environmental issues such as climate change, urban heat island effect, water quality
deterioration, vegetation degradation, increased flooding risk, decreased air quality
(Bhatta 2010; Sexton et al. 2013a, b; Li et al. 2011; Tan et al. 2009; Sundarakumar
et al. 2012; Thapa and Murayama 2009). Therefore, consistent monitoring of land
use and land cover (LULC) change at local and regional scale is an urgent need for
planners and policy makers to understand change dynamics of an area to make it
more appropriate and effective decisions of planning and environmental management in the future.
With recent development of remote sensing technologies and accessibility to
remotely sensed data, the study of identifying detailed spatial and temporal changes
of urban area and monitoring urban growth have become more cost-effective and
successful (Huang et al. 2011; IRS 2013; Jensen 2006; Thapa and Murayama 2009;
Patino and Duque 2013; Lunetta et al. 2004). To date, various change detection
methods have been explored and developed for detecting LULC change analysis
(Singh 1989). Technically, image algebra (i.e. image differencing and image
ratioing), principal component analysis (PCA), post-classification change detection
(PCCD), direct multi-date classification, and change vector analysis (CVA) are
most widely used methods for change detection (Singh 1989; Alumutairi and
Warner 2010; Coppin et al. 2004; Jensen 2005). From an application perspective,
most of the previous studies on urban growth and LULC change detection were
based on bi-temporal and coarsely multi-temporal analyses.
The previous researchers indicated that the bi-temporal and coarsely multitemporal analyses have their own advantages of providing useful change information. They are unable to observe dynamic change patterns and higher-order complexities, such as acceleration, deceleration of specific LULC change within a longterm time span (Sexton et al. 2013a, b). The dynamic change patterns include
spatially and temporally complex changes in water, forest, agriculture, and urban
built-up area caused by natural and anthropogenic processes (Sexton et al. 2013a,
b). Moreover, the impacts on ecosystems caused by frequent human activities
exhibit nonlinearities, time lags, and legacy effects, and the change dynamics is
only able to be detected by long-term repeatedly measurements (Sexton
et al. 2013a, b).
With the opening of Landsat archive from United States Geological Survey
(USGS) in 2009 (Sexton et al. 2013a, b; Wulder et al. 2011), an increased demand
of long-term time-serial analysis of urban growth and LULC change dynamics can
be met (Sexton et al. 2013a, b; Hansen and Loveland 2012). Therefore, with a free
access of Landsat archive, processing dense datasets with high frequency will shift
research focus from analyzing static bi-temporal change to comprehending more
detailed long-term change dynamics in which planners, policy makers and resource
managers are much more interested (Sexton et al. 2013a, b).
In this chapter, the role of satellite data and Landat archive data for change
detection analysis will be introduced. In addition, an overview of change detection
methods will be provided. To reveal the superiority of long-term change dynamics
analysis using high-dense Landsat images, this chapter focuses on a case study of
change detection analysis of the Region of Waterloo. Also, based on a case study
64
A. Fu et al.
deterioration, vegetation degradation, increased flooding risk, decreased air quality
(Bhatta 2010; Sexton et al. 2013a, b; Li et al. 2011; Tan et al. 2009; Sundarakumar
et al. 2012; Thapa and Murayama 2009). Therefore, consistent monitoring of land
use and land cover (LULC) change at local and regional scale is an urgent need for
planners and policy makers to understand change dynamics of an area to make it
more appropriate and effective decisions of planning and environmental management in the future.
With recent development of remote sensing technologies and accessibility to
remotely sensed data, the study of identifying detailed spatial and temporal changes
of urban area and monitoring urban growth have become more cost-effective and
successful (Huang et al. 2011; IRS 2013; Jensen 2006; Thapa and Murayama 2009;
Patino and Duque 2013; Lunetta et al. 2004). To date, various change detection
methods have been explored and developed for detecting LULC change analysis
(Singh 1989). Technically, image algebra (i.e. image differencing and image
ratioing), principal component analysis (PCA), post-classification change detection
(PCCD), direct multi-date classification, and change vector analysis (CVA) are
most widely used methods for change detection (Singh 1989; Alumutairi and
Warner 2010; Coppin et al. 2004; Jensen 2005). From an application perspective,
most of the previous studies on urban growth and LULC change detection were
based on bi-temporal and coarsely multi-temporal analyses.
The previous researchers indicated that the bi-temporal and coarsely multitemporal analyses have their own advantages of providing useful change information. They are unable to observe dynamic change patterns and higher-order complexities, such as acceleration, deceleration of specific LULC change within a longterm time span (Sexton et al. 2013a, b). The dynamic change patterns include
spatially and temporally complex changes in water, forest, agriculture, and urban
built-up area caused by natural and anthropogenic processes (Sexton et al. 2013a,
b). Moreover, the impacts on ecosystems caused by frequent human activities
exhibit nonlinearities, time lags, and legacy effects, and the change dynamics is
only able to be detected by long-term repeatedly measurements (Sexton
et al. 2013a, b).
With the opening of Landsat archive from United States Geological Survey
(USGS) in 2009 (Sexton et al. 2013a, b; Wulder et al. 2011), an increased demand
of long-term time-serial analysis of urban growth and LULC change dynamics can
be met (Sexton et al. 2013a, b; Hansen and Loveland 2012). Therefore, with a free
access of Landsat archive, processing dense datasets with high frequency will shift
research focus from analyzing static bi-temporal change to comprehending more
detailed long-term change dynamics in which planners, policy makers and resource
managers are much more interested (Sexton et al. 2013a, b).
In this chapter, the role of satellite data and Landat archive data for change
detection analysis will be introduced. In addition, an overview of change detection
methods will be provided. To reveal the superiority of long-term change dynamics
analysis using high-dense Landsat images, this chapter focuses on a case study of
change detection analysis of the Region of Waterloo. Also, based on a case study
64
A. Fu et al.