nbul 2004 
n values 
map the 
yan scene 
ngle-date 
| process 
stratified 
dentify a 
lation of 
imate the 
deled by 
ction is 
between 
il photos 
growing 
iis paper 
rate in a 
of fuzzy 
hange is 
la with 
:curately 
om areas 
ogic that 
h values 
sets are 
between 
1975). 
clement 
ted by a 
set to a 
smaller 
ual date, 
way by 
| ] from 
ractional 
d cover 
sess the 
odology 
ge (high 
change 
icept of 
l| (or S- 
ctive in 
99). The 
"Zy sets 
ncrease, 
ecrease. 
sted and 
| of the 
change- 
member 
the five 
ixel-by- 
cal map 
rmation 
egorical 
perty of 
present 
discrete areas of relatively homogeneous environmental 
conditions, the definition of which is artificially imposed 
according to a phenomenon of interest and only meaningful 
when referenced to a particular scale. For example, the urban 
landscape of Los Angeles can be described as a mosaic of 
census tracts. The census tract in this case can be thought of as 
a patch that is relatively homogeneous in terms of social and 
physical characteristics. Similarly, at a larger scale, a census 
tract can be viewed as a mosaic (or landscape) of its own, 
consisting of smaller patches of land cover classes represented 
by a collection of pixels in a remotely sensed image. While 
individual pixels (the construction blocks of patches) possess 
uniform spatial characteristics (e.g., identical size, perimeter, 
and shape), the aggregation of these pixels provides a rich set of 
properties. These properties depend on whether the pixels are 
aggregated over a single land cover class (patch) or multiple 
classes, and whether the aggregation is considered within a 
specified census tract. Landscape metrics make use of these 
properties to reveal the spatial character and distribution of 
patches, and thus to quantify landscape patterns (O'Neill et al., 
1988). 
The proposed methodology uses a subset of landscape metrics 
as a way of quantifying the configuration and composition of 
spatial variation of land cover fraction changes produced by 
MESMA. Calculating these metrics at the census tract level 
(ie., each tract is considered as a collection of land cover 
patches) provides an additional means of establishing and 
testing the link between both social and physical drivers of 
urban land cover change and sustainable policies implemented 
at the local level. The temporal differences in land cover 
fractions produced by MESMA are typically represented in 
terms of the change of the areal percentage occupied by a 
fractional class of land cover within a pixel. However, 
landscape metrics operate on the assumption that individual 
patches are homogeneous at the patch level. Therefore, before 
landscape metrics can be applied, fractional differences have to 
be reclassified such that each pixel within any census tract 
corresponds to one, and only one, magnitude of land cover 
change (i.e. higher increase, lower decrease, etc). To do so, 
each individual pixel in the fuzzified layers of change produced 
in the previous phase is screened in terms of whether or not a it 
meets a threshold of the degree of membership (e.g. 0.7 in the 
case study presented herein) in a certain magnitude of change. 
If a pixel value (i.e. the degree of membership in a specific 
magnitude class) is equal to or greater than this threshold, the 
pixel is classified under this magnitude of change. Thus, there 
may exist up to five classes of change magnitude (higher 
increase, lower increase, no change, lower decrease, and higher 
decrease) calculated for each class within any census tract. 
The next step is to select a subset of landscape metrics that best 
quantifies the ecological patterns of land cover change within 
the census tracts in a given study area. The ecological patterns 
are quantified in terms of the configuration of patches of pixels 
of a given magnitude of change in a given land cover class 
within a census tract (i.e. class level metrics). Table 1 shows the 
subsets of metrics that have been used in the case study 
presented in this paper. As shown in the table, there are metrics 
that essentially measure different properties in the same way 
and at the same level such as PLADJ and AI. Thus, we should 
expect that some of the measures resulting from these metrics 
would be highly correlated with each other. This redundancy is 
deemed important in the proposed methodology because each 
metric points to a slightly different aspect of the spatial 
structure of urban places. In the case study presented below, the 
calculation of all these metrics was done through a software 
505 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004 
package called FRAGSTATS (version 3), designed to compute 
a wide variety of landscape metrics for categorical map 
patterns. 
Table 1: Description of landscape metrics applied at the 
land cover class level within a census tract 
: Class Metrics 
Metric Property Measured 
  
PD - Patch density Areal composition 
  
LPI - Largest patch index Areal composition 
PAFRAC - Perimeter-Area Fractal | Shape complexity 
Dimension 
  
ggregation of land cover 
PLADJ - Percentage of Like Degree of a 
Adjacencies class 
  
AL Indéx of Aggregation Degree of aggregation of land cover 
dam class 
IJI - Interspersion and Juxtaposition | Degree of interspersion or intermixing 
Index of land cover class 
DIVISION Diversity of land cover class 
Physical connectedness of the land 
cover class 
  
  
  
COHESION 
  
  
3. TESTING THE METHODOLOGY IN LOS ANGELES 
3.1 Study Area and Data 
The study area used to test the proposed methodology is Los 
Angeles County, California, a dynamic and data-rich region that 
has undergone dramatic changes towards sustainable policies. 
In the 1980s, urban planners and policy makers began to 
formulate broader policies for sustainability (e.g. the 1989 Air 
Quality Management Plan) and implement tangible processes of 
technological and procedural change designed to allow cities in 
this region to use less energy and materials, to pollute less, and 
to create more durable social relationships with nature (Keil and 
Desfor, 2003). Nurtured by the region’s strong growth control 
movement in residential neighborhoods, Los Angeles’ planning 
and policies in the late 1990s began to reflect a leaning towards 
vertical growth and urban greenness (Pincetl et al, 2003). 
Projects have been conducted to transform small interstitial 
spaces into greened open spaces, particularly in parts of the 
region that were rated park poor. Although it is too early for 
one to assess the degree to which these sustainable policies and 
efforts have been successful in such a complex and diverse 
region as Los Angeles County, there are indicators that can be 
used to assess the progress. In the present case study, the 
proposed methodology has been used to extract physical 
indicators associated with neighborhood densification in the 
regions to quantify the ecological patterns of change associated 
with this process. 
The data utilized in the application of proposed methodology 
included subsets (3113 lines X 4801 samples) from two Landsat 
TM and ETM+ images acquired in September 1990 and July 
2000 respectively (path 41, row 36). Both images have 0% 
cloud cover. In addition to the multispectral images, the case 
study utilized a set of 1.0 m spatial resolution aerial photos to 
aid in the validation of the resultant endmember fractions. 
These photos represented a 1:12 000 color reproduction of high 
resolution visible color aerial photography acquired in late 1993 
by I. K. Curtis Services, Inc., from an altitude of 2743.2 m 
using an RC10 aerial survey camera. 
3.2 Results from Applying MESMA 
Two individual subsets of image endmembers were selected 
independently for two dates, one for each image. These subsets 
were chosen according to a modified VIS model: vegetation 
(V), impervious surface (1), bare soil (S), and, the modification, 
water or shade. The latter endmember type, shade, was used 
here as a proxy for building heights based on the assumption