nbul 2004 
rtant and 
s on geo- 
es of the 
rpretation 
XML has 
isfer and 
ts. XML 
cture for 
arding to 
s for the 
S, free or 
racter of 
systems, 
enges for 
services, 
Internet 
k client 
a format 
es. 
own S., 
5 4s 
W3C 
address: 
mat for 
national 
ary 3- 
ition of 
, Thesis 
for the 
sity of 
ity, GIS 
ALIZE 
HONE, 
DETECTING AND MODELLING DYNAMIC LANDUSE CHANGE USING 
MULTITEMPORAL AND MULTI-SENSOR IMAGERY 
Q. Zhou^ *, B. Li*, C. Zhou. 
* Dept. of Geography, Hong Kong Baptist University, Kowloon Tong, Kowloon, Hong Kong - qiming@hkbu.edu.hk 
? State Key Laboratory of Environment and Resources Information System, Institute of Geographical Sciences and 
Resources Research, Chinese Academy of Sciences, Beijing 100101, P.R. China 
Commission IL, Working Group 11/5 
KEY WORDS: Land Cover, Change Detection, Monitoring, Modelling, Multisensor, Multitemporal 
ABSTRACT: 
It is now common to use data from two or more sensors for land cover change detection. Since the spatial and spectral resolutions of 
different sensors vary significantly, the ability to discriminate the land cover also varies greatly. In this paper the applications of 
landuse change detection including area statistics, temporal trajectories and spatial pattern are discussed. The area statistics show the 
general landuse change pattern, but with quite significant uncertainty. The results of this study show that if the area of detected 
landuse change accounts for less than 5% of the total area, the uncertainty of change detection can be very significant. Temporal 
trajectory analysis was also conducted with the particular focus on the analysis of unchanged and “stable” change trajectories, 
because they generally show the trend of landuse change that is irreversible. Unstable change trajectories, on the other hand, show 
relatively less significance since they largely contain reversible temporary changes (c.g. seasonal cropping and bare ground) and 
classification errors. The study results show overall accuracy of 85-90% with Kappa coefficients of 0.66-0.78 in classification and 
change detection. On spatial patterns, the landuse pattern metrics demonstrate a reasonable result, but most other patch metrics do 
not show recognisable patterns. 
1. INTRODUCTION on spatial pattern metrics statistics based on the same stage 
multi-resolution imagery (Benson and MacKenzie 1995, 
Land cover change plays a pivotal role in regional socio- Wickham and Riitters 1995), in which the focus of discussion 
economic development and global environment changes (Chen was on the effects of spatial resolution on the landscape spatial 
2002). In arid environment, where fragile ecosystems are pattern metric using multi-resolution remotely sensed imagery 
dominant, the land cover change often reflects the most with the same acquisition period. Less attention, however, was 
significant impact on the environment due to excessive human paid to the effect of multi-resolution and multitemporal data on 
activities. area statistics, trajectory statistics and spatial pattern metrics. 
This study evaluates the effect of multi-resolution data on the 
When monitoring natural environment and land cover change, change detection in an arid environment over a monitoring 
three aspects are focused (Singh 1989, MacLeod and Congalton timeframe of 30 years. The focus of the discussion is on the 
1998): statistics of area extent, temporal trajectories and spatial pattern. 
* areal extent of the change, measuring the magnitude of the 
change; 2. METHODOLOGY 
e the nature of the change, measuring the temporal trajectory 
of the change; The generic approach of this study is based on post- 
* spatial pattern of the change, measuring spatial distribution classification comparison method, ; which is commonly 
and relationship of the change. employed in land cover change detection studies (Miller et al 
1998, Larsson 2002, Yang and Lo 2002, Zhang et al 2002, Liu 
Numerous works have been reported in these fields (Miller et al and Zhou 2004). A unified land cover classification scheme was 
1998, Mertens and Lambin 2000, Petit et a/ 2001, Maldonado et established for classification of images. The classified images 
al 2002, Pereira er al 2002). For landuse change detection, ^ Were then used to derive class area statistics, temporal 
imagery data from various sensors such as Landsat MSS, TM, trajectories and spatial pattern in the past 30 years. 
ETM, SPOT HRV, IRS and AVIRIS are often used, and it is 
common that images from two or more sensors were used 2.1 Study area and data 
(Prakash and Gupta 1998, Luque 2000, Masek er al 2000, 2: p : 
Mertens and Lambin 2000, Roy and Tomar 2001, Ustin and The study area is centred at ar N and 85°43’E and located in 
Xiao 2001, Yang and Lo 2002). Since the spatial and spectral Donghetan Township, Yuli County, Xinjiang Uygur 
resolutions of different sensors vary. significantly, the ability to Autonomous Region, € hina. It locates at the middle reach of 
discriminate the land cover also varies greatly. Some research Tarim River, the longest inland river of China (figure 1). At the 
: = ; ; : ring an Desert, the “green corridor” of Tarim 
work has been reported on the effect of multi-resolution sensors fringe of Taklimakan De green co or 
Basin is one of the most important habitation areas in aridzone 
  
* Corresponding author. 
697