MONITORING AND ANALYZING THE GEOSPATIAL PATTERNS OF THE PEARL RIVER DELTA (PRD) FROM 1960 TO 2012

The Geospatial Patterns Pearl River Delta (PRD), one of the most economically-important and fastest-growing regions in China, have changed remarkably and continuously during the past decades. In this research, the change of landuse, coastline during 1960~2012 were closely investigated to provide better description and explanation of the geospatial pattern. And the relationships between them were explored. Finally, the impact that urban expansion brought to the coastal environment was quantitatively analyzed. The main remarks of this research are summarized into the following points: (1) In PRD, construction land expanded 33 times. In the meantime, the area of forest and farmland decreased 47.53% and 56.70%, respectively. (2) The land demand for agricultural development was the key factor that changed coastline in PRD before 2000. Since 2000, land demand for urban construction has been the key factor that changed coastline in PRD. (3) During 1960 to 2012, the length of coastline in PRD increased from 1134.95 km to 1508.02 km with an increasing speed of 7.17 km per year. Relatively, the coastline changed more obvious in three periods (2004 to 2006, 2006 to 2008 and 2008 to 2010). (4) The type of coastline changed remarkably from 1960 to 2012. 82.94% of the natural coastline in PRD disappeared until 2012. Known from this research, the geospatial patterns of the PRD changed remarkably during 1960 to 2012, mainly driven by human activities. The coastal environment is facing serious risks and challenges under the rapid process of urbanization. * Corresponding author


INTRODUCTION
Pearl River Delta (PRD) is one of the most prosperous and fastest growing regions in China.Guangzhou, Shenzhen and other important cities located in this area.PRD is the world largest metropolitan area according to the annual report (2015) from UN-Habitat.It created nearly 10% of GDP of mainland China with only 0.57% of the land area at 2014 (Bureau 2015).The geospatial patterns of landuse and coastline remarkably changed under rapid industrialization in PRD.
The coastline is the sea-land demarcation line in coastal regions.The position and shape of the coastline exert remarkable influence on the coastal environment by changing coastal wetland, river estuary, geomorphology of the shore, coastal climate, etc (Dellepiane et al. 2004;Rasuly et al. 2010).It is one of the most important linear features on the earth's surface, which has a dynamic nature (Winarso and Budhiman 2001).In the coastal area, the change of landuse has close relationships with coastline (Freire et al. 2009).The geospatial patterns of coastline and landuse are both the key factor of the coastal environment.
In this research, the change of landuse and coastline in PRD during 1960 to 2012 was closely investigated to provide better description and explanation of the geographical spatial pattern based on RS and GIS.First, the temporal and spatial variation of landuse, coastline and coastal wetland were monitored and analyzed with the quantitative and spatial approach.Furthermore, the relationships among them were explored.Finally, the impact that urban expansion brought to the coastal environment was quantitatively analyzed.

STUDY AREA & DATA SOURCE
The study area in this research was defined by the 10 km buffer zone of the coastline of PRD in 1960.Five important cities in this area are therefore covered, namely Guangzhou (GZ), Dongguan (DG), Shenzhen (SZ), Zhongshan (ZS) and Zhuhai (ZH).With regional excellence, natural resources abundance and economic prosperity, it is the core area of Guangdong-Hong Kong-Macau Greater Bay Area, home to 120 million (Habitat 2016).
Multi-source remote sensing imageries, large-scale topographic maps and historical maps were used in this research, together with DEM data, field survey data and administrative maps as auxiliary data.The main data sources for each periods are listed in Table 1.Atmosphere correction and geometry correction were performed for all 37 imageries.Multi-temporal imagery adaptive regression model was applied for fixing the gap in Landsat 7 imageries (Zeng et al. 2013).The source data used for monitoring geospatial patterns in 1960 were topographic maps with a scale of 1:50000.Additionally, the coastline in 1960 was extracted via visual interpret.

Landuse
The study area was classified into six landuse categories, including buildup area, vegetation, farmland, water body, fishery cultivation (aquaculture) and exclamation land.Supervised classification was applied for mapping the landuse of PRD in 1979~2004 and 2010~2012 from Landsat imageries.For mapping landuse of 2006~2008 from SPOT 5 imageries, O-O classification was applied in order to take full advantage of the plentiful information of geometry, texture, topology in highresolution imagery.Landuse in 1960 was mapped by visual interpretation of topographic maps.In the post classification step, all the results of landuse were carefully corrected based on field survey and temporal consistency was examined.Finally, the results of 1960, 1979, 2006 and 2008 were resampled into 30m for matching with results of other periods.

Accuracy assessment
400 validation points were randomly selected from the study area.To assess the accuracy of landuse classification and coastline extraction respectively, their landuse type and land/marine attribute were assigned based on field survey and high-resolution imageries.Indicated by cross validation, the landuse type of 342 points and the land/marine attribute of 379 were accurately estimated.The accuracy of landuse classification and coastline extraction is 85.5 % and 94.75%, respectively.

Change of coastline form
It can be shown that the length of coastline increased from 1034.22 km to 1508.02 at an annual increasing speed of 6.79km/a (Figure 2).With annual change rate of about 3% (Table 2), 2004~2010 is the most drastically changing period.

Change of coastline type
The extracted coastline was further divided into natural/artificial coastline according to classification criterion used in previous research (Zhu et al. 2013).The results show, 42.95% of the coastline in PRD was natural in 1979 (Figure 3).

CONCLUSION
In this research, the geospatial patterns of PRD were monitored and analyzed based on remote sensing imagery.The relationships between the change of coastline and landuse were explored with quantitative and spatial approaches.Furthermore, we also found out the reason for the dramatic changing of the coastline in PRD.The main outcomes of this research are summarized in the following points: (1) This research demonstrated the ability of approaches based on RS and GIS for monitoring and analyzing the geospatial pattern in the coastal area, which is applicable, convenient and accurate.
(2) The period when the coastline in PRD changed the most dramatically is 2004~2010.Spatially, Guangzhou, Zhuhai and Shenzhen are the places where coastline experienced the most dramatic change.

Figure
Figure 1.The location of the study area

Figure 2 .
Figure 2. Change of coastline length in the PRD (1960~2012) It decreased to  20.19% in 1990.Most of the natural coastline in the north part of PRD and Shenzhen city were converted into artificial.By 2012, almost all the coastline has become artificial.The proportion of natural coastline decreased to 6.07%, all distributed in the remote area of Zhuhai city.

Figure 5 .
Figure 5. Statistics of landuse change in the PRD

Figure 7 .
Figure 7. Landuse state in expanded land

Table 2
. The average annual change of coastline length The coastline in the city of Zhuhai experienced the greatest change in PRD.It reached 709.51km in 2012, 1.61 times of that in 1960.From 2004 to 2008, first, it decreased sharply and then followed by a remarkable increase.The coastline of Zhongshan