat-bed or drum scanners. Features on these images are not identified. These high-resolution images often are stored on CD-ROMs. Any of these images can be retrieved by the GIS and viewed on, a color monitor. Feature information from these scanned images can be extracted with feature classification and extraction algorithms, or information (such as the distance between two points or areas) can be calculated with available software of the GIS. Many GISs in the future will support scanned-image libraries.
Advances in remote sensing and GIS technology have followed the advances in computer capabilities since the late 1960s. In many situations, remote sensing techniques that use satellite data are the only feasible means for collecting data for GIS applications over large regions. Remote sensing can be defined as any technique by which we gather data about an object without directly touching the object. Remotely sensed data for GIS applications are obtained from satellites or aircraft.
The most effective techniques of remote sensing used for GIS applications are those that provide digital data for the study area. These digital raster data sets can be obtained by satellite sensors, by digital sensors mounted in aircraft, or from scanned aerial photographs.
Remote sensing can employ active or passive systems. Satellite systems, such as Landsat and SPOT, use passive sensors, which measure the intensity of natural radiation. Active systems, such as radar and laser systems, transmit energy to the ground, then measure the energy returned from the ground to the sensor. Photographic cameras, video cameras, and multi-spectral sensors in aircraft or satellite are examples of passive systems. Some satellite and aircraft remote sensing systems use active sensors such as radar. Numerous passive and active remote sensing systems mounted on satellites or aircraft are currently available for acquiring data for GIS applications, and many additional remote sensing systems will become available to GIS users in the near future.
LANDSAT
Landsat, the U.S. land remote sensing satellite system, began as an experimental
program conducted by the National Aeronautics and Space Administration (NASA).
Landsat 1, launched on 23 July 1972, was expected to function for about 1
year and finally ceased operating in 1978 after nearly 5 years of continuous
operation. During that time, it returned digital data for some 300,000 images
of the earth's surface. The Landsat system was declared an operational system
in 1993 and turned over to the National Oceanic and Atmospheric Administration
(NOAA), U.S. Department of Commerce. In 1984, the Land Remote Sensing Commercialization
Act (Landsat Act) was established to transfer the commercial operation of
the Landsat program to the private sector. Earth Observation Satellite Company
(EOSAT) was selected as the commercial operator for the Landsat program.
Landsat 1 through 3 satellites had two sensors. The Return-Beam Vidicom (RBV) sensor, which is similar to the television camera, recorded red, green, and infrared energy reflected from the surface of the earth. The Multi-Spectral Scanner (MSS) was the main instrument carried on these satellites and is still operating in Landsat 4 and 5 satellites. The MSS sensor collects data by scanning the earth from west to east with an oscillating mirror. Radiation from four different spectral bands (green, red, and two in the near infrared) is recorded. The radiation is transferred by fiber optics to filters that permit only certain wavelengths of radiation to strike the sensor's detectors. The picture element (pixel) sampled by the MSS is about 79 X 56 m (the size of a football field in the U.S.). Landsat satellites 2 and 3 ceased operating in 1983. Landsat 1, 2, and 3 satellites orbited the earth at 900 km and provided repeat coverage for any location on earth every 18 days.
Landsat 4 and 5 satellites were launched in 1982 and 1984, respectively. Landsat 4 is used sparingly because of an electrical problem that developed shortly after its launch. As of July 1993, Landsat 5 was still operating. Landsat 4 and 5 satellites circle the earth every 98.9 minutes in a near polar orbit of 705 km. Each satellite provides repeat coverage for any area every 16 days, at,the same local time of day. Landsat 4 and 5 satellites weigh nearly 2,000 kg each and carry the MSS sensor and the Thematic Mapper (TM) sensor. The TM sensor has excellent capabilities for meeting the data needs of many GIS applications for large regions. Along each orbital path, the TM and MSS sensors can continually scan a swath 185 km wide. The scanned data are systematically divided into an area termed a "Landsat Scene," which encompasses approximately 185 X 170 km. Each scene covers approximately 3.2 million ha. Users of Landsat data can purchase data from an existing archive maintained by EOSAT or can schedule the collection of data for any site. The images from the TM sensor on Landsats 4 and 5 satellites have significantly better geometric quality than images from sensors on earlier Landsat missions due to engineering enhancements to the spacecraft. This has facilitated geodetic rectifications of the images to the accuracy standards for 1:24,000-scale map products (Welch et al. 1985).
The TM sensor provides significant improvements in spatial, spectral, and radiometric resolution compared to the MSS. The instantaneous field of view (IFOV) of the TM is square and results in a ground-resolution cell and image pixel of approximately 30 m on a side. The TM measures the intensity of reflected radiation in six spectral bands--three in the visible wavelengths, blue (0.45-0.52 µm), green (0.52-0.60 µm), red (0.63-0.69 µm); one in the near infrared (0.76-0.90 µm); and two in the shortwave infrared (1.55-1.75, 2.08-2.35). The TM also measures emitted thermal radiation (10.4-12.5 µm), although the IFOV for this spectral band is 120 m on a side. The greater radiometric resolution is achieved by the analog-to-digital conversion of the electrical signal to 8 bits or 256 gray levels compared to the 127 gray levels of the MSS on the first three Landsat satellites. Figures 9-12 provide examples of the raster data collected by the Landsat TM sensor and types of information that can be extracted for use in GISs. Landsat 6 was launched on 5 October 1993; however, communication with the satellite was not established.
SPOT
The first SPOT (Systeme Pour I'Observation de la Terre) satellite was launched
by France in 1986. The SPOT program was designed to be a long-term, operationally
commercial program, whereas the Landsat program was designed initially as
an experimental system. The SPOT program was established by the French government
in 1981 under the French space agency, CNES. France, several European banks,
and industries from Belgium and Sweden have invested in this commercial entity.
SPOT Images, S.A., which is partly owned by the French government, operates
the SPOT system. SPOT Image Corporation was formed to market SPOT data in
the U.S., and Radarsat, Inc., markets the data in Canada.
SPOT-1 carries two identical, high-resolution visible (HRV), pushbroom scanners. Each scanner can operate in one of two modes. In the panchromatic mode, 10-m resolution data can be obtained. This single band records visible energy ranging from 0.51 µm to 0.73 µm. In the multi-spectral mode, three bands are recorded at 20-m spatial resolution: green (0.5 µm-0.59 µm), red (0.61 µm-0.73 µm), and near infrared (0.79 µm-0.89 µm).
SPOT orbits at 832 km and repeats the orbit every 26 days. Each of the two sensors images a 60-km-wide swath. Pointed vertically, the two sensors can record a 117-km-wide swath. The sensors can be pointed, which provides two major advantages. First, a particular site can be imaged not only from the path directly over the site but also from adjacent satellite paths. This allows the potential for acquiring data for a site more frequently than every 26 days. Secondly, stereo images can be produced by acquiring scenes for the same area from two widely separated locations.
SPOT's panchromatic band has nine times more spatial detail than Landsat TM data. Combining Landsat's spectral data with the spatial advantages of SPOT's panchromatic data can produce spectacular images. A short-wave infrared spectral band is planned for the fourth satellite of the SPOT series.
Coastal Zone Color Scanner
The Coastal Zone Color Scanner (CZCS) was launched on the Nimbus-7 satellite
by the U.S. Government in 1978 and operated until June 1986. The CZCS measures
ocean color and temperature with six spectral bands, including four bands
measuring narrow portions of the visible spectrum, a near-infrared band, and
a thermal-infrared band. This sensor provides spatial resolution of 0.825
km2 at nadir and a scan width of 1,600 km. CZCS data have been
used successfully to map suspended sediments and phytoplankton in coastal
regions (Clark and Maynard 1986, Tassan and Sturn 1986) and in the detection
of acid-waste pollution (Elrod 1988).
AVHRR Sensor
In 1979 the NOAA-6 satellite and all subsequent satellites in the NOAA series
carried the Advanced Very High Resolution Radiometer (AVHRR) sensor. The spatial
resolution of the AVHRR varies from 1.1 km2 at nadir to 12.6 km2
at the end of the scan line. The sensor scans throuah 110.8° as it examines
the earth, producing a scanline of 2.925 km. This wide scan angle, +
54° of nadir, permits daily views of the earth. The AVHRR measures reflected
radiation in the red and near infrared wavelengths and emitted thermal radiation
in three spectral bands. Two data formats; are available from NOAA--local
area coverage (LAC) at full spatial resolution and reduced resolution global
area coverage (GAC) with a spatial resolution of 4 km2 at nadir.
Originally designed to provide improved determination of hydrologic, oceanographic,
and meteorological parameters, AVHRR data, because of their high temporal
frequency, also have proven useful for study of the phenology and productivity
of terrestrial ecosystems on continental and global scales (Justice et al.
1985).
The digital data and photographic images are used in a variety of time-critical applications over large areas. AVHRR data have been used by the U.S. Fish and Wildlife Service to monitor snow cover in the arctic region of Canada for use in forecasting production of arctic nesting geese. LAC data were used to monitor water distribution for waterfowl wintering in the Central Valley of California (L. Strong, U.S. Fish Wildl. Serv., unpubl. data). Other uses of AVHRR data were described by Lillesand and Kiefer (1987) and Aronoff (1989).
GOES Satellite
Geostationary Operational Environmental Satellites (GOES) orbit the earth
at an altitude of 36,000 km in the same direction as the earth's rotation.
In this orbit, they maintain a stationary position relative to the earth (geostationary
orbit). Two GOES satellites are operated by the U.S. and cover the western
and eastern parts of North America. Europe and Japan operate additional GOESs.
GOESs provide continuous monitoring of temperature, humidity, and cloud cover
for weather forecasting. GOES data have been used for some GIS applications
over huge regions (Meisner and Arkin 1984).
GOES collects two bands of data, a visible band (0.55-0.75 µm) and a thermal infrared band (10.2-12.5 µm). NOAA can provide data from the visible band at 1-, 2-, 4-, and 8-km resolution and thermal-infrared imagery at 8-km to 14-km resolutions.
MOS-1
The first Japanese remote sensing satellite, the Marine Observation Satellite
1 (MOS-1), was launched in February 1987. It has three sensors: a multi-spectral,
self-scanning radiometer (similar to the Landsat MSS), a visible and thermal-infrared
radiometer (similar to NOAA AVHRR), and a microwave scanning radiometer. No
data tape recorders are on MOS-1, consequently data can be collected only
when the satellite is in view of a ground receiving station. The U.S. has
no such stations, but data are available from two Canadian receiving stations.
JERS-1
Japan launched JERS-1 (Japanese Earth Remote Sensing Satellite) on 11 February
1992. The three sensors of JERS-1 are an L-band (horizontal polarization synthetic
aperture radar system), a visible and near-infrared radiometer, and a short-wave
infrared radiometer. All provide 18-m resolution data.
ERS-1
The European Space Agency's first remote sensing satellite, ERS-1, was launched
in 1991. ERS-1 carries a C-band, vertical polarization synthetic aperture
radar instrument. Both high-resolution (25-35 m) and low-resolution (100 m)
data are available in digital and photographic forms. Canada is planning to
collect and distribute ERS-1 data for much of North America.
RADARSAT
Canada is developing RADARSAT, a radar remote sensing system to be deployed
in 1995, the first Canadian remote sensing satellite. RADARSAT will assume
a sun-synchronous orbit at approximately 800 km. The repeat cycle will be
every 24 days, but with a change in the look angle, data can be collected
for a specific site every 3 days. RADARSAT's synthetic aperture radar (SAR),
a C-band with horizontal polarization, is designed to operate in several modes
to provide numerous options in terms of swath widths, spatial resolutions,
and angles of incidence. The standard beam mode will provide coverage with
approximately 100-km-wide swath with a spatial resolution of 28 m. The wide
swath beam will collect 28-m data over a 150-km swath. In the fine-resolution
beam mode, 10-m data for a 50-km swath will be collected.
In addition to the SAR instrument, RADARSAT will include a scatterometer and two optical instruments. The scatterometer is a microwave sensor that collects data on wind speed and direction for a 600-km swath. One of the optical instruments is a multilinear array sensor, which records four spectral bands at 30-m resolution for a 400km swath. The other optical instrument is an AVHRR sensor capable of collecting five spectral bands at 1,300m resolution over a 3,000-km swath.
The radar data collected by RADARSAT could be valuable for mapping and monitoring wetlands, because radar is an active sensor creating its own illumination source, and data can be collected for areas covered with clouds or even at night. Place (1985) reported that the accuracy of mapping forested wetlands was improved by 85%, when radar images collected from SEASAT were used to complement conventional aerial photography used by photo-interpreters for mapping wetlands (SEASAT was launched in 1978, but failed only 99 days after launch).
Aircraft Sensors
Satellite-based sensors have many advantages for meeting GIS data needs. Satellite
data have low cost per hectare of coverage, a geometric fidelity that facilitates
registration of images to various maps projections, and freedom from mission
planning. However, for some applications, the spatial resolution's of satellite-based
sensors may be too gross, and the temporal frequency or clouds prevent data
acquisition during optimum times. The time of data acquisition can be critically
important for the successful use of the data. For example, temporary wetlands
may be inundated for only a few weeks. Acquiring satellite data when the temporary
or seasonal wetlands are dry makes detection and identification of temporary
or seasonal wetlands difficult. Aerial photography similarly acquired when
the wetland basins are dry will not provide acceptable delineation of wetlands.
Aircraft sensors offer great flexibility of spatial resolution, timing, and wavelengths of spectral data. Aircraft sensor data can be scheduled to be collected at optimum time for extracting information from desired features. Spatial resolution can be as fine as 1 m or as coarse as 50 m and is dependent on the aircraft altitude, the optical system, and the size of the sensor's detector elements. However, when compared to satellite data, aircraft sensors provide a relatively narrow swath width. A major problem with aircraft scanner data is the poor geometry of the data. These data are adversely affected by variations in aircraft attitude (roll, pitch, and yaw) and deviations from the flight line. Digital elevation models (DEMs) and GPS can be used to suppress the geometric problems inherent in aircraft multi-spectral data. Lee (1991) provided an excellent review of applications of aircraft multi-spectral data for classifying and mapping wetlands.
Videography
Airborne videography recently has been used successfully for assessing wetland
and riparian habitats in North Dakota (Cowardin et al. 1988a) and for evaluating
rangeland and other vegetation in Texas (Driscoll 1990). Sidle and Ziewitz
(1990) described the use of aerial videography for wildlife studies. Lee (1991)
described many of the advantages of videography: (1) imagery can be captured
by microcomputer for immediate use; (2) in-flight error-proofing can be done;
(3) narrow-band filters for fine spectral resolution can be used; (4) data
can be acquired in a wide range of atmospheric conditions; (5) data can be
acquired any time; 6) cost of videography systems is low; and (7) standard
digital image-processing techniques, which are typically used on satellite
data, can be used to analyze the video data. Disadvantages of videography
include: (1) images provide coverage of only small areas; (2) resolving power
is much less than that of aerial photography; (3) geometric distortion from
motion in the plane is difficult to correct; (4) spectral resolution of solid-state
detectors is limited to visible and near infrared wavelengths; (5) multi-spectral
data collection is difficult because single cameras have problems with focus
for different wavelengths, and multiple cameras require accurate bore sighting
and large camera ports on the aircraft; (6) calibration of video data is difficult
because of automatic gain control; and (7) images are vignetted.
Various video systems have been developed (Mausel et al. 1992), including single-band panchromatic systems, single-band color systems, and multi-spectral systems, some of which include near-infrared capabilities. Everitt and Escobar (1989) described many of the available systems. In GIS applications, the best use of videography may be to update existing information layers.
Because of the expense of acquiring digital data layers for GIS applications by digitizing existing maps or by remote sensing techniques, GIS users always should search for existing digital data sets to meet their data needs, before capturing the data themselves. Sources of existing databases include third-party vendors, federal, government agencies, and state and local government agencies.
Before searching for existing databases, one must have a clear vision of what kind of information is required and exactly how the information is to be used. Knowing the exact data requirements is critical to identifying good potential information sources.
The available digital data sets were produced to satisfy a wide range of users. Consequently, the data are not always suitable for a specific GIS application. The cost, accuracy, and currency of the data vary greatly with existing sources. By the time the data have been collected, reviewed, digitized, edited, and distributed, they may be out of date for some applications. Dulaney (1987) reviewed many of the problems associated with existing databases. Descriptions for some of the more widely used databases follow. The GIS World Source Book (Parker 1991), which is published yearly, is an excellent source of information on existing data available for GIS applications.
Land Use and Land Cover and Assosiated Maps
The Land Use and Land Cover (LULC) and associated data files are available
from the U.S. Geological Survey (USGS) and provide information on five data
layers: (1) land use and land cover, (2) political units, (3) hydrologic units,
(4) census county subdivisions, and (5) federal land ownership. These files
are derived from maps at scales of 1:250,000 and 1:100,000.
Land use and land cover areas are classified into nine major classes: urban or built-up land, agricultural land, rangeland, forestland, water, wetland, barren land, tundra, and perennial snow or ice. Each major class is composed of several minor classes (e.g., forestlands are further classified as deciduous, evergreen, or mixed). This classification system (Anderson et al. 1976) was reviewed by a committee of representatives from the USGS, NASA, Soil Conservation Service (SCS), the Association of American Geographers, and the International Geographical Union. The classification system (Table 2) was designed to be used with data obtained from remote sensors on aircraft and satellites.
The minimum mapping area (smallest area mapped) for all urban areas, bodies of water, surface mines, quarries, gravel pits, and certain agricultural areas is 4 ha. The minimum mapping area for all other categories is 16 ha. Thus, a residential area <4 ha would not be recorded in these files, nor would an area of cropland or pastureland <16 ha. Aerial photographs and satellite data serve as the primary sources used in compiling the LULC maps. Some areas on each map are field checked for accuracy.
The four associated maps are prepared at the same scale as the LULC files. The political units file contains county and state boundaries as shown on USGS maps. The hydrologic-units file was digitized from the 1:500,000-scale state maps delineating hydrologic units, which were compiled by the Water Resources Council and published by USGS's Water Resources Division. The census county subdivisions file shows minor unit divisions or equivalent areas. Census tracts within Standard Metropolitan Statistical Areas (SMSA) are represented in this file. The federal land ownership file delineates surface ownership for all areas >16 ha. Federal subsurface ownerships are not delineated.
The LULC and associated data files are available in vector and raster formats on 9-track tapes or CD-ROMs. The raster format uses a cell size of 4 ha. More information on these files can be obtained from the regional USGS Earth Science Information Centers (ESIC) offices (see Appendix I for the addresses and telephone numbers for USGS ESIC offices).
Table 2. Land use and land cover categories used by the U.S. Geological Survey (Anderson et al. 1976). | |
|
|
Digital Line Graphs
Digital line graphs (DLGs) are the digital representation of the planimetric
information (line map data) shown on a map. DLGs have been compiled by USGS
from 1:2,000,000-scale maps, some 1:250,000- to 1:100,000-scale maps, and
some of the 1:24,000- and 1:62,500-scale maps.
DLGs compiled from 1:2,000,000-scale maps are available for three categories: (1) boundaries, which include state and county boundaries and federally administered lands; (2) transportation, which includes roads, railroads, and airports; and (3) hydrography, which includes streams and water boundaries. A CD-ROM that contains data for all 50 states organized into 21 geographic regions can be purchased for only $32 from the ESIC in Reston, Virginia.
The DLGs compiled from 7.5- and 15-minute topographic quadrangles include nine thematic categories: (1) boundaries; (2) transportation; (3) hydrography; (4) U.S. Public Land Survey System (PLSS) (including township, range, and section information); (5) hypsography, including contours and supplemental spot elevation; (6) vegetative surface cover, including woods, scrubs, orchards, vineyards, and marshes and swamps; (7) nonvegetative features including lava, sand, and gravel; (8) survey and control markers, including horizontal and vertical positions of benchmarks; and (9) hurnamnade features, including cultural features not collected in other major data categories, such as buildings. Any feature shown on a 7.5 or 15-minute topographic map will be delineated on the DLGs. These data are not available currently for many locations. DLG data do not carry quantified accuracy statements. However, the data are inspected for attribute accuracy and topological fidelity.
Digital Elevation Data
Elevation, slope, and aspect can be important information for a variety of
wildlife applications of GISs. Digital elevation data, frequently termed digital
elevation model (DEM) or digital terrain model (DTM), provide elevation information
along a contour or at regularly spaced sample points. Aronoff (1989) described
four basic formats for capturing and storing elevation data. These data can
be used to derive information about the morphology of the landscape such as
slope and aspect, which are important to solar insolation and microclimate.
Algorithms have been developed to extract the drainage network from DEMs and
to partition the landscape into watersheds, subcatchments, and hillslopes
for hydrologic modeling (Jenson and Domingue 1988, Band 1989).
The U.S. Defense Mapping Agency produced the first DEM for the entire U.S. by scanning the contour overlays for all 1:250,000-scale topographic maps. From the scanned contour lines, elevations were sampled every 3 arc-seconds of latitude and longitude (approximately every 90 m). The elevation accuracy of these data ranges from 15-m RMSE (Root Mean Square Error) to 60-m RMSE, depending on the terrain. RMSEs are usually lower in flat terrain and increase in steep terrain. These DEM data are sold by USGS in sections 1 X 1 degree in size.
USGS is compiling elevation data from the 7.5-minute topographic maps. From these maps, elevation is sampled every 30 m. Vertical accuracy varies from 7 m to 15 m. DEMs derived from the 7.5-minute topographic maps are available for about 50% of the U.S. as of 1993. Errors in the elevation data can introduce significant errors into calculations of slope and aspect. Errors tend to occur in areas of rapid change in slope and exposure such as along ridges and ravines (Davis and Dozier 1990).
National Wetlands Inventory
More than 30,000 detailed wetland maps have been produced by the National
Wetland Inventory Program (NWI). NWI maps cover nearly 70% of conterminous
U.S., 21 % of Alaska, and all of Hawaii. Most of the maps cover the same area
as covered by the 7.5-minute, 1:24,000-scale topographic maps distributed
by USGS, but some NWI maps have been produced at scales as small as 1:100,000
(Gravatt 1991).
The maps have excellent consistency, because one classification system (Cowardin et al. 1979), one set of photo-interpretation conventions, and one set of cartographic conventions were used. In April 1991, more than 1,100,000 copies of NWI maps had been distributed (Gravatt 1991). The USFWS is on schedule to complete the mapping of the conterminous U.S. by 1998 as required by the Emergency Wetland Resource Act of 1986. Mapping in Alaska should be completed by 2000.
In 1991, digital data files were available for more than 6,200 maps representing 10.5% of the continental U.S. An index map that shows the current availability of digital NWI data can be obtained from NWI in St. Petersburg, Florida. Digital data are available for sale from the USGS's ESIC offices for $25/map; they are available on magnetic tape in MOSS export, DLG3 optional, or GRASS formats (Gravatt 1991).
Digital Soils Data
The SCS has the responsibility for the National Cooperative Soil Survey (NCSS),
which includes collecting, storing, maintaining, and distributing soils information
for privately owned lands in the U.S. (Nielsen 1991).
The SCS has established three digital geographic databases for soil. Each consists of a spatial component that describes the location of the named soil unit and an attribute component that describes characteristics of the soil unit in detail. These digital data help facilitate the storage, retrieval, analysis, and display of soil data in a highly efficient manner. These data can be integrated readily with other spatial and demographic data in GISs. A soils data layer may be one of the most important GIS data layers for wildlife applications.
The Soil Survey Geographic database (SSURGO) is a vector database describing soil delineation boundaries. The boundaries of the soil units are delineated from aerial photographs ranging in scale from 1:15,840 to 1:31,680 combined with extensive fieldwork. The delineated soil boundaries are transferred to 7.5-minute orthophotoquads or topographic maps before the digitizing proceeds.
State Soil Geographic database (STATSGO) was digitized from 1:250,000-scale topographic maps, on which a generalization of the detailed soil surveys was mapped. For areas for which detailed soil survey maps were not available, the generalized soils information was compiled from existing data on geology, topography, vegetation, and climate. STATSGO data are distributed as complete coverage for a state.
The National Soil Geographic database (NATSGO) was derived from general soil maps for each state. NATSGO data were digitized from a map covering all of the U.S. at a scale of 1:7,500.
The Soil Interpretations Record (SIR) database provides attribute data describing the characteristics for each map-unit component and interpretative data for numerous uses. The accuracy of these maps is not determined. Data standardization between field surveys has been a problem, and as a result soil types and properties at the boundaries of adjacent maps often disagree (Burke et al. 1991). GIS technology not only will revolutionize the way the data are analyzed and displayed, but also the way data are collected. SSURGO, STATSGO, and NATSGO data files and the associated attribute files are available from SCS. NATSGO costs $500 for the entire U.S., STATSGO costs $500/state, and SSRUGO costs $500/county. More information on the availability and distribution of these databases can be obtained from USDA's National Cartographic Center, Fort Worth, TX 76115.
DIME Files
The U.S. Bureau of the Census created a spatial data set describing street
networks, street addresses, political boundaries, and major hydrographic features
for approximately 350 major cities and suburbs in the U.S. These files were
created with the Dual Independent Map Encoding (DIME) system to automate the
processing of the 1970 and 1980 U.S. censuses. DIME files have limited application
as a digital map base. For example, streets are represented by straight lines
connecting adjacent intersections. Even a curved street is represented by
a straight line.
TIGER Files
To overcome the limitations of the DIME files and to prepare for the 1990
census, the U.S. Bureau of the Census developed the TIGER (Topologically Integrated
Geographic Encoding and Referencing) system. The TIGER files provide vector
data for hydrography, transportation, political, and statistical areas (such
as county, incorporated area, census tract, and census block). Data collected
from the 1980 and 1990 censuses, such as population, number of housing units,
income, occupation, and housing values, serve as attribute data for these
files. Nearly all commercially available GIS software systems have procedures
for importing TIGER data. Various companies have developed inexpensive GIS
systems strictly for the use of TIGER files and the associated census data.
These companies sell hardware, software, TIGER, and census data as a complete
package. TIGER data also can be purchased directly from the Bureau of the
Census, Washington, DC 20233. With the release of the TIGER files, the Census
Bureau no longer supports or sells the DIME files.
The TIGER files comprise one of the most detailed computerized digital map databases ever developed for the U.S. More than 7 years and $200 million dollars were required to complete the TIGER files. The complete TIGER files for the entire U.S. contain nearly 40,000,000 line segments and require more than 15,000 megabytes of storage for the vectors alone (Anonymous 1989).
Appendix II provides the addresses and telephone numbers for many sources of digital data in the U.S. and Canada.
Understanding of remote sensing models and their interrelationships can benefit from a system view of the image-forming process (Swain and Davis 1978). An important concept is the distinction between the scene, which is real and exists on the earth's surface, and the image, a collection of spatially arranged measurements from the scene (Strahler et al. 1986). The purpose of a remote sensing model is to provide a conceptual and explicit framework for inferring the characteristics of the scene from the image. A remote sensing model may be generalized as having three components: a scene model, an atmospheric model, and a sensor model.
A scene model quantifies the relationships of the objects or targets of interest and their interactions with radiation through the processes of reflectance, transmittance, absorbance, and emittance. Characteristics of the scene objects could include their type, size, number, and spatial and temporal distributions. The model also must consider the background or nontarget components of the scene, including shadow.
An atmospheric model describes the transformation of the radiance due to scattering by molecules and aerosols, and gaseous absorption during the path from the sun to the earth's surface and between the surface and the spacecraft. If an atmospheric model is omitted, the parameters developed to extract information from the image are not transferable and the entire procedure must be repeated for other images. Several methods for the normalization or radiometric calibration of remotely sensed data have been developed (Ahern et al. 1987, Schott et al 1988, Chavez 1989, Tanre et al. 1990).
The sensor model quantifies how the instrument collects the measurements of the scene and includes four key parameters: spectral, spatial, and temporal resolution, and view angle (Duggin 1985). The spectral resolution of the sensor specifies what wavelengths of the electromagnetic spectrum are measured. The spatial resolution specifies the size of the area on the ground from which the measurements that comprise the image are derived. The spatial resolution relative to the spatial structure of the scene objects determines the appropriate analysis methods for scene inference (Woodcock and Strahler 1987). The temporal resolution specifies the frequency with which images are obtained in time. View angle is an important component of the imaging geometry. View angle and illumination geometry (solar zenith and azimuth angles) are important determinants of the measured reflectance since adjustments in observation and illumination geometry result in different sampling of the bidirectional reflectance distribution function, the most fundamental property describing the reflection characteristics of a surface (Silva 1978). Multidirectional observation of this reflectance anisotropy will be possible with the new generation of sensors (Ormsby and Soffen 1989).
Digital image processing, the numerical manipulation of digital images, includes procedures for preprocessing, enhancement, and information extraction. Preprocessing involves procedures applied to the original data before enhancement or information extraction. Calibration of image radiometry for atmospheric conditions and illumination and view geometry, the correction of geometric distortions and georegistration of the image, and noise suppression are examples of image-preprocessing procedures (Schowengerdt 1983).
Image enhancement involves the application of procedures designed to facilitate the interpretation of images. These procedures include contrast and color manipulations and spatial-filtering methods (Schowengerdt 1983). The "Tasseled Cap" is a well-known spectral transformation, which derives new variables that allow vegetation and soils information to be extracted, displayed, and understood more easily (Crist et al. 1986). Hodgson et al. (1988) used this transformation with Landsat TM data in a study of wood stork foraging habitat. Jackson (1983) provided a general procedure to develop spectral indices for user-defined features in a scene.
The development of scene models for extracting information from remotely sensed data requires an understanding of the image-forming process. Strahler et al. (1986) provided a framework for identifying appropriate scene models given the characteristics of the image and the scene. The most common information-extraction methods used with remote sensing data are spectral classifiers in which each pixel is processed independently of its neighbors or location in the image. A discrete scene model is appropriate when the scene objects are larger than the spatial resolution of the sensor.
The parameter estimation process for spectral classifiers can be generalized as being supervised or unsupervised (Swain and Davis 1978, Schowengerdt 1983). In supervised classification, a sample of image elements for each land cover class is used to estimate parameters, typically a mean vector and covariance matrix, for input to the classifier. In unsupervised training, a clustering algorithm is used to partition a sample of the data into populations of pixels with similar reflectance, which are referred to as spectral classes and parameters estimated for these spectral classes (Richards and Kelly 1984). In unsupervised training, the analyst then attempts to establish a correspondence among the spectral classes and the land-cover classes. A statistics file consisting of a mean vector and covariance matrix for each land-cover class then is input to a classification algorithm. The output from a maximum likelihood classification, a common method that produces results having the minimum probability of error over the entire set of data classified, is an image in which each pixel is assigned the label of the land-cover class for which the a posteriori probability was the maximum. An enhancement to the standard output from the maximum likelihood classification would be to create a raster for each land-cover class wherein the pixel value would be the a posteriori probabilities of membership for the category. The result is a probabilistic digital map of the geographic distribution for each land-cover class. This would increase the computational and storage requirements, but technological progress in these areas is great (Faust et al. 1991).
In a continuous-scene model, the scene objects are smaller than the resolution element of the sensor. A relationship between the reflectance and a property of a scene, such as canopy coverage, is established and used to estimate the property in each pixel in a continuous fashion. Mixture models are a type of continuous-scene model, in which the objective is to estimate the proportions of scene objects in each pixel. Mixture models have been used for a variety of resource inventories, including waterfowl habitat (Work and Gilmer 1976), rangeland vegetation and soil cover (Pech et al. 1986), and wintering geese (Strong et al. 1991).
Spectral-spatial scene models exploit the spatial structure of images as well as their spectral characteristics to infer the properties and processes at the land surface. A variety of spectral-spatial models is available. Some of these scene models segment the image into contiguous groups of pixels that meet a spectral similarity criterion and perform the classification using all the pixels of the feature (Strahler et al. 1986). Other spectral-spatial models exploit a measure of image texture or the spatial autocorrelation function as an additional feature in the classification process (Shih and Schowengerdt 1983, Pickup and Chewings 1988).
Spectral-temporal models use the change in the spectral properties of images acquired at different times to infer properties or processes at the land surface. The "Tasseled Cap" is an example of a spectral-temporal model of the phenological development of agricultural crops that can be used to identify crops and forecast yields (Kauth and Thomas 1976, Wiegand et al. 1986). Time series of the normalized difference vegetation index (NDVI), calculated from the red and infrared spectral reflectance measurements of the AVHRR sensor, have been used to describe and map the intra- and inter-year phenological dynamics of biomes at regional, continental, and global scales (Justice et al. 1985), to infer net primary productivity (Goward et al. 1985), and to measure the dynamics of vegetation at the transition zones between biomes (Tucker et al. 1991). Various techniques for detecting change (Singh 1989) use images acquired at different times to infer changes in land cover.
The flow of information between remote sensing and GIS should not be one-way. The accuracy of information derived from remote sensing can benefit from access to accurate spatial data within a geographic information system. Integration of the parallel technologies of GIS and RS will be important to the fullest maturation of both areas.
The analysis capabilities of the GIS provide the answers to the five basic questions defined by Walker and Miller (1990). These five basic questions, which were discussed previously, should be reviewed before the analytical capabilities of GISs are studied.
Aronoff (1989) described four major GIS analysis capabilities: (1) maintenance and analysis of spatial data, (2) maintenance and analysis of attribute data, (3) integrated analysis of spatial and attribute data, and (4) output formatting. The first two functions deal with maintaining the data and the fourth function is for producing output from the system. The real power of the GIS is its ability to integrate the analysis of spatial and attribute data. Spatial data can be stored as raster or vector structures. Some analytical functions can work equally well with raster or vector data, some analytical functions will perform better with raster data, and some functions will perform better with vector data. GIS systems of the future will have improved user interfaces and expert systems to advise the user on how to utilize the existing database and software to obtain the desired resource information (Coulson et al. 1987, Goodenough et al. 1987, McKeown 1987).
Maintenance and analysis capabilities of the GIS will permit transforming spatial data files, editing spatial data, and assessing the quality and accuracy of the data. Aronoff (1989) described seven maintenance and analysis functions for spatial data: format transformation, geometric transformation, transformation between map projections, conflation, edge matching, editing of graphic elements, and line coordinate thinning.
Because data for the GIS may come in many formats (e.g., DLG, TIGER, MOSS, GRASS, raster run-lengths encoded, and DIME), GISs must be able to transform data among various formats. Conversions from various formats and from raster to vector and vector to raster are examples of format transformation. Because of inaccuracies in source maps, slight differences in map projection, or digitizing errors, data layers may not be precisely registered to a common coordinate system or a standard data layer. Geometric transformations are procedures used to ensure that each data layer precisely overlays the other data layers. Some data may be supplied in geometric coordinates (latitude and longitude), whereas other data may be supplied in UTM or state plane coordinates. Functions to transform from one map projection to another are required to transform all sources of data to a common map projection.
Conflation functions are used to ensure that common boundaries between different data layers share the exact coordinates. For example, the edge of one end of a wetland is adjacent to a gravel road, and the gravel road is the boundary on a wildlife management area. The location of the wetland was digitized for the wetland data layer, the road was digitized for the transportation data layer, and the wildlife management area was digitized for the public lands data layer. When all three layers were plotted as one map, the common boundaries of the wetlands, road, and wildlife management area are similar, but not exactly the same. Conflation procedures ensure that common boundaries are defined with exactly the same coordinates.
Usually all features on one map sheet will be digitized and then all features on the adjacent map will be digitized. Because of slight error in the maps and imprecisions in the digitizing process, the boundary of a feature crossing onto the adjacent map may not match precisely. Edge-matching functions ensure that the edge of a feature on one map perfectly matches the feature edge on the adjacent map.
GISs should have numerous editing functions to aid in adding, deleting, and changing the geographic positions of features. In the digitizing process, more coordinates are collected for a line than are actually required to represent the line. Line-coordinate thinning processes can eliminate coordinates that are not essential to represent the line and can greatly reduce the disk space required to store the coordinates needed to represent the line.
Just as various functions are required to edit, convert, and maintain spatial data, functions for editing, converting, and maintaining the nonspatial attribute data also are required. Attribute editing functions should allow for retrieving, editing, and changing the attributes. Attribute query functions allow records in the attribute databases to be selected. Using the attribute data query functions on the attribute database for the soils layer would allow one to select all soils that are sandy and have a "T" value >4 (T values represent the tons of topsoil likely to be eroded per 0.4 ha/year). Attribute query functions should support the use of complex Boolean expressions on the attributes of one or more data layers.
The ability to effectively process spatial and attribute data is what primarily distinguishes GISs from automated mapping and computer-aided drafting systems (Aronoff 1989). The integrated analysis of spatial and attribute data can be divided into four categories (Berry 1987, Aronoff 1989): retrieval/classification/measurement, overlay, neighborhoods, and connectivity of network functions.
The retrieval processes are used to answer the first three basic questions that a GIS can address (Walker and Miller 1990): what exists at a particular location; where are certain conditions met; and what changes have occurred and where have changes occurred over time? Retrieval processes operate on spatial and attribute data. Output of the selective searches can create new layers in the database, tabular reports, interactive displays, or maps.
Classification processes frequently are performed after a retrieval process and are used to assign a new attribute. If deciduous, coniferous, and mixed forest types are retrieved from a vegetational data layer, a classification function could be used to assign a new attribute, forests, to the areas retrieved. Classification functions can be applied on one layer or on multiple data layers.
Measure functions calculate distances between points, lengths of lines, perimeters and areas of polygons, and sizes of continuous areas of the same feature. Selecting all wetlands >8 ha in size from a data layer describing the various wetlands types would require a retrieval process (retrieve all wetlands types), a classification process (classify the various retrieved wetlands types as one category), a measurement process (determine the size of all continuous area of wetlands), and a final retrieval process (select all wetlands >8 ha).
Overlay operators are some of the most fundamental and most frequently used processes in GIS applications. Aronoff (1989) described two types of overlay operators-arithmetic and logical. Arithmetic operators are used to add, subtract, divide, or multiply values in one data layer by a constant or by values in another data layer in corresponding location (Fig. 13). Logical overlays are used to identify areas where one feature or condition exists in one data layer and another feature or condition exists in another data layer. The logical overlays can use various logical Boolean operators such as "and" where both condition A and condition B are met, "or" where condition A or condition B is met, and "and not" to idenffy areas where condition A is met, but not condition B (Fig. 14).
Neighborhood processes evaluate the characteristics of the area surrounding a specific location (Aronoff 1989). Measuring the length of edge between habitat types within 2 km of the location of a radio-telemetered animal is an example of a neighborhood function. All neighborhood processes require three parameters: the location of one or more target areas, the size of the neighborhood surrounding the targeted area, and a function to perform in the defined neighborhood and assigned to the targeted area. Five numerical functions can be applied in the neighborhood: average, diversity, majority, maximum/minimum, and total (Aronoff 1989).
Point-in-polygon and line-in-polygon operations are fundamental neighborhood process for vector-based GISs (Aronoff 1989). A point-in-polygon function is used to determine which features (as defined by polygons) cover a specific point. In analysis of radiotelemetry data, point-in-polygon routines are used in vector-based GISs to ascertain various habitat parameters occurring at each location made for the animal being studied. In a similar manner, line-in-polygon routines are used to ascertain which polygon contains a specified line.
Various topographic functions can be computed from elevation information provided by DEMs or DTMs. Topographic functions usually are computed from raster data of elevation. The elevation for a particular cell (X), plus the elevation of the eight orthogonal (O) and diagonal (D) neighbors (Fig. 15), can be used to ascertain slope, aspect, and topographic position (ridge, valley, knoll). These topographic parameters frequently are highly correlated with the distribution of plant and animal species, and they frequently can be used in remote sensing applications to distinguish spectrally similar habitats. For example, spectrally, coastal dunes often cannot be separated from sandy flats. Slope and topography position parameters can readily separate these two spectrally similar sandy habitats.
Other topography functions include illumination techniques to enhance display of elevation or relief, viewshed modeling to determine what areas can be seen from a specific point for appraising visual impacts, watershed analysis to determine the extent of the watershed, and perspective views to produce illustrations of the relief as viewed from a given point. These topography functions usually are classified as connectivity functions. Various other neighborhood functions are available for interpolating information from various point locations.
Connectivity functions accumulate values over a connected area (Aronoff 1989). Connectivity functions require (1) a procedure for connecting the areas and (2) a measure for the connected area. Continuity measures are connectivity functions that typically measure the size of a continuous area. Evaluating habitat for a particular avian species using bottomland hardwood forests will require the locations of all bottomland hardwoods. However, if the species is known to use only bottomland hardwood forests that are at least 50 ha in size, a continuity measure would be used to ascertain the size of each continuous block of bottomland hardwood forests. Continuity measures are a valuable tool for measuring habitat fragmentation.
Proximity to a habitat type also may be valuable in evaluating the availability of habitat to certain species. Proximity analysis is the measure of distance between features. Figure 16 illustrates the use of proximity function to define the area within 250 m of all streams.
![]() |
Fig. 16. -- A proximity function was used to define all areas within 250 m of streams. |
Various other connectivity functions are used in GIS applications. Network functions are processes to optimize vehicle routing and to divide areas into service districts for optimizing limited resources (Aronoff 1989). Spread functions can be used to evaluate the cost of transversing an area and often are used to evaluate alternative routes for transportation and utilities. Such routines will find the lowest cost route between two points. Cost can be measured separately for economic, social, or environmental cost, or total costs can be calculated by summing the weighted economic, social, and environmental costs.
The previous sections described many of the processes of a GIS. Tomlin and Berry (1979) and Berry (1987) developed the term "cartographic modeling" to describe the use of basic GIS processes in a logical sequence to solve complex spatial problems. A series of standard processes, such as reclassifying, overlaying, distance, and neighborhood operations, is used to subdivide the spatial problems into a series of primitive operations. A cartographic model can be depicted with a flowchart. The initial data layers are shown, followed by a standard process performed on the data layer, followed by an output (often a data layer) created by the processing function performed. An output data layer then may be used as input to another process. Various inputs, processes, and outputs are diagrammed until the solution is derived. The cartographic model provides the pathway for solving the specified spatial problem. Prior to implementing the cartographic model with the GIS, the flowchart of the cartographic model can be reviewed and easily revised. Cartographic models are an excellent means to help determine which data layers will be required to solve any spatial problem.
Berry (1987) identified many advantages of the use of cartographic models. These models are capable of dynamic simulations and provide spatial "what if" analysis. Koeln (1980) used a GIS to identify the "best" (or least bad) sites for general aviation airports. Three different sets of weights for economic, social, and environmental impacts were used to simulate the attitude of three different groups of decision makers. One set of simulated decision makers (the misers) was concerned primarily with the economic costs but recognized some social and environmental costs. Another set of simulated decision makers (the environmentalists) weighted the environmental costs of general aviation airport the highest, followed by social costs, then economic costs. The altruist group of decision makers weighted economic, social, and environmental costs nearly equally. Through the cartographic modeling approach, three different sets of "best sites" for general aviation airports were selected based upon weights assigned by the simulated decision makers. Because the choice of location for general aviation airports is quite constrained, many of the best sites as determined from the misers' weights were identical to the best sites chosen with the weights of the environmentalists and altruists.
Another advantage of the cartographic modeling approach is its flexibility. New considerations can be added easily or existing ones can be refined. The cartographic modeling approach also provides an effective means for communicating the process used, including consideration of the specific application and fundamental procedures applied. A flowchart provides an excellent tool for communicating the logic, assumptions, and relationships used in the analysis.
Information from GISs must be reported. The reports can be statistical tabulations, such as reports of various habitat parameters for a study area or the polygon defining the home range of an animal, interactive displays of data on color and monochrome monitors, and hard-copy maps. Both hard-copy and soft-copy (images on a color or monochrome monitor) are essential. The reporting capabilities of GISs vary greatly, depending upon the software used. Modem GISs have interactive capabilities to display images on color or monochrome monitors. Most modem GISs have improved, and are improving, hardcopy output capabilities. Some systems have map automation procedures, text-labeling abilities, and graphic capabilities that nearly match the map-making capabilities of automated mapping and drafting systems. The reduction in the cost of image display devices, color plotters, and film recorders allows many GIS users to afford quality output devices. The expansion of software output capabilities seen for most commercially available GISs reflects the availability of an increasing array of quality output devices.
The same advances in computer technology that have led to recent developments in the use of GISs have led to increases in the use of simulation models that attempt to portray the population biology of waterfowl species. Population data required by these simulation models usually are difficult and expensive to obtain. For many species, habitat availability and quality and various population parameters are correlated. Fortunately, recent advances in remote sensing have made inexpensive estimates of the availability of habitat over vast areas available to waterfowl managers. A marriage between GIS and simulation modeling is the logical outcome of these technological developments. As an example, we will illustrate the use of such a combined GIS and population simulation model for evaluating management practices for mallards in the prairie pothole region of North Dakota.
For the following example, we selected a 10.4-km2 plot from typical duck habitat in central North Dakota. The area is glaciated and has numerous small wetland basins. The uplands are used almost exclusively for agriculture. Landsat TM data from May and September of 1986 processed by methods developed by Koeln and Wesley (1987) and Koeln et al. (1988) were used. The digital image processing techniques created a GIS data layer for land use and land cover. The retrieval and reporting functions of a GIS were used to report the areas for three wetland and 10 upland classes. This habitat information was altered to represent the habitats available to birds arriving in early May. For example, areas of growing grain were assumed to have been bare soil (recently plowed ground) when birds arrived in late April. Habitat classes conformed with the classes (Cowardin et al. 1988b) required as data input for a mallard productivity model (Johnson et al. 1987). That model allows the user to vary population parameters and habitat availability to predict the resulting production of young. Cowardin et al. (1988b) illustrated the use of the model in combination with other models that predict the number of breeding birds settling in an area. The same models have been linked with habitat data derived from high-altitude photography and aerial video (Cowardin et al. 1988a).
Population parameters used in the example were based on information from Cowardin et al. (1988b) and Klett et al. (1988). The habitat data (Table 3) can be displayed on the monitor of a microcomputer. These data then were transmitted to the mallard productivity model, and the model was executed to obtain estimates of expected breeding population and young produced as well as other important population parameters (Table 4). For our demonstration, the control data represent current conditions.
Table 3. Availability of nesting habitat for a 10.4-km2 area in central North Dakota used to illustrate linkage of a population simulation model to a GIS with habitat data derived from Landsat TM. | ||||
Autumn-plowed crop | ||||
Grain stubble | ||||
Summer fallow | ||||
No-till winter wheat | ||||
Grassland | ||||
Hayland | ||||
Right-of-way | ||||
Shallow wetlandb | ||||
Deep wetlandb | ||||
Permanent water | ||||
Fenced cover | ||||
Conservation reserve | ||||
Barren | ||||
a The treatments were: (0) control, no management;
(1) creation of an impoundment; (2) half of cropland converted to conservation
reserve; (3) addition of a 148-ha predator barrier fence. b Only the portion of the wetland representing nesting cover is shown; the remainder is included in barren. |
The problem for the waterfowl manager is to select from an array of potential management alternatives the one most likely to maximize production from an area. The simulation models also can be combined with economic models to produce a planning tool (Nelson and Wishart 1988). The system demonstrated here combines GIS and simulation modeling. It cannot tell the manager what management should be applied, but it is a tool to assist evaluation of the outcome of various alternatives. The software has the advantage of allowing this to be done by modifying the habitat displayed on the computer monitor. For illustration, we used the editing functions of the GIS to simulate three management techniques. These techniques were represented by three different land use and land cover data files. In one simulation, we created a large, semipermanent wetland. In another simulation we converted one-half of the cropland to planted cover under the USDA's CRP. In the third simulation, we constructed a predator barrier fence around 148 ha of tall, dense nesting cover (Figs. 17-19). The predicted results are highly dependent on the assumptions made for many population parameters in the model. As a first approximation we used the same assumptions made by Cowardin et al. (1988b) and unpublished nest survival rates prepared for data in Klett et al. (1988).
Results of the simulations are summarized in Table 4. All treatments show some increase in the number of recruits produced. Creation of the large impoundment was the only treatment that increased the amount of wetland used by breeding pairs. The model predicted a corresponding increase in breeding population and recruits produced, but the increase in nest success was negligible. Conservation reserve was modeled under two different nest survival rates for that cover because there were no published data. We therefore used nest survival of 13.2% (unpubl. data, Northern Prairie Wildl. Res. Cent., Jamestown, N.D.) for planted cover in simulation (a) because that cover is similar to what is expected under conservation reserve. Some might argue that nest success in conservation reserve should be higher than planted cover because the large amount of conservation reserve may dilute the effect of predation. In conservation reserve (b), without real data, we chose a nest success rate of 20% which is relatively high but reasonable. The treatment then produced as many recruits as the impoundment from the same number of pairs used in the control. Addition of the predator barrier fence produced the most recruits of all treatments.
Table 4. Estimates of mallard population parameters for a 10.4-ha area produced by using habitat availability data deriverd from Landsat TM, a GIS, and a population simulation mmodel applied to a control and three habitat enhancement practices. | ||||
Control | ||||
Impoundment | ||||
Conservation reserve (a) | ||||
Conservation reserve (b) | ||||
Barrier fence | ||||
a The treatments were: (0) control, no management; (1) creation of an impoundment; (2) half of cropland converted to conservation reserve, for (a) nest success in conservation reserve = 13.3%, for (b) nest success in conservation reserve = 20.0%; (3) addition of a 148-ha predator barrier fence. |
Results obtained from the model are extremely sensitive to the input data (Johnson et al. 1987, Cowardin et al. 1988b). It was designed as a tool to assist in comparing the potential outcome to be obtained from treatments. In the example, we used a range of reasonable input data, but there was an absence of data for conservation reserve. Both the input data and the assumptions upon which the model was based must be considered carefully when model results are interpreted.
The model is also sensitive to the amounts of habitat present in the area to be evaluated. Fortunately, these areas can be measured more accurately than some of the mallard population parameters by use of remote sensing techniques. In previous applications of the model, construction and modification of data sets that simulated various treatments were time- and labor-intensive. The combination of GIS technology with the model overcame this problem. Furthermore, the display of a real landscape and the rapid modification of that landscape by means of GIS editing functions are easily understandable by a land manager. Through the efforts of the North American Waterfowl Management Plan (NAWMP), many waterfowl habitat-enhancement techniques are being applied to the prairie pothole region. By comparing land cover and land use data derived from the 1986 satellite data with data derived from more recent satellite data, we will have an excellent record of the success of the NAWMP in changing the landscape for improving waterfowl habitat. The effects of these actual changes will be evaluated with the GIS and mallard model approach described above.
The presence of error on maps begins with the process of map projection (Vitek et al. 1984). A map projection is a systematic representation of all or part of the three-dimensional earth to a two-dimensional plane. Since this cannot be done without distortion, the user must choose the map property to be shown accurately at the expense of others, or a compromise of several properties (Snyder 1987). Error is introduced to maps in the process of cartographic abstraction. A map is a model of reality, and map contents are often elegant misrepresentations of changes that are often gradual, vague, or fuzzy (Burrough 1986). Map errors involve both positional and attribute accuracy, which in some instances can be difficult to separate. Furthermore, both position and attribute error are a function of scale.
Errors accumulate during the analysis process in a GIS. Newcomer and Szajgin (1984) demonstrated a method for estimating the error propagation during the map overlay process. They concluded that the accuracy of the final map is a function of the number of map layers, the accuracy of these layers, and the coincidence of errors at the same position from several map layers. The accuracy of spatial databases and the propagation of error during data analysis are complex issues (Newcomer and Szajgin 1984, Vitek et al. 1984, Walsh et al. 1987, Goodchild and Gopal 1989, Lunetta et al. 1991).
The future for the integration of GIS, remote sensing, and expert systems is promising. In 1988, the National Science Foundation announced the formation of the National Center for Geographic Information Analysis, a consortium of universities including the University of California at Santa Barbara, The State University of New York at Buffalo, and the University of Maine at Orono. The Center has outlined a program aimed at the systematic removal of perceived impediments to the adoption and use of GIS technology. The program consists of a series of initiatives, several of which already have resulted in publications and symposia.
At least 100 available GISs of various types for a wide array of computers are available from many government agencies, universities, and commercial vendors (Parker 1991). It is hoped that wildlife managers desiring to use GIS technologies will not get bogged down in the bits, bytes, and ". . . primordial ooze of system development or primitive promotions" (Giles 1991:5), but will become excited about the capabilities provided by GISs to gain ".. . explanatory, descriptive, and predictive control ..."(Giles 1991:5) of ecosystems.
Printed in the United States of America for The Wildlife Society by Allen Press, Inc., Lawrence, Kansas
All rights in this book are reserved. No part of the book may be used or reproduced in any manner whatsoever without written permission except in the case of nonprofit educational reproduction, use by agencies of the U.S. Government, or brief quotations embodied in critical articles and review. For information, address: The Wildlife Society, Inc., 5410 Grosvenor Lane Bethesda, Maryland 20814
Mention of trade names does not imply endorsement of the product.
This book is the fifth in a series on wildlife techniques published by The Wildlife Society
Editor, Henry S. Mosby
Manual of Game Investigational Techniques
(1) First Edition-May 1960
Second
Printing-February 1961
Wildlife Investigational Techniques
(2) Second Edition-May 1963
Second
Printing (Revised)-March 1965
Third
through Sixth Printing-March 1966 to September 1968
Editor, Robert H. Giles, Jr.
Wildlife Management Techniques
(3) Third Edition-June 1969
Second
Printing (Revised)-January 1971
Third
Printing-May 1972
Editor, Sanford D. Schemnitz
Wildlife Management Techniques Manual
(4) Fourth Edition-September
1980
Editor, Theodore A. Bookhout
Research and Management Techniques for Wildlife
and Habitats
(5) Fifth Edition-January
1994
Suggested citation formats:
Bookhout, T. A., Editor. 1994. Research and management techniques for wildlife and habitats. Fifth ed. The Wildlife Society, Bethesda, Md. 740pp.
Johnson, D. H. 1994. Population analysis. Pages 419-444 in T. A. Bookhout, ed. Research and management techniques for wildlife and habitats. Fifth ed. The Wildlife Society, Bethesda, Md.
This book was produced on the Penta DeskTopPro/UX® and output to an AGFA SelectSet 7000 imagesetter. The text is Adobe Times Roman. The text paper is 50 pound Simpson Offset (50/10 recycled). The Roxite cloth cover was printed by Allen Press, Inc. and the case binding was done by Prizma Industries, Denver, Colorado. This book was printed on a Hantscho full-sized waterless web press by Allen Press, Inc.
ISBN 0-933564-10-4
Library of Congress Catalog Card Number: 93-61624
ABLER, R. F. 1987. The National Science Foundation National Center for Geographic Information and Analysis. Int. J. Geogr. Inf. Syst. 1:303-326. AGEE, J. K., S. C. F. STITT, M. NYQUIST, AND R. ROOT. 1989. A geographic analysis of historical grizzly bear sightings in the North Cascades. Photogram. Eng. Remote Sens. 55:1637-1642. AHERN, F. J., et al. 1987. Radiometric correction of visible and infrared remote sensing data at the Canada Centre for Remote Sensing. Int. J. Remote Sens. 98:1349-1376. ANDERSON, J. R., E. E. HARDY, J. T. ROACH, AND R. E. WITMER. 1976. A land use and land cover classification system for use with remote sensor data. U.S. Geol. Surv. Prof Pap. 964. 28pp. ANONYMOUS. 1989. Using the TIGER files. U.S. Stat. Newsl. 5(12):1-5. ARONOFF, S. 1989. Geographic information systems: a management perspective. WDL Publ., Ottawa, Ont. 294pp. BAND, L. E. 1989. A terrain-based watershed information system. Hydrological Processes 3:151-162. BARNARD, T., R. J. MACFARLANE, T. NERAASEN, R. P. MROCZYNSKI, J. JACOBSON, AND R. SCHMIDT. 1981. Waterfowl habitat inventory of Alberta, Saskatchewan and Manitoba by remote sensing. Proc. Can. Symp. Remote Sens. 7:150-158. BERRY, J. K. 1987. Fundamental operations in computer-assisted map analysis. Int. J. Geogr. Inf. Syst. 1:119-136. BREININGER, D. R., M. J. PROVANCHA, AND R. B. SMITH. 1991. Mapping Florida scrub jay habitat for purposes of land-use management. Photogram. Eng. Remote Sens. 57:1467-1474. BROSCHART, M. R., C. A. JOHNSTON, AND R. J. NAIMAN. 1989. Predicting beaver colony density in boreal landscapes. J. Wildl. Manage. 53:929-934. BURKE, I. C., T. G. F. KITTEL, W. K. LAUENROTH, P. SNOOK, C. M. YONKER, AND W. J. PARTON. 1991. Regional analysis of the Central Great Plains. BioScience 41:685-692. BURROUGH, P. A. 1986. Principles of geographical information systems for land resources assessment. Oxford Univ. Press, New York, N.Y. 193pp. CANNON, R. W., F. L. KNOPF, AND L. R. PETTINGER. 1982. Use of Landsat data to evaluate lesser prairie chicken habitats in western Oklahoma. J. Wildl. Manage. 46:915-922. CHAVEZ, P. S., Jr. 1989. Radiometric calibration of Landsat Thematic Mapper multi- spectral images. Photogram. Eng. Remote Sens. 55: 1285-1294. CLARK, D. K., AND N. G. MAYNARD. 1986. Coastal zone color scanner imagery of phyto- plankton pigment distribution in Icelandic waters. Pages 350-357 in Proc. SPIE ocean optics VII. Int. Soc. Optical Eng., Billingham. Wash. COULSON, R. N., L. J. FOLSE, AND D. K. LOH. 1987. Artificial intelligence and natural resource management. Science 237:262-267. COWARDIN, L. M., P. M. ARNOLD, T. L. SHAFFER, H. R. PYWELL, AND L. D. MILLER. 1988a. Duck numbers estimated from ground counts, MOSS map data, and aerial video. Pages 205-219 in J. D. Scurry, comp. Proc. Natl. MOSS Users' Conf. 5. Louisiana Sea Grant College Program, Baton Rouge, and U.S. Fish Wildl. Serv., Slidell, La. ------, V. CARTER, F. C. GOLET, AND E. T. LAROE. 1979. Classification of wetlands and deep water habitats of the United States. U.S. Fish Wildl. Serv. Rep. FWS/OBS-79/31. 103pp. ------, D. H. JOHNSON, T. L. SHAFFER, AND D. W. SPARLING. 1988b. Applications of a simulation model to decisions in mallard management. U.S. Fish Wildl. Serv. Fish Wildl. Tech. Rep. 17. 28pp. CRAIGHEAD, J. J., F. L. CRAIGHEAD, AND D. J. CRAIGHEAD. 1986. Using satellites to evaluate ecosystems as grizzly bear habitat. Pages 101-112 in Proc. grizzly bear habitat symposium, Missoula, Mont. CRIST, E. P., R. LAURIN, AND R. C. CICONE. 1986. Vegetation and soils information contained in transformed thematic mapper data. Pages 1465-1470 in Proc. IGARSS' 86 Symp. ESA SP-254. CURRAN, P. J. 1985. Principles of remote sensing. Longman Group Limited, London, U.K. 282pp. DAVIS, F. W., AND J. DOZIER. 1990. Information analysis of a spatial database for ecological land classification. Photogram. Eng. Remote Sens. 56:605-613. ------, D. M. STOMS, J. E. ESTES, J. SCEPAN, AND J. M. SCOTT. 1990. An information systems approach to the preservation of biological diversity. Int. J. Geogr. Inf. Syst. 4:55-78. DE STEIGUER, J. E., AND R. H. GILES, Jr. 1981. Introduction to computerized land- information systems. J. For. 79:734-737. DRISCOLL, D. 1990. Remote sensing: USFS pest management group. GIS World Mag. 3(5):94-96. DUEKER, K. J., AND D. KJERNE. 1989. Multipurpose cadastre: terms and definitions. Am. Soc. Photogram. Remote Sens. and Am. Congr. Surv. Mapping, Falls Church, Va. 5:94-103. DUGGIN, M.J. 1985. Factors limiting the discrimination and quantification of terrestrial features using remotely sensed radiance. Int. J. Remote Sens. 6:3-27. DULANEY, R. A. 1987. A geographic information system for large area analysis. Pages 206-215 in Proc. of GIS '87. Am. Soc. Photogram. Remote Sens., Falls Church, Va. ELROD, J. A. 1988. CZCS view of an oceanic acid waste dump. Remote Sens. Environ. 25:245-254. ESTES, J. A., E. J. HAJIC, AND L. R. TINNEY, editors. 1983. Fundamentals of image analysis: analysis of visible and thermal infrared data. Pages 987-1124 in R. N. Colwell, ed. Manual of remote sensing. Seconded., Vol. 1. Am. Soc. Photogram., Falls Church, Va. EVERITT, J. H., AND D. E. ESCOBAR. 1989. The status of video systems for-remote sensing applications. Proc. Biennial Workshop on Color Aerial Photography and Videography. Am. Soc. Photogram. Remote Sens. 12:6-29. FAUST, N. L., W. H. ANDERSON, AND J. L. STARR. 1991. Geographic information systems and remote sensing future computing environment. Photogram. Eng. Remote Sens. 57:655-668. GILES, R. H., Jr. 1991. Nine thoughts about geographic information systems. Nat. Resour. Comput. Newsl. 6(4):3-5. GOODCHILD, M., AND S. GOPAL, editors. 1989. The accuracy of spatial databases. Taylor and Francis, London, U.K. 290pp. GOODENOUGH, D. G., M. GOLDBERG, G. PLUNKETT, AND J. ZELEK. 1987. An expert system for remote sensing. IEEE Trans. Geoscience Remote Sens. GE-25:349-359. GOSSELINK, J. G., AND L. C. LEE. 1989. Cumulative impact assessment in bottomland hardwood forests. Wetlands 9. 174pp. GOWARD, S. N., C. J. TUCKER, AND D. G. DYE. 1985. North American vegetation patterns observed with the NOAA-7 advanced very high resolution radiometer. Vegetation. 64:3-14. GRAETZ, R.D. 1990. Remote sensing of terrestrial ecosystem structure: an ecologist's pragmatic view. Pages 5-30 in R. J. Hobbs and H. A. Mooney, eds. Remote sensing of biospheric functioning. Springer-Verlag, New York, N.Y. GRAVATT, G. 1991. National Wetlands Inventory. Pages 29-31 in K. K. Reay, ed. Proc. Natl. Conf. Integrated Water Inf. Manage. Virginia Polytechnic Inst. State Univ., Blacksburg. HEINEN, J. T. AND R. A. MEAD. 1984. Simulating the effects of clear cuts on deer habitat in the San Juan National Forest, Colorado. Can. J. Remote Sens. 10:17-24. HODGSON, M. E., J. R. JENSEN, H. E. MACKEY, Jr., AND M. C. COULTER. 1988. Monitoring wood stork foraging habitat using remote sensing and geographic information systems. Photogram. Eng. Remote Sens. 54:1601-1607. JACKSON, R. D. 1983. Spectral indices in n-space. Remote Sens. Environ. 13:409-421. JENSON, S. K., AND J. O. DOMINGUE. 1988. Extracting topographic structure from digital elevation data for geographic information system analysis. Photogram. Eng. Remote Sens. 54:1593-1600. JOHNSON, D. H., D. W. SPARLING, AND L. M. COWARDIN. 1987. A model of the productivity of the mallard duck. Ecol. Model. 38:257-275. JOHNSTON, C. A., AND J. BONDE. 1989. Quantitative analysis of ecotones using a geographic information system. Photogram. Eng. Remote Sens. 55:1643-1647. ------, N. E. DETENBECK, J. P. BONDE, AND G. J. NIEMI. 1988. Geographic information systems for cumulative impact assessment. Photogram. Eng. Remote Sens. 54:1609-1615. ------, AND R. J. NAIMAN. 1990a. Aquatic patch creation in relation to beaver population trends. Ecology 71:1617-1621. ------, AND ------ 1990b. The use of a geographic information system to analyze long- term landscape alteration by beaver. Landscape Ecol. 4:5-19. JUSTICE, C. O., J. R. G. TOWNSHEND, B. N. HOLBEN, AND C. J. TUCKER. 1985. Analysis of the phenology of global vegetation using meteorological satellite data. Int. J. Remote Sens. 6:1271-1318. KAUTH, R. J., AND G. S. THOMAS. 1976. The tasselled cap-a graphic description of the spectral-temporal development of agricultural crops as seen by Landsat. Pages 4B41-4B51 in Proc. symposium machine processing of remotely sensed data. Purdue Univ., West Lafayette, Ind. KLETT, A. T., T. L. SHAFFER, AND D. H. JOHNSON. 1988. Duck nest success in the prairie pothole region. J. Wildl. Manage. 52:431-440. KOELN, G. T. 1980. A computer-assisted general aviation airport location and evaluation system for Virginia. Ph.D. Thesis, Virginia Polytechnic Inst. State Univ., Blacksburg. 235pp. ------, AND E. A. COOK. 1984. Applications of geographic information systems for analysis of radio-telemetry data on wildlife. Pecora 9:154-158. ------, J. E. JACOBSON, D. E. WESLEY, AND R. S. REMPLE. 1988. Wetland inventories derived from Landsat data for waterfowl management planning. Trans. North Am. Wildl. Nat. Resour. Conf. 53: 303-310. ------, AND D. E. WESLEY. 1987. Ducks Unlimited's wetland inventory. Pages 225-233 in J. Zelazny and J. S. Feierabend, eds. Proc. increasing our wetland resources. Natl. Wildl. Fed., Washington, D.C. KORTE, G. B. 1991. How GIS relates to CADD, CAM and AM/FM. Point of Beginning 16:56-66. LAUER, D. T., J. E. ESTES, J. R. JENSEN, AND D. D. GREENLEE. 1991. Institutional issues affecting the integration and use of remotely sensed data and geographic information systems. Photogram. Eng. Remote Sens. 57:647-654. LECKENBY, D. A., D. L. ISAACSON, AND S. R. THOMAS. 1985. Landsat application to elk habitat management in northeast Oregon. Wildl. Soc. Bull. 13:130-134. LEE, K. H. 1991. Wetlands detection methods investigation. U.S. Environ. Prot. Agency Rep. 600/4-91/014. 73pp. LILLESAND, T. M., AND R. W. KIEFER. 1987. Remote sensing and image interpretation. John Wiley & Sons, New York, N.Y. 721pp. LUNETTA, R. S., R. G. CONGALTON, L. V. FENSTERMARKER, J. R. JENSEN, K. C. McGWIRE, AND L. R. TINNEY. 1991. Remote sensing and geographic information system data integration: error sources and research issues. Photogram. Eng. Remote Sens. 57:677-687. LYON, J. G. 1983. Landsat-derived land-cover classifications for locating potential kestrel nesting habitat. Photogram. Eng. Remote Sens. 49:245-250. MAUSEL, P. W., J. H. EVERITT, D. E. ESCOBAR, AND D. J. KING. 1992. Airborne videography: current status and future perspectives. Photogram. Eng. Remote Sens. 58:1189-1195. MAYER, K. E. 1984. A review of selected remote sensing and computer technologies applied to wildlife habitat inventories. Calif. Fish Game 70:101-112. McHARG, J. L. 1969. Design with nature. Doubleday and Company, Inc., Garden City, N.J. 197pp. McKEOWN, D. M., Jr. 1987. The role of artificial intelligence in the integration of remotely sensed data with geographic information systems. IEEE Trans. Geoscience Remote Sens. GE-25:330-348. MEISNER, B. N., AND P. A. ARKIN. 1984. The GOES precipitation index: large scale tropical rainfall estimates using infrared data. Proc. Conf. Hurricanes Tropical Meteorol. 15:203-206. MILLER, K. V., AND M. J. CONROY. 1990. SPOT satellite imagery for mapping Kirtland's warbler wintering habitat in the Bahamas. Wildl. Soc. Bull. 18:252-257. MURPHY, D. D., AND B. D. NOON. 1991. Coping with uncertainty in wildlife biology. J. Wildl. Manage. 55:773-782. NELSON, J. W., AND R. A. WISHART. 1988. Management of wetland complexes for waterfowl production: planning for the prairie habitat joint venture. Trans. North Am. Wildl. Nat. Resour. Conf. 53: 444-453. NEWCOMER, J. A., AND J. SZAJGIN. 1984. Accumulation of thematic map errors in digital overlay analysis. Am. Cartographer 11:58-62. NIELSEN, R. D. 1991. Digital soils data: 1:15,840 to 1:7,500,000 scale digital soils information from SSURGO, STATSGO, and NATSGO data bases. Pages 66-68 in K. K. Reay, ed. Proc. national conference integrated water information management. Virginia Polytechnic Inst. State Univ., Blacksburg. ORMSBY, J. P., AND R. S. LUNETTA. 1987. Whitetail deer food availability maps from Thematic Mapper data. Photogram. Eng. Remote Sens. 53:1081-1085. ------, AND G. A. SOFFEN. 1989. Foreword: special issue on the Earth Observing System (Eos). Inst. Electrical Electronics Eng. Trans. Geoscience Remote Sens. 27:107-108. PALMERIM, J. M. 1987. Automatic mapping of avian species habitat using satellite imagery. Oikos 52:59-68. PARKER, H. D., Editor. 1991. GIS world source book. GIS World, Inc., Ft. Collins, Colo. 597pp. PECH, R. P., R. D. GRAETZ, AND A. W. DAVIS. 1986. Reflectance modelling and the derivation of vegetation indices for an Australian semi-arid shrubland. Int. J. Remote Sens. 7:389-403. PETERSON, L., AND I. MATNEY. 1986. Data management. Pages 727-740 in A. Y. Cooperider, R. J. Boyd and H. R. Stuart, eds. Inventory and monitoring of wildlife habitat. U.S. Dep. Inter. Bur. Land Manage. Serv. Cent., Denver, Colo. PICKUP, G., AND V. H. CHEWINGS. 1988. Forecasting patterns of soil erosion in arid lands from Landsat MSS data. Int. J. Remote Sens. 9:69-84. PIWOWAR, J. M., AND E. F. LeDREW. 1990. Integrating spatial data: a user's perspective. Photogram. Eng. Remote Sens. 56:1497-1502. PLACE, J. L. 1985. Mapping of forested wetland: use of SEASAT radar images to complement conventional sources. Prof. Geogr. 37:463-469. RICHARDS, J. A. 1986. Remote sensing digital image analysis. Springer-Verlag, West Berlin. 281pp. ------, AND D. J. KELLY. 1984. On the concept of spectral class. Int. J. Remote Sens. 5:987-991. RIPPLE, W. J., G. A. BRADSHAW, AND T. A. SPIES. 1991. Measuring forest landscape patterns in the Cascade Range of Oregon, USA. Biol. Conserv. 57:73-88. ------, AND S. WANG. 1989. Quadtree data structures for geographic information systems. Can. J. Remote Sens. 15:172-176. RODCAY, G. 1991. GIS a "natural" for wildlife management. Pages 365-369 in H. D. Parker ed. GIS world source book. GIS World, Inc., Ft. Collins, Colo. SAXON, E. C., AND DUDZINSKI, M. L. 1984. Biological survey and reserve design by Landsat mapped ecoclines--a catastrophe theory approach. Aust. J. Ecol. 9:117-123. SCEPAN, J., F. DAVIS, AND L. L. BLUM. 1987. A geographic information system for managing California condor habitat. Proc. Int. Conf., Exhibits Workshops Geogr. Inf. Syst. 2:276-286. SCHOTT, J. R., C. SALVAGGIO, AND W. J. VOLCHOK. 1988. Radiometric scene normalization using pseudoinvariant features. Remote Sens. Environ. 26-1-16. SCHOWENGERDT, R. A. 1983. Techniques for image processing and classification in remote sensing. Academic Press, Inc., New York, N.Y. 249pp. SHAW, D. M., AND S. F. ATKINSON. 1990. An introduction to the use of geographic information systems for ornithological research. Condor 92:564-570. SHIH, E. H., AND R. A. SCHOWENGEPDT. 1983. Classification of and geomorphic surfaces using Landsat spectral and textural features. Photogram. Eng. Remote Sens. 49:337-347. SIDLE, L G., AND J. W. ZIEWITZ. 1990. Use of aerial videography in wildlife studies. Wildl. Soc. Bull. 18:56-62. SILVA, L. F. 1978. Radiation and instrumentation in remote sensing. Pages 21-135 in P. H. Swain and S. M. Davis, ed. Remote sensing: the quantitative approach. McGraw-Hill Book Co., New York, N.Y. SINGH, A. 1989. Digital change detection techniques using remotely-sensed data. Int. J. Remote Sens. 10:989-1003. SNYDER, J. P. 1987. Map projections--a working manual. U.S. Geol. Surv. Prof. Pap. 1395. 383pp. STEENHOF, K. 1982. Use of an automated geographic system by the Snake River Birds of Prey Research Project. Computer-Environ. Urban Syst. 7:245-251. STENBACK, J. M., C. B. TRAVLOS, R. H. BARRETT, AND R. G. CONGALTON. 1987. Application of remotely sensed digital data and a GIS in evaluating deer habitat suitability on the Tehama Deer winter range. Proc. Int. Conf., Exhibits Workshops Geogr. Inf. Syst. 2:440-445. STRAHLER, A. H., C. E. WOODCOCK, AND J. A. SMITH. 1986. On the nature of models in remote sensing. Remote Sens. Environ. 20:121-139. STRONG, L. L., D. S. GILMER, AND J. A. BRASS. 1991. Inventory of wintering geese with a multispectral scanner. J. Wildl. Manage. 55: 250-259. SWAIN, P. H., AND S. M. DAVIS, Editors. 1978. Remote sensing: the quantitative approach. McGraw-Flill Book Co., New York, N.Y. 396pp. TANRE, D., et al. 1990. Description of a computer code to simulate the satellite signal in the solar spectrum: the 5S code. InL J. Remote Sens. 11:659-668. TASSAN, S., AND B. STURM. 1986. An algorithm for the retrieval of sediment content in turbid coastal waters from CZCS data. Int. J. Remote Sens. 7:643-655. TOMLIN, C. D., AND J. K. BERRY. 1979. A mathematical structure for cartographic modeling in environmental analysis. Proc. Am Congr. Surv. Mapping 39:269-284. TUCKER, C. J., H. E. DREGNE, AND W. N. NEWCOMB. 1991. Expansion and contraction of the Sahara Desert from 1980 to 1990. Science 253:299-301. VITEK, J. D., S. J. WALSH, AND M. S. GREGORY. 1984. Accuracy in geographic information systems: an assessment of inherent and operational errors. Pecora 9:296-302. VONDEROHE, A. P., R. F. GUPDA, S. J. VENTURA, AND P. G. THUM. 1991. Introduction to local land information systems for Wisconsin's future. Wisc. State Cartographic Off., Madison. 59pp. WALKER, P. A. 1990. Modelling wildlife distributions using a geographic information system: kangaroos in relation to climate. J. Biogeogr. 17:279-289. WALKER, T. C., AND R. K. MILLER. 1990. Geographic information systems: an assessment of technology, applications, and products. Vol. L SEAI Tech. PubL, Madison, Ga. 166pp. WALKLETT, D. C. 1992. Investing in GIS and remote sensing holds the keys to understanding global change. Earth Observation Mag. l(l):70. WALSH, S. J., D. R. LIGHTFOOT, AND D. R. BUTLER. 1987. Recognition and assessment of error in geographic information systems. Photogram. Eng. Remote Sens. 53:1423-1430. WELCH, R., T. R. JORDAN, AND M. EHLERS. 1985. Comparative evaluations of the geodetic accuracy and cartographic potential of Landsat-4 and Landsat-5 Thematic Mapper image data. Photogram. Eng. Remote Sens. 51:1249-1262. WIEGAND, C. L., et al. 1986. Development of agrometeorological crop model inputs from remotely sensed information. IEEE Trans. Geoscience Remote Sens. GE-24:90-98. WOODCOCK, C. E., AND A. H. STRAHLER. 1987. The factor of scale in remote sensing. Remote Sens. Environ. 21:311-332. WORK, E. A., Jr., AND D. S. GILMER. 1976. Utilization of satellite data for inventorying prairie ponds and lakes. Photogram. Eng. Remote Sens. 42:685-694. YOUNG, T. N., J. R. EBY, H. L. ALLEN, M. J. HEWITT, III, AND K. R. DIXON. 1987. Wildlife habitat analysis using Landsat and radiotelemetry in a GIS with application to spotted owl preference for old growth. Proc. Int. Conf., Exhibits Workshops Geogr. Inf. Syst. 2:595-600.
Addresses and telephone numbers for the ESIC offices are shown below:
Reston-ESIC U.S. Geological Survey 507 National Center Reston, VA 22092 (703) 648-4000 |
Salt Lake City-ESIC 8105 Federal Bldg. 1245 S. State St. Salt Lake City, UT 84138 (801) 524-5652 |
Rolla-ESIC 1400 Independence Road Rolla, MO 65401 (314) 341-0851 |
Menlo Park-ESIC Building 3, MS 532 345 Middlefield Road Menlo Park, CA 94025 (415) 329-4309 |
Lakewood-ESIC Federal Center Box 25046, MS 504 Denver, CO 80225-0046 (303) 236-5829 |
Anchorage-ESIC 4230 University Drive Anchorage, AK 99508-4664 (907) 786-7011 |
Stennis Space Center-ESIC Building 3101 Stennis Space Center, MS 39529 (601) 688-3544 |
Washington, D.C.-ESIC Dept. of the Interior Bldg. 1849 C St., N.W., Rm. 2650 Washington, DC 20240 (202) 208-4047 |
Installation: Extract all files and open index.htm in a web browser.research.zip () -- Research and Management Techniques for Wildlife Habitats -- Geographic Information Systems