Hunan Province, China. It is also one of biggest and rapidly de- 
veloping cities in Central China. Therefore, this study choses 
the core area of Changsha-Zhuzhou-Xiangtan metropolitan as 
the case study area. Using land use and land cover data derived 
from remotely sensed data from 1993 to 2006, the land use and 
land cover changes of four periods are calculated using GIS 
analyses. The dynamic changes of urban landscape from 1993 
to 2006 in Changsha-Zhuzhou-Xiangtan metropolitan areas is 
then quantified and expressed using landscape expansion index 
(LED, which contains information of the formation processes of 
landscape pattern. 
2. DATA AND METHODS 
This section briefly describes the data used in the study and the 
calculation of LEI, and discusses why LEI is better than the tra- 
ditional methods for analyzing dynamic changes of multi-time 
landscape patterns. 
2.1 Land Use and Land Cover Data 
Landsat TM/ETM- images in 1993, 1996, 1998, 2001 and 2006 
were selected for this study. For detecting urban expansion, the 
images were rectified to a Transverse Mercator coordinate sys- 
tem based on 1:5,000 scale topographic maps, and were re- 
sampled to a pixel size of 30m for all bands using the nearest 
neighbour algorithm. A root mean square error of less than 0.5 
pixels was obtained for all of the rectifications. 
Support Vector Machine (SVM) method was used for land use 
and land cover classification in this study. The images were 
classified in five land use and land cover types, that is, built-up, 
forest, cropland, bare land, and water. The overall accuracy and 
kappa variance were calculated based on an error matrix for 
each classified map. The overall accuracy and kappa coefficient 
of land use and land cover map was about 85% and 0.81, re- 
spectively. Land use and land cover data derived from remotely 
sensed data have a reasonable level of accuracy and are suffi- 
cient for urban landscape analysis and change detection. 
Using land use and land cover data from 1993 to 2006, the 
changes of land use and land cover in the study area can be 
identified in four different periods: 1993 to 1996, 1996 to 1998, 
1998 to 2001 and 2001 to 2006. 
2.2 Landscape Expansion Index (LEI) 
LEI is defined by using the minimum bounding box. The box 
represents the spatial extent of newly grown path, which is 
identified using maximum and minimum coordinates of the 
path. If a newly grown patch belongs to the infilling growth 
type, the box is mostly occupied by newly grown patch and the 
rest of box is taken by existing urban patch. If a newly grown 
patch belongs to the edge-expansion growth type, the box is 
partially occupied by newly grown patch, and the rest of box is 
occupied by existing urban patch and other land use patch. If a 
newly grown patch belongs to the outlying growth type, the box 
is partial occupied by newly grown patch and there is no other 
land use type except newly grown patch. In short, the LEI for a 
newly grown patch can be defined and calculated through ex- 
amining the characteristics of its minimum bounding box. 
When the newly grown patch is not rectangular, LEI is calculat- 
ed using equation (1); otherwise, using equation (2) (Liu et al., 
2009). 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B7, 2012 
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia 
100x—5.. (7) 
A4. 
LEI = a 
100 x — He... (2) 
LE -À, 
where LEI = the landscape expansion index for newly grown 
patch, 
Ag - the area of minimum box 
A, = the area of newly grown patch 
A, = the existing urban patch 
When the newly grown patch is rectangle, 
Ar. = the area of enlarged minimum box 
Ap, = the area the existing urban patch in enlarged 
minimum box. 
LEI range in 0 < LEI < 100. 
In contrast with conventional landscape indices, LEI contains 
information of the formation processes of landscape patterns. 
Since conventional landscape indices only reflect information 
for one particular temporal phase of landscape patterns, LEI can 
be used to analyze the dynamic changes of two or more tem- 
poral phases of landscape patterns. 
Using land use and land cover change data in the study area 
from 1993 to 2006, the landscape expansion indices (LEI) are 
calculated in the four different periods. These indices and their 
threshold of different urban expansion types are then used to 
identify different urban expansion types. 
3. RESULTS AND DISCUSSION 
Based on LEI described in the previous section, this section 
presents the study results related to the identification of differ- 
ent types of urban expansions and expansion modes in different 
time periods. 
3.1 Urban Landscape Expansion Type 
According to the landscape expansion index (LEI) value and 
threshold of different types of urban expansions, the distribution 
of urban expansion types are identified in the four different pe- 
riods from 1993 to 2006. The results are shown in Figure 1. 
The results indicate that urban landscape expansion had been 
taking place in the study area from 1993 to 2006. Many non- 
urban lands were converted to urban uses. There were three 
types of urban expansions identified, i.e., infilling type, edge- 
expansion type and outlying type in the study area.