Dynamic Pattern of Agricultural Landscapes in Response to Urbanization across Hangzhou Metropolitan Region: A Remote Sensing Approach

This paper applied multiple endmember spectral mixture analysis (MESMA) to interpret Landsat satellite imageries and then analyzed the spatiotemporal dynamics of agricultural landscapes within Hangzhou metropolitan region. Between 1994 and 2003, this region has witnessed accelerated urbanization based on four indicators: total population, percentage of non-agriculture population, GDP per capita, and area of impervious surface. Meanwhile, conversion, fragmentation, irregularity and isolation of agricultural landscapes were also captured according to four class-level metrics: total area, patch density, area-weighted mean shape index, and Euclidean nearest neighbor distance. Statistics showed that increased impervious area was significantly correlated with declined area and configuration of agricultural landscapes. Growth in GDP per capita was also correlated with increased irregularity of agricultural landscapes. Changes of total population and percentage of non-agriculture population were weakly associated with all the agricultural landscape metrics, denoting that the two indicators did not have direct influences on agricultural landscape patterns. Further, distance to the nearest road was also a good indicator of how the cultivated land was lost and transformed. Our results demonstrated complexities of relationships between urbanization and agricultural landscapes, and highlighted the importance of selected variables, spatiotemporal scales and land-use classification when quantifying these relationships.