Ecological modeling in urban environments: predicting changes in biodiversity in response to future urban development - Urban Remote Sensing: Monitoring, Synthesis and Modeling in the Urban Environment - page 346

model. UrbanSim is designed to model and predict many different
land use and development variables, several of which are impor-
tant in predicting land cover change in the LCCM including:
number of residential units built, area of commercial or industrial
land use added, and occurrence of a development event each year
(Hepinstall, Alberti and Marzluff, 2008). Other variables such as
change in land use type, density of residential land use, density of
commercial land use and age of development can be derived from
these variables and are important drivers of land cover change.

Creek, Snohomish, and Monroe, and fertile agricultural lands
surrounding the Snohomish River.
The basic LCCM specification allows the user to develop (i.e.,
specify and parameterize) multinomial logit equations of land
cover change from empirically derived land cover and potential
explanatory variables that are believed to drive land cover using
the time-step interval of input data (Hepinstall, Alberti and
Marzluff, 2008). For my example here, I selected two dates of
land cover (1995 and 1999; 4-year time step) to derive estimates
of plausible land cover transitions (including no-change). I
developed multinomial equations for each transition using 65
candidate explanatory variables measuring landscape composi-
tion, configuration, contagion/distance, development intensity,
and indicator variables defining land use, parcel value, parcel size.
A full listing of the variables available for the Puget Sound imple-
mentation of the LCCM can be found in Table 2 of Hepinstall,
Alberti and Marzluff (2008). I used 1999 land cover as the base
land cover to predict to 2027 (the last date of available UrbanSim
predictions for the study area). The predictive accuracy of LCCM

25.2.3 Land cover changemodel

I used LCCM (Hepinstall, Alberti and Marzluff, 2008) applied
to the Central Puget Sound, Washington, to predict land cover
change for a 15 . 5 × 33 . 8 km swath of Snohomish county, Wash-
ington (Fig. 25.2; north of the city of Seattle). This area includes
a portion of the city of Everett as well as the towns of Mill

FIGURE 25.2 Land cover in 2002 for example 508 km 2 study area and for portions of Snohomish and King Counties,

Washington, USA.