|
GIS MODELLING | |
? Simplified representation – phenomenon,
system
? Model examples
? USGS quad map
? Vector/raster data models to represent spatial
features
? Relational database tp represent a database
system
? Model – helps to better understand a
phenomenon/system by retaining significant
features or relationships of reality
Use of GIS in the process of building
models with spatial data
? Models of spatial data – use spatially
explicit data
? Use of GIS in modeling, rather than the
models
Integrate different data sources
? Maps, DEMs, GPS data, images, tables
? Data displayed together & dynamically
linked
? Models are vector-based or rasterbased
? Depends on nature of the model, data
sources and the computing algorithm
Raster-based models – use when …
? Spatial phenomenon to be modeled
varies continuously over the space
? Soil erosion, snow accumulation
? Satellite images & DEMs constitute a
major part of the input
? Modeling involves intense & complex
computation
Vector-based models – use when …
? Travel demand modeling –requires use
of topology-based road network
? Recommended for spatial phenomena
that involves well-defined locations and
spheres
Integrate both types of data
? Raster-to-vector & vector-raster conversion
algorithms are available
? Depends on efficiency & expected results,
rather than format of original data
? Ex: vector-based model requires precipitation
map as input
? easier to prepare a precipitation grid through
spatial interpolation from known point data
? derive a map showing precipitation zones from the
grid
GIS
modeling takes place …
? In a GIS
? GIS packages have analytical functions for
modeling
? ArcView – ModelBuilder
? GIS not as efficient as a statistical
analysis package
? GIS can’t perform dynamic simulation
efficiently
Linking
Linking a GIS to other computer programs
? Loose coupling
? Data exported from GIS/imported to statistical
package
? Results returned to GIS for data visualization or
display
? Tight coupling
? Gives a GIS & other programs a common user
interface – program resides outside the GIS
? Embedded system
? bundles the GIS & other programs with shared
memory and a common menu interface
Types of models
? Binary models
? Index models
? Regression models
? Process models
Binary model
? Uses logical expressions to select map
features from a composite map or
multiple grids
? Output of binary model is in binary
format
? 1 – True for map features that satisfy the
logical expression
? 0 – False for map features that do not
Index model
? Uses the index value calculated from a
composite map or multiple grids to
produce a ranked map
? Evaluate selected variables at 2 levels
? Relative importance of each variable
evaluated against other variables (weight)
? Observed values of each variable evaluated
and given numeric scores
? Grouped into classes – each class given a score
Index model expressed as a linear equation
? Index variable on left – selected variables and
their weights on the right
? Index value calculated by summing the
weighted numeric scores from each variable
? Can assign lowest score, highest score, most
frequent score
? Normalize the index values so scaled from 0
to 1
? (x – lowest index value) / (range of index values)
Commonly used in suitability analysis
(habitat) and vulnerability analysis
? Examples:
? DRASTIC model – EPA
? DRASTIC: Depth to water, net Recharge,
Aquifer media, Soil media, Topography, Impact
of the vadose zone, hydraulic Conductivity
? Rate each parameter, multiply the rating by a
weight, sum the total score
Regression models
? Relates a dependent variable to a # of
independent variables in an equation,
use for prediction or estimation
? Use map overlay operations in a GIS to
combine all the independent variables
needed for analysis
Linear regression
? Dependent & independent variables are all
numeric variables
? Logistic regression
? Dependent variable is a binary phenomena
– presence or absence
? Independent variables are categorical or
numeric variables
Process models
? Integrates existing knowledge about
environmental processes in the real world
into a set of relationships & equations for
quantifying the processes (Beck et al. 1993)
? Offers an explanation inherent in the
proposed processes
? Typically raster-based
? Simple process may be prepared in GIS –
normally linked to other computer programs
for complex & dynamic analysis
GIS modellind
Project example
A distributed hydrological modeling system linking GIS and hydrological models |