Dynamics of Impervious Surface Pattern in Cixi County and Its Simulation
【摘要】：Hangzhou Bay, is located on the East coast of China, had become the major social, cultural, business and intellectual stimulus in China so far. Also Hangzhou Bay has pioneered the nation in economic growth and rapid urbanization since China commenced economic and policy reforms in 1978. In the recent decades, however, the environment of the bay has degraded due to the rapid urbanization and industrialization.
In this respect, there is a long history of remote sensing research on the land use change. Most remote sensing studies have mainly focused on the declined of agriculture land and loss of biodiversity caused by rapid urbanization. However, less attention has been directed to monitor urban growth in terms of pattern and process. As a matter of fact, monitoring urban growth, can help us to understand how an urban landscape changing with time. Probably, there might be some reasons which make using of remote sensing more difficult for urban growth analysis. First of all, the characteristics of urban landscapes make mixed pixels a common problem in medium spatial resolution data. Second conventional classification technique may not extract characteristics of urban area. This is because the physical composition of land use classes may vary dramatically from region to region due to the different building materials. Also traditional land change detection ignores quantifiable changes at the subpixel level or within-class change in urban regions.
Recently, issue related to impervious surface index have attracted interest among a wide variety of researches ranging from those who favor examining environmental impacts of land conversion and those who try to understand causes and consequences of urban growth. Impervious surface describes the whole of the impermeable surface including roads, building, parking lots, railways and side walks. This thesis demonstrated application of impervious surface for monitoring internal modification of urban regions and simulating urban growth.
First, imperviousness index (%) was derived from Landsat images(i.e.,1974,1981,1987,1995,2002 and 2009) by using multi-endmember spectral mixture analysis (MESMA). MESMA provided reasonable estimates of abundances of impervious surface for multiple years of Landsat dataset over Cixi County. However, the results showed that the performance of this technique was susceptible to morphology of Chinese cities.
Second, so far most impervious surface change detection analysis based on either MESMA, or other techniques have mainly focused on two dates of satellite images or more than two dates of imagery using dates comparison in sequence rather than a composite analysis of three dates as multi-temporal. This situation worsens if time series remote sensing data (more than three) are used for urban land cover change detection, especially for detection inter-class and intra-class changes. To overcome this shortcoming, this research developed a methodology based on the concept of Red, Green, Blue(RGB) model to monitor time series of internal changes of urban regions including between-class changes and within-class change. We presented here a novel method of combining remote sensing tools at the subpixel level for accurate identification of impervious surface changes. This methodology was called RGB-Impervious Surface(IS). The performance of this technique was compared with RGB-NDVI and Post-Classification technique.
RGB-IS derived a richer amount of urban changes information including between-class changes and within-class changes in comparison to RGB-NDVI and post-classification. Our investigation showed that RGB-IS(urban modification Table based on IS) improved overall result, namely visual interpretation and overall accuracy compared to the RGB-NDVI(urban modification Table based on NDVI). Also the result was accordance with previous research which depicted that impervious surface was stable, and almost independent of seasonal change and atmospheric conditions. Moreover, our finding indicated that RGB-IS can be able to serve as invaluable alternative for quick mapping potential hotspot growth and monitoring imperviousness change.
Third, in terms of change detection, results showed that most of the urban evolution was concentrated in the 1974-1981(8.97% year-1),1981-1987 (20.49% year-1) and 1987-1995(18.52 year-1). However, urbanization rate dropped slightly to approximately 6.8%, in the periods 1995-2002 and 2002-2009. In this research, it is confirmed that urban expansion exhibits an alternating process of diffusion and coalescence, and spatial metrics showed changeable behavior during this process. Diffusion and coalescence were detected in periods 1974-1987 and 1987-2002, respectively. Since 2002, urban regions have grown more compact and saturated by inflling the gaps within the urban areas.
Finally, urban growth was simulated by using SLEUTH to 2015 based on the four specific scenarios to explore the potential hotspot impervious surface growth. The first scenario was to simulate the current trend (CT) of urban growth. The second scenario was to project the limited urban growth trend by taking environmental protection (EPG) into account, particularly for agricultural areas. The third scenario was to predict accelerated urban development and business growth (BG). The last scenario was based on Chinese policy for protecting rural regions(CPPR). It was found that urban regions would grow outward the edge of existing urban areas if BG scenario and CT scenarios were applied to Cixi County. Also the result from the EPG scenario showed that much more land could be preserved in the county. Moreover, the result from the CPPR scenario indicated urban development would decline in rural regions for the next 7 years. In this regard, most impervious surface growth would be inside or close to the major towns or cities rather than rural regions or small villages.
Above all, although the results of change detection and modeling based on impervious surface promising, its accuracy was not yet to reach the level of accuracy required by urban planners for practical application. Much of this low accuracy may be related to the environment of the study site, which was located in the coastal regions. Nevertheless, impervious surface performance in highly heterogeneous nature of urban surface is comparable with conventional methods. However, more research is required to adapt this index for accurately mapping of Chinese cities.