A data model is a set of rules to identify and symbolize features of the real world (called entities) into digitally and logically represented spatial objects consisting of the attributes and the geom- etry. The attributes are characterized by thematic or semantic structures, while the geometry is represented by geometric-topological structures.
There are two basic categories of data involved: spatial and attribute. Spatial data include the locations of features, such as the latitude/longitude of dams, gauging stations, etc. Spatial data are often represented as objects such as points, lines, and polygons, which are used to represent the differing types of features. For example, the location of a well is a point, a stream path is a line (or vector), and a basin boundary is a polygon. Spatial data may also be represented as fields or images, such as might be derived from satellite imagery. Attribute data include numerical and character-type data that characterize the resource. Geographic data are characterized by a series of attribute and behavioral values that define their spatial (location), graphical, textual, and numeric dimensions (Worboys 1995). These include identifiers, names, and physical capacities for features of the water resources system, such as dams and reservoirs, pipelines, drainage basins, pumps, and turbines. Time-series data on river flows, reservoir releases, pumping rates, and other time variables are also managed in the attribute database.
3.3.2 rastersanD vectors
The two major types of geometric data models are raster and vector, as shown in Figure 3.5. These categories are sometimes related to how people think about things (objects) versus how they see
Point Objects Line Objects Area Objects Raster Vector A B C A A A A A A A A A B B B A A A A A A A A B B B B A A A A A A A B B B B B A A A A A A A B B B B B A A A A A A B B B B B B A A A A A A C B B B B B A A A A A C C C B B B B A A A A C C C C C B B B A A A A C C C C C C B B A A A C C C C C C C C B A A A C C C C C C C C C A A C C C C C C C C C C
FiGure 3.5 Raster and vector data models; the raster data model represents point, lines, and areas as col-
the world (field). The first general data model is the raster or field model, which is often defined on an x,y grid with each cell, or pixel (i.e., picture element), specifying the value of the data. The uniform grid is also referred to as a raster data structure. A field model is often used to represent variables that vary continuously over a region, such as temperature, rainfall, and elevation. In a raster, a point is given by a point identifier, its coordinates (i,j), and the attribute value. A line is given by a line identifier, a series of coordinates forming the line, and the attributes. A raster area segment is given by an area identifier, a group of coordinates forming the area, and the attributes.
A vector data structure maps regions into polygons, lines into polylines, and points into points (Figure 3.6). The vector model is used to represent spatial entities such as river and pipe networks and facilities (e.g., fire hydrants). These data often derive from land surveying and CAD (computer- aided design) drawings, as well as conversion of imagery data through processing. A key aspect of the vector data model is that the topology of relationships between features be established. Topology refers to the relationships or connectivity between spatial objects. The geometry of a point is given by two-dimensional (2-D) coordinates (x,y), while line, string, and area are given by a series of point coordinates.
Node
• : an intersection of more than two lines or strings, or the start and end point of an arc with a node number
Arc
• : a line or a string with an arc number, start and end node numbers, or left- and right- neighbored polygons
Polygon
• : an area with a polygon number, or a series of arcs that form the area with direc- tionality (e.g., positive in clockwise order)
a 1 5 1 4 1 6 3 2 6 6 b 12 13 11 13 11 12 12 11 d c Chain Geometry Chain
Chains Chain From To Left Polygon Left Polygon
Node
Start Coordinates End Polygon
Polygon Topology
Chain (Xa, Ya)
(Xb, Yb) (Xi, Yi) (Xi, Yi) . . . (Xj, Yj)
(Xj, Yj)
(Xb, Yb) (Xd, Yd)
Node Topology Chain Topology
. . . . . . . . . . . a b 0 0 b d 1, 2, 5 1, –5, –4 –1, 2, 6 –2, 3, 5 –3, 4, –6 4, –5, 3 6, –3, –2 a b c d . . . . . . . . .. ..
A vector database must be topological before it can be reliably used for spatial queries or to support modeling. The vector data model is often preferred for sophisticated modeling projects because the objects can be manipulated in a logical manner. For example, a river network can be searched up or down.
Vector and raster structures both have advantages and disadvantages. Each approach tends to work best in situations where the spatial information is to be treated in a manner that closely resembles the data structure. Vector structures are generally well suited to represent networks, connected objects, and features that are defined by distinct boundaries. Raster structures work best when the attributes they represent are continuously and smoothly varying in space. The finer the grid used, the more geographic specificity there will be in the data matrix. However, the loca- tion precision for a map feature in the raster model is limited by the cell resolution. Generally, the advantages of vector structures include a good representation of point, line, and polygon features (streams, lakes, drainage divides, etc.); compactness of data storage; accurate graphics; relational representation of objects; and the capability of updating, modifying, and generalizing graphics and attributes. Disadvantages include complexity of data structure, elaborate processing for over- lays and simulation, comparatively expensive technology and data, and difficulty in representing spatially varying attributes (Meijerink et al. 1994). The advantages of raster structures include the simplicity of the data structure, easy overlay and spatial analysis, availability of data, and comparatively cheap technology. Disadvantages include inefficient use of computer storage; inac- curacies in point, line, and area definitions; difficulty in establishing networks and topology; and unattractive visualization in low-resolution rasters (Meijerink et al. 1994). Table 3.1 summarizes the advantages and disadvantages of each data structure; Burrough and McDonnell (1998) provide a more comprehensive review.
The decision to use an object or field-data model is based on the requirements of the application, on tradition, and on the original source data. However, advances in GIS database technologies make the distinctions between the vector and raster data structures less compelling. Data-storage capaci- ties continue to increase, which makes the issue of raster storage volume less critical, although the huge volumes of image data from various sensors still require difficult decisions on storage priorities. For many applications, the increasing resolution of imaging systems provides adequate accuracy for locating land features. Additionally, the long-term growth in computer capacities continues to mitigate processing issues.
Representations of the spatially varying characteristics of a watershed are often contained in a series of raster data sets, which may include such information as elevation, soil properties, and taBle 3.1
vector versus raster data structures
data structure advantages disadvantages
Vector Compact storage
Topology is explicit and powerful Represents entities
Integrates with DBMS
Easy coordination of transformations Accurate graphics at all scales
Complex data model
Overlay operations complex and CPU intensive Requires attributes for entities
Display and plotting can be time consuming Display accuracy may be misleading
Raster Simple data structure
Easy manipulation of attribute values Many analysis functions
Easy math modeling Many data forms available Economical technology
Large data volumes
Accuracy limited by cell resolution Limited attribute representations Graphics coarse with zoom Coordinate transforms difficult
land use. However, the spatial resolution of raster elevation data may make it difficult to accurately locate smaller streams and other bodies of water using only the elevation data. The finest spatial resolution of widely available elevation data is 10 m; at such a scale, many smaller streams simply may not be identified, and drainage boundaries may be inaccurately determined. Vector stream data can help to eliminate some of these difficulties. Sources of raster and vector data for land surface and stream paths are described below.