Geography Reference
In-Depth Information
WRF model. Keeping lateral boundary conditions and all other parameters of WRF
model unchanged, processed land use/cover underlying surface data are input into
WRF model to simulate regional climate change in Wuhan Metropolitan. After
these processes, regional climate change under different urbanization patterns in
Wuhan Metropolitan can be figured out. Through scenario analysis of regional
climate change under different urbanization patterns, optimal urbanization patterns
which can mitigate climate change in Wuhan Metropolitan can be worked out.
6.3.2.2 Partitioned and Asynchronous Cellular Automata Model
Cellular automata model has the ability to simulate spatial and temporal evolution
of complex systems. The ''bottom-up'' research idea fully reflects the concept that
local individual behaviors of complex systems will produce global and orderly
pattern. Therefore, cellular automata model has natural advantages in urban land
expansion simulation (Li et al.
2007
). However, most cellular automata models
have some limitations in simulating urban land expansion. On one hand, it ignores
spatial heterogeneities existing in urban land expansion and its influencing factors
to use the unified cellular transformations rules for all cells in urban land
expansion simulation. On the other hand, it ignores the spatial heterogeneities of
urban land expansion speed to employ same evolving speed for all cells. Both of
them become the barrier of simulation accuracy improving for cellular automata
model (Ke and Bian
2010
). In this model, spatial data mining methods are
employed to dig out partitions for cellular automata model and separately cellular
transformation rules for each partition are dig out by Decision Tree Algorithm.
Transformation rules for each partition are made up of three sections: trans-
formation probability for each partition, unit constraints, and neighborhood
development density (Li et al.
2007
). It can be showed by the following formula.
P
t
d
;
ij
¼½1
þð
lnc
Þ
a
P
g
con
ð
s
ij
Þ
X
ij
where p
t
d
;
ij
is transformation probability for each partition, c is random number
ranging from 0 to 1, a is the parameter which controlling random variable effect
level. It is an integer which ranges from 1 to 10, P
g
is transformation probability
which decided by urban expansion influencing factors, con
ð
s
ij
Þ
is constraint con-
dition of unit, X
ij
is neighborhood function which means effect of neighborhood to
cellular automata transformation probability.
In these above parameters, c and a is introduced to add random factors in
cellular automata model to imitate effect and intervention of all kinds of uncertain
factors in land use processes. P
g
is obtained from geographical phenomenon
change data and is related to impact factors by the method of spatial data mining. It
remains unchanged in the process of the whole simulation. X
ij
is a very important
factor. It changes over time and can be performed by the following formula.
