Land Use Primer:
What kind of models do scientists use to project future Land Use and/or Land Cover?
Several different approaches are used to project future Land Use and/or Land Cover. These can be divided into two broad categories: a) models that predict total land use change for a region, and b) models that predict Land Use for specific parcels or grid cells. If the analyst wants to know the total amount of land use change that will occur in a large region like a state, then the first type of model is appropriate. On the other hand, if the analyst wants to know where in the region land use change will occur, or to project what will happen at a specific place, then the second type of model is appropriate.
a.) Models to predict total Land Use
These models predict the total amount of land use or land cover change that will occur for a region (county, state, etc.), but do not predict where in the region that change will occur. For example, these models use equations or formulas to predict how much new development will occur, based on projections of population. A simple model could assume that for each new resident of a region, a fixed amount of land will be developed. A more complex model will use an equation to project land use change, based on initial population, population growth, available land, and other factors. Such a model has been developed for the CARA region, and is available here .
b) Models that predict Land Use for Specific Parcels or Grid Cells
This category of models can be subdivided in three broad categories:
1) Build out analysis, 2) Cellular Automata Models and 3) Agent based models.
b.1.) Build Out Analysis
A build out analysis projects future land use based on zoning regulations within the region of analysis. A build out analysis is used to estimate levels of residential, commercial and industrial uses that might happen in the future, given a zoning specification, and to visualize where that development will occur. Build out refers to a hypothetical future where all the available land has been developed at the maximum density allowed according to its zoned use.
b.2.) Cellular Models
In these models, a region is subdivided into square cells that form a regular grid. For each cell, an initial land use is measured. The analyst defines a set of rules that describe the probability that a cell will transition from one land use to another. These rules depend on the topography of the cell, its proximity via roads and highways to business centers, and the pattern of land use in neighboring cells. The transition probabilities can be estimated from historical data on land use. These models can be used to simulate the spatial pattern of land use change over time.
b.3.) Agent Based Models
In contrast to cellular models, where the focus is on the interaction among neighboring cells, Agent Based Models focus on the interactions among decision makers (individual land owners, firms or institutions). Most Agent Based Models combine cellular models, representing different land uses in the study area, with a model representing human behavior. For example, an agent based model might simulate the decisions made by a large number of farmers, each of whom is considering converting their farm to housing. Because the price of vacant land depends on how much of it is available, each farmer’s decision affects the profits of all the other farmers.