Urban Land-Cover Change Detection through Sub-Pixel Imperviousness Mapping Using Remotely Sensed Data

We developed a Sub-pixel Imperviousness Change Detection (SICD) approach to detect urban land-cover changes using Landsat and high-resolution imagery. The sub-pixel percent imperviousness was mapped for two dates (09 March 1993 and 11 March 2001) over western Georgia using a regression tree algorithm. The accuracy of the predicted imperviousness was reasonable based on a comparison using independent reference data. The average absolute error between predicted and reference data was 16.4 percent for 1993 and 15.3 percent for 2001. The correlation coefficient (r) was 0.73 for 1993 and 0.78 for 2001, respectively. Areas with a significant increase (greater than 20 percent) in impervious surface from 1993 to 2001 were mostly related to known land-cover/land-use changes that occurred in this area, suggesting that the spatial change of an impervious surface is a useful indicator for identifying spatial extent, intensity, and, potentially, type of urban land-cover/land-use changes. Compared to other pixel-based change-detection methods (band differencing, rationing, change vector, post-classification), information on changes in sub-pixel percent imperviousness allow users to quantify and interpret urban land-cover/land-use changes based on their own definition. Such information is considered complementary to products generated using other change-detection methods. In addition, the procedure for mapping imperviousness is objective and repeatable, hence, can be used for monitoring urban land-cover/land-use change over a large geographic area. Potential applications and limitations of the products developed through this study in urban environmental studies are also discussed. Introduction Rapid urbanization and urban sprawl have significant impact on conditions of urban ecosystems. Accurate and updated information on the status and trends of urban ecosystems is needed to develop strategies for sustainable development and to improve the livelihood of cities. The ability to monitor urban land-cover/land-use changes is highly desirable by local communities and by policy decision makers alike. With increased availability and improved quality of multi-spatial and multi-temporal remote sensing data as well as new analytical techniques, it is now possible to monitor urban landcover/land-use changes and urban sprawl in a timely and cost-effective way. Urban Land-Cover Change Detection through Sub-Pixel Imperviousness Mapping Using Remotely Sensed Data Limin Yang, George Xian, Jacqueline M. Klaver, and Brian Deal Using multi-date satellite remote sensing data to detect land-cover change goes back to the early 1970s (Singh, 1989). Many remote sensing change-detection methods have been evaluated (e.g., Kam, 1995; Jensen, 1995; Ridd and Liu, 1998; Sohl, 1999). There is no consensus as to a single method/ algorithm that is universally applicable. The most commonly used change-detection methods are either spectrally based (image-to-image) or classification-based (map-to-map) method (e.g., Green et al., 1994; Yang and Lo, 2002; Loveland et al., 2002). Most urban land-cover/land-use change studies utilized Landsat data due to the uniqueness of the dataset as the only long-term digital archive with a medium spatial resolution and relatively consistent spectral and radiometric resolution. Urban change studies using Landsat Multispectral Scanner (MSS) or Landsat Thematic Mapper (TM) data have been conducted either at a regional scale encompassing several urban areas (Todd, 1977; Royer et al., 1988) or a single metropolitan area (Gomarasca et al., 1993; Johnston and Watters, 1996). Recently, long-term urban land-cover/land-use changes (over two decades or longer) have been studied using the methodology of post-classification comparison using the Landsat archive as a baseline data source (Chen et al., 2002; Yang and Lo, 2002; Loveland et al., 2002). Despite continued improvements in methodology, several limitations are recognized regarding some commonly used change-detection methods. First, existing change-detection techniques have distinct advantages and disadvantages. For example, the post-classification (map-to-map comparison) method identifies conversion from one land-cover/land-use type to another with little information on the intensity of such changes. This method often involves intensive manual interpretation and relies heavily on the skills of the interpreter. The spectrally based (image-to-image) method of change detection provides quantitative information on spectral change over time. However, interpretation of the spectral difference images with regard to the type of land-cover/land-use change is not always straightforward (Sohl, 1999). Second, the majority of urban change studies using remotely sensed data assumed homogeneity within a single pixel, resulting in no quantifiable changes at the sub-pixel level. In actuality, most Landsat pixels in urban areas are mixed and composed of several land-cover/land-use types. Ignoring the sub-pixel variation of Landsat imagery can lead to a biased estimate in urban change analysis. Finally, conventional methods allow users PHOTOGRAMMETR IC ENGINEER ING & REMOTE SENS ING September 2003 1003 L. Yang, G. Xian, and J.M. Klaver are with SAIC, U.S. Geological Survey EROS Data Center, Sioux Falls, SD 57198 ([email protected]). B. Deal is with the Department of Urban and Regional Planning, Building Research Council, University of Illinois, 1 E. St Mary’s Rd, Champaign, IL 61820. Photogrammetric Engineering & Remote Sensing Vol. 69, No. 9, September 2003, pp. 1003–1010. 0099-1112/03/6909–1003$3.00/0 © 2003 American Society for Photogrammetry and Remote Sensing 03-916.qxd 8/7/03 5:30 PM Page 1003