Document 10679219

advertisement
R F Y ~ I NM~A P P I N E
joseph K. Berry
Logic and Extent Elevate
suitability Models t o N e w Levels
P
revious "Beyond Mapping" disc o n s t r a i n e d areas t o generate the
set of terrain considerations for an overoverall suitability map shown in Figcuss~onson s~utabilityrnodel~ng
all habitat map shown in Figure 2.
used wildlife hahitat mapping to
ure I . Thi? recult contains continimui
Note that a large part of the model's
illustrate the development of progreshabitat values-considerably more
strength or weakness is established in
information than simply the spatial
sively m o r e powerful m o d e l i n g
Step 1) Calibrate Criteria Maps. As
approaches: hinary, rankinp,, permutac o i n c i d e n c e o f d i s c r e t e areas o f
much as possible, the identification of
tion and rating models. A l l four
goodlbad classiiications.
map criteria needs to reflect good sciapproaches used the same set of basic
ence andlor expel[ opinion to caplure
criteria: Hugag preierence for gentle
Separate Sub-Modds
factors that are Important and easily
slopes, southerly aspects and lower clcAlthough the processing apploach is
measurable. Similarly, calibrating the
v;ltioins (see Figure 1). The discusbions
a11 i r n p u r t d n t c u r i s ~ d e r a t ~ u r the
i,
rnd[~siritu the 1-9 p r e i e r e r i ~ erdnge
f o c ~ i s e do n d i f f e r e n c e s
needs to capture realistic reli n h o w the processing
l u ~ . ~ . s r ~ . ~ ~ ~ n n . . ~ o s o ~ . . o ~ l m i ative values and not whimsi..p.rt.
I"" ,on.r.**."an.
butrlmn
tdkeb place.
me v
.urnV. mep."l.t.r
cal or biased assignments.
I n the case o f a b i n a r y
model, each cons~derat~on
,
is treated as good or bad
Step 2) Combine Calibrated
and results in a habitat map
Maps also r e q ~ ~ i r econsiders
that identifies just good and
able understanding of the sysbad hahitar areas. A ranking
tem being modeled. A simple
model, howeve), uses the
average of the cal~bratedmap
Callbratc Cllhrla Ma r. ~ r o h i r , or ~r ~ map
a
cnmr!eaa,ea8anod.rnrtm,r~mivw&drne vnyed.
same goodlbad criteria, but
layers assumes that all c r l mna ComblnsC~lbratrdM~s.~hccallb~toddma~s.lara
it identifies the number of
teria are equally important.
'"'""d"b.noMr~!ini#abiilNmrpdNbb,
good
,or eat,,
The right inset i n Figure 3
~~l~~~cannra~nn.~~~fw~~r~mdhimmatd~v
map location, and higher
shows the habitat results for
.*rwn,wlhcmr nbgdoand",dd~,m.
..,--D
,
values indicate increasingly
, "cxpert thinking" that Hugags
I l i g h r r l i d l ~ i l d tranking. A
are "1 0 times more concerned
Figure 1. A graphic describes thc modcl logic far basic Hugag habitat
perniutation m o d e l prosuitability mapping.
about slope, iorest and water
con5iderations than they are
vidcs additional information
by identifying the unique combinations
model's logic and extent can be more
about aspect, elevation and roads considerations."
o f good and bad factors occurring at
important in determining accuracy. In
each location.
p r a c t i c a l applications, the habitat
model likely w o u l d consider many
more factors than simply lerrain conGood Ratings
A idling model is the most p o w e r i i ~ l figiiratlon.
approach, and it breaks the goodlbad
Figure 2 shows a flowchart o i the
exlended rnodel logic used to evaluate
dichotomy into a preference gradient
most often expressed as 1 - very bad
the additional criteria that "Hugags
to Y = very good. For cxarnplc, the
would prcicr to b c in iorcstcd areas''
preierelice lur gentle slupeb ( S Pref in
(Furebt rndp), "Hugdg, would prefer to
he near water" ( p r o x i m i t y t o Water
ligure 1 ) was assigned as 1 (very had)
= greater than 40 percent; ? = 10-40
map), and "Hugags would prefer to he
percent; 5 = 20-30 percent; 7 = 10-20
far i r o m roads" (proximity to Roads
K. BERRY is a principal in
Berry &Associates, consultants in GI
pel-cent; a n d 9 (very g o o d ) = 0-1 0
map). in suitability modeling, these contechnology, 2000 S. College Ave., Sui
percent.
siderations are treated as separate sub300, Fort Collins. CO 80525; e-mail
models to derive the necessary criteria
In a similar manner, categories for
jkber'y@du.edu,Web:
and then calibrated on the 1 9 preferaspect and elevation are calibrated
ence scale and averaged with the basic
and then averaged and masked for
-
The procedure for determining relative importance involves computing
the weighted average of the six map
layers and is analogous to a professor
grading some exams more importantly
than others in determining a class
grade. in this particular example, map
values correspond to student grades
on each exam; each student i s represented as a grid cell on the map-kind
of like their desk seats in the classroom
floor plan.
Note in Figure 3 the similarities and
differences in the maps induced by the
additional criteria (Extended) and relative weighting of map layers Weighted).
Provided that expert opinion is sound,
the weighted map on the right would
be considered the most accurate representation of Hugag habitat.
Alw note that calibratingand weighting are critical steps in suitability modeling. Procedures such as Delphi and
AHP can be used to derive these fact o n in a quantitative, objective, consistent and comprehensive manner (see
"Author's Note," below). In addition,
factors can be changed to reflect different assumption scenarios analogous
to "what iP questions applied to traditional spreadsheet analysis.
From this perspective, it's how the
suitability maps change that becomes- I
information about a project area's sensitivity to the interplay of criteria, calibrations and weights.This takes us
w e l l "beyond mapping" and into
assessing the spatial relationships
w i t h i n a system and their logical
expression within a CIS.
As CIS technology matures, the focus
is shifting from access of static map
products depicting physical features
for navigation and inventory t o a
dynamic environment that enables
"thinking with a stack of maps" within
decision-making contexts.
Author's Note: For a discussion of
Delphi andAHPprocedures as well as
a fbwerfbint slide sec insfructions and
free evaluation soiiware for ciassroom
or individual "hands-on" experience in
suitability modeling, visit the Web a t
-
1
:
'
http://www.innovativegis.com/basis,
selecting "Column Supplements." Selm
A e ! ! b 4 a @ A n a -~ ~ ~ ~
compilation ofprevious "Beyond
Mapping" columns.
9W
1I
',
8
'
8 ' 1
Fieure 2. Extended
Hugag preferencesfor
beiq in b-, mu
water and far fmm
Download