to Introduction GIS
Lecture 3:
Part 1. Understanding Spatial Data Structures Part 2. An introduction to the Vector data model
Lecture by Austin Troy, University of Vermont
Introduction to GIS
Part 1. Understanding Spatial Data Structures
Introduction to GIS
Perception, Semantics, and Space
How do we deal with representing semantic constructions of spatial objects, like "mountain," "river," "street," "city," How about representing more conceptual semantic constructions like "temperature," "migration pattern," "traditional homeland," "habitat," "geographic range," etc? Answer: we have various data models which use different abstractions of reality
Entities and Fields
There are two general approaches for representing things in space:
Entities/ Objects: precise location and dimensions and discrete boundaries (remember, points are abstractions). Fields, or phenomena: a Cartesian coordinate system where values vary continuously and smoothly; these values exist everywhere but change over space
Introduction to GIS
Entities and Boundaries
There are two general types of boundaries, bona fide and fiat (D. Mark, B. Smith, A. Varzi) Pure bona fide boundaries represent real discontinuities in the world, like roads, faults, coastlines, power lines, rivers, islands, etc. Pure Fiat boundaries are a human cognitive or legal construction, based on a categorization, such as administrative unit, nation state, hemisphere Some have elements of both, like soil type areas
Introduction to GIS
Two major data models
Entity approach roughly corresponds with the vector model Field approach roughly corresponds with raster model Any geographic phenomenon can be represented with both, but one approach is usually better for a particular circumstance
Introduction to GIS
Raster
Spatial features modeled with grids, or pixels Cartesian grid whose cell size is constant Grids identified by row and column number Grid cells are usually square in shape Area of each cell defines the resolution Raster files store only one attribute, in the form of a "z" value, or grid code. Consider the contrary....
Introduction to GIS
Vector
Vector layers either represent:
Points (no dimensions) Lines, or "arcs" (1 dimension) or Areas, or "polygons" (2 or 3 dimensions)
Points are used to define lines and lines are used to scribe polygons Each point line or polygon is a "feature," with its own record and its own attributes
Introduction to GIS
Raster and Vector representations of the same terrain
Raster: great for surfaces
Vector: limited with surfaces
Introduction to GIS
Raster and Vector representations of the same land use
Introduction to GIS
Raster and Vector representations of the same land use: closer in
Introduction to GIS
Vector vs. Raster: bounding
Raster: bad with bounding
Vector: boundary precision
Introduction to GIS
Vector vs. Raster: Sample points Cancer rates across space
Introduction to GIS
In Arc View and Arc GIS, we can covert vector layers to grids, based on an attribute, or grids to vector layers
Moving between vector and raster
The disadvantage of vector to raster is that boundaries can be imprecise because of cell shape Each time you convert, you introduce more error too
Introduction to GIS
WHEN TO USE RASTER OR VECTOR???
Introduction to GIS
Raster data analysis is better for representing phenomena:
where boundaries are not precise that occur everywhere within a frame and can be expressed as continuous numeric values where change is gradual across space where the attribute of a cell is a function of the attributes of surrounding cells
Introduction to GIS
Raster technical advantages :
Simple file structure Simple overlay operations Small, uniform unit of analysis
Raster technical disadvantages :
Big file size, especially for fine-grained data Difficult and error-prone reprojections Square pixels are unrealistic
Introduction to GIS
Vector analysis is better :
Where there are definable regions Where the relative position of objects is important Where precise boundary definition is needed Where multiple attributes are being analyzed for a given spatial object For modeling of routes and networks For modeling regions where multiple overlapping attributes are involved EG: units with man-made boundaries (cities, zip codes, blocks), roads, rivers
Introduction to GIS
Vector technical advantages :
Smaller file size (in general) More graphically interpretable Allows for topology (see further on)
Vector technical disadvantages :
Complicated file structure Minimum mapping units are inconsistent between overlapping layers
Introduction to GIS
Specific Vector Usages
All legal and administrative boundaries (zip codes, states, property lines, land ownership) Building footprints and 3-D models Roads Bedrock geology Pipelines, power lines, sewer lines Flight paths and transportation routes Coastlines
Introduction to GIS
Specific Raster Usages
Terrain modeling where micro-locational variability is present and matters Groundwater modeling, where surface flow outside of channels is important Representation of slope and aspect Representations of distance and proximity to features Spatial representation of probabilities (logit) Modeling phenomena in nature with continuous spatial variability and numeric attributes, like soil moisture, depth to bedrock, percent canopy cover, vegetative greenness index, species richness index
Introduction to GIS
Tossups
In many cases, though, the choice between raster and vector may not be so clear. Often it depends on the application The following are some examples where you could go either way:
Introduction to GIS
Soil
Soil type: Vector
Soil types are meant to represent discrete and homogeneous areas and are qualitative. There is no "slight gradation" between soil types like with pH
Soil pH: raster
pH is numeric, not categorical, and that number may vary slightly within a single soil type polygon If pH were turned into categories, like High, Medium and Low, vector might be better
Introduction to GIS
Rivers
Most people think of a river as a discretely bounded entity, hence vector What about where the river size fluctuates seasonally, e.g. desert rivers? Or where the location of the river bed changes slowly and gradually the over years Or where the river becomes delta, and the distinction between "river" and "swamp" becomes fuzzy? Or where the river has a certain probability of flowing or being dry at any given location and time
Introduction to GIS
Rivers
Depends on the type of analysis being done With vector can do network modeling of stream and river system, but only in the arcs
Vector stream model can take advantage of topologically enabled analysis tools
With raster, can do surface flow modeling
More realistic, because when it rains water flows everywhere, not just in channels, shows accumulation Think of every piece of land as mini stream channel
Introduction to GIS
Vegetation Mapping
Vector works well for modeling vegetation stand type where categories are broad, e.g. mixed conifer, deciduous hardwood Raster works better where there is micro-locational heterogeneity in species distribution Raster also works better for representing ecotones, or edges between two stands The more specific and variable the classification, the more likely the raster approach will be needed
Introduction to GIS
Part 2. Brief Intro to the Vector Data Model
Introduction to GIS
Intro to Vector
Recall that there are three basic "feature" or "object" types in the vector data type:
Point Arc Polyons
In general a given layer holds a given feature type (e.g. "roads" is a line layer, "counties" is a polygon layer, "weather stations" is point)
Introduction to GIS
Intro to Vector
A point layer only consists of a bunch of (x,y) coordiantes In a line (arc) layer, points define lines In a polygon layer, lines define areas Hence each level of vector features builds on the last
Introduction to GIS
Intro to Vector
Each point has a unique location 2 points define a line segment One or several line segments define an arc The endpoints of an arc are "nodes The angle points are "vertices" (sing. Vertex) The feature is the arc, not the line Two arcs meet at the nodes
Introduction to GIS
Intro to Vector
Several arcs can scribe a polygon Polygons are closed regions whose boundaries are made up of line segments connected at many angles. Polygons generally define an area of homogenous phenomena (e.g. forest stand, building, zip code, lake) These phenomena can be described by one or more stored attributes
Introduction to GIS
Vector Representation:lines
Ring: this is a series of line segments (a string) that close upon each other It is NOT a polygon!! The computer does not know that the area inside "belongs" to that object
Introduction to GIS
Vector Representation:lines
A polygon is encoded differently, because the computer "knows" that the areas within those arcs "belongs" to that polygon, while it does not with a ring
Introduction to GIS
Vector: Topology
Topology: spatial relationships between objects are encoded; the spatial location of each point, line and polygon is defined in relation to every other point, line and polygon Topology allows for behaviors of objects in relation to other objects to be defined Topology allows for powerful analysis tools and can significantly reduce error and increase quality Vector files in ARC INFO are topologically encoded. Arc GIS 8.3 geodatabases will be as well. Currently geodatabases are partially topological
Introduction to GIS
Vector Topology: purpose
One of the most important functions of topology is ensuring data quality and "logical consistency" When you bring in line and polygon data from external sources, you will often find errors such as lines (arcs) that dangle or overshoot, polygons that don't close, adjacent polygons that show up as not sharing a border (we'll return to this later in the semester)
Introduction to GIS
Vector ...