Spatial Models and Modeling June 7, 2013 Institute of Space Technology, Karachi

Chapter 13: Spatial Models and Modeling

Model Description of reality For this class Static reproduction that represents basic shape and form of an object A conceptual description of key elements and processes in a system Sophisticated replica of objects, events, or processes For this class Restricted to computer-based models of spatial features It may be….. Spatial models

Computer Based Spatial Models Combination of GIS Computer programming languages Spatial and non-spatial analytical tools

Two Broad Classes Spatio-temporal Models Cartographic Models Dynamic in both space and time Example: analysis of oil after spill Cartographic Models Static models Involve application of spatial operations Example: buffer, interpolation, reclassification, overly Combine data from multiple data layers Analysis of oil after spill: current, winds, wave action, and physics of oil separation and evaporation on exposure to air might be combined in a model to predict the changing location of an oil slick. The actions of objects as they move across an environment may also be represented in a spatio-temporal model.

Cartographic Models Represent spatial features at a fixed point or points of time Most GIS models are cartographic models that are temporally static Provides information through a combination of spatial data sets, functions and operations Reclassification Overlay Interpolation Terrain analysis Buffering, Map algebra, etc. Temporally static: Although we may use them to analyze a change over a period of time e.g. vegetation change over a 10-year period. Each data layer represents the vegetation boundaries at a fixed point in time. Polygon boundaries for a given layer do not change. Data in base layers are mapped for a given period. These data are the basic for spatial operations that may create new data layers.

Example: Suitability Analysis Suitable park sites based on the proximity to Roads and Lakes and the absence of Wetlands Most common type of ~models

Flow Chart A graphic representation of the spatial data, operations and their sequence of use in a cartographic model Cartographic models may be represented by Flowcharts: Near road, near lake, not wetland. Data layers represented by rectangles, operations by ellipses, and sequences by arrow.

Application Land use planning Transportation route and corridor studies Design and development of water distribution systems Human disease spread Site selection Pollution response planning Endangered species preservation Useful to managers, public and policy makers.

Designing a Cartographic Model Spatial functions and operations are mixed and matched in cartographic models Variation in sequence of same operations will result in entirely different outputs With a small set of data layers and tools, a huge number of models can be created Usually produces a large number of ‘intermediate’ or temporary data layers that are not needed in final output or decision making

Designing the BEST Cartographic Model Selection of appropriate spatial tools and specification of their sequence Most important and often the most difficult process in ~.

Conti.. Designing a Cartographic Model Based on a set of criteria These criteria are usually defined in qualitative terms The slope must not be too steep Interpretation /translation of criteria into selection and sequence of spatial operations What is meant by “too steep” Need to be converted to specific, quantitative measures Sometime conversion from qualitative to quantitative criteria involves iterative process with repeated interaction between the analyst and manager/decision maker.

Example: Home Site Selection Problem: Ranking sites by suitability for home construction Criteria: Slopes should not be too steep Southern aspect is preferred to enhance solar warming Soils suitable for on-site septic systems Sites should be far enough from a main road to offer some privacy but not so far as to be isolated The area to be analyzed has steep terrain and is in a seasonally cold climate. Steep slopes may substantially increase costs There is a range of soil types in the study area, with a range of suitabilities for septic system installation. These criteria must be converted to more specific restrictions prior to the development and application in a cartographic model. Discussion between the analyst and decision maker: what type of classification required, what range of classes, what type of ranking, ordinal, ratio/interval? Measurement scale of results? Slope threshold beyond which construction is infeasible can be defined by studying how slopes affect construction costs.

First convert these criteria into more specific quantitative terms Check availability and quality of data Do the required data layers exist for the study area? Are spatial accuracies, spatial resolution and attributes appropriate for analysis What level of map generalization? If required data is not available Obtain or develop the required data OR Modify the goals Explicit ranking of the relative importance of different classes or types of criteria Map generalization: will inclusion of different soil types in a soil polygon lead to inappropriate results? Is the minimum mapping unit appropriate?

General Criteria Refined Criteria Slopes Not Too Steep Slope < 30 degrees Southern Aspect Preferred 90 < aspect < 270 Soils Suitable for Septic System Specified list of septic-suitable soil units Far enough from Road to Provide Privacy, But Not Isolated 300 meters < distance to road < 2000 meters

Ranking and Weighting Ranking: Assignment of relative values within the same layer How we rank a sandy soil vs. a silty soil in a soil layer ‘Weighting’ – assigning the relative values of different layers How we weight the values in an elevation layer vs. the values in a land use layer

Ranking Within Criteria Each Criterion in cartographic model is usually expressed by a data layer or ‘criterion layer’ Each criterion layer is a spatial representation of some constraint or selection condition Select site outside floodplain: Floodplain sites = 0, Upland sites = 1 select site outside floodplain may consist of a set of numbers in a layer identifying flood plain locations. Obtain floodplain maps Interpret codes in the map to delineate the most flood-prone areas. Rank areas based on the likelihood of flooding.

Discrete vs. Continuous Ranking Discrete: when input data are interpreted such that criterion data layer is a map of discrete value Soil = Good, Bad Slope = steep or acceptable Continuous: ranks vary along a scale Soils: Rated from 1 to 100 for construction suitability When there are clear discrete classes to be represented in criteria. Continuous: ranks or scores range over real or large integer interval e.g. 1 to 100 or 1 to 1000. Not necessarily a linear relationship but may be complex relationship between an input value and output ranking scores.

Discrete and Continuous ranking

Non-linear relationship. Mercury concentration in drinking water

At 1150 meters = rank 1 Criterion: desirable sites are greater than 300 but less than 2000 meters from road

Weighting Among Criteria Criteria combined in spatial analysis – in overlay or addition process How to weigh one layer over another? How important is slope relative to aspect? Will an optimum aspect offset a moderately steep site? The relative weightings given to each criterion will influence the result Different weights are likely to result in different suitability rankings Easy to define when importance of the various criteria expressed on a common scale – money Reducing all criteria to a common scale removes differential weighting among criteria Continuous ranking may be combined, often through a weighted addition process to generate a combined suitability score. Assign slopes a monetary values as it affects construction cost. Costs can be calculated Nuisance cost for noise and distance cost in lost time or travel might be quantifiable

Assigning Weights based on Importance Ranking

In our example all criteria are equally weighted and all criteria are binary. 3 base layers are needed Elevation, Soils and Roads. 3 branches in the flowchart. Reclassify: areas that do and don't meet the respective criteria.

Left most branch of the flowchart. Low elevation = black Low slopes = light shades Aspect= light to dark from 0 – 360 Reclassify based on the threshold values. Two main river systems in the study area: Running from west to east Running from south to north

Major roads: all divided and multilane roads.

Spatio-temporal Models Includes time driven processes within the framework of model Feature boundaries, point feature locations, and attribute variables may change within model run Complicated models Involves lots of computer coding.