Watershed Ecosystem Dynamics In Response To Urbanization

Based on the concept of Ecosystem Based Management, the whole watershed was regarded as an integrated ecosystem. The watershed ecosystem was divided into several sub-systems named vegetation (agricultural land and forest), soil and water, with a case of Qiantang River watershed in Zhejiang Province. The spatiotemporal dynamics of urbanization across the Qiantang River watershed were characterized by remote sensing and geographical information system at four temporal intervals (1979-1985,1985-1994,1994-2003,2003-2009) and three spatial scales (watershed, administration, and ecoregion). The dynamics of the four sub-systems were analyzed by direct (monitoring data based) and indirect (landscape ecology based) approaches, and their dynamic response to urbanization were investigated by geographically weighted regression. A composite indicator model was then developed by catastrophe theory, and the spatial response of watershed ecosystem to urbanization was finally analyzed by geographically weighted regression. The major findings and conclusions were summarized as follows:Between1979and2009, the total population, non-agriculture population (%), and Gross Domestic Product (GDP) all increased significantly. Built-up land expanded vigorously. Under such rapid urbanization, agricultural landscapes became lost, fragmented, irregular, instable and less aggregated. Relationships between agricultural landscape patterns and urbanization presented significant spatial non-stationarity. Generally,(1) at watershed scale, urbanization indicators like total population, non-agriculture population (%), GDP, and area of built-ups could all effectively predict agricultural landscape pattern changes;(2) at administrative scale, urbanization intensity index (UII) and GDP could effectively predict agricultural landscape pattern changes. Changes of agricultural landscape fragmentation and instability could be explained by non-agriculture population (%);(3) at ecoregional scale, UII and GDP could effectively predict agricultural landscape pattern changes. Between1985and1994, total population was a good indicator for agricultural landscape pattern changes.Between2003and2009, the forest NDVI at all scales increased significantly across the zones with elevation less than200m or zones with slope below5°. Changes of plantation structure may account for such results. For zones with elevation greater than600m or zones with slope greater than5°, the forest NDVI at all scales decreased significantly. These results may relate to mining activities and rainfall.Between1979and2009, soil landscape patterns became fragmented, irregular, instable, and became less aggregated, connected and diverse. Relationships between soil landscape patterns and urbanization presented significant spatial non-stationarity. Generally,(1) at watershed scale, urbanization indicators like total population, non-agriculture population (%), GDP, and area of built-ups could all effectively predict soil landscape pattern changes;(2) at administrative scale, UII could effectively predict soil landscape pattern changes. Changes of soil landscape fragmentation and instability could be explained by GDP and total population;(3) at ecoregional scale, UII and GDP could effectively predict soil landscape pattern changes. Total population was a good indicator for soil landscape fragmentation and instability.Between1996and2004, the river water pollution of petroleum, and phosphorus gradually became a regional problem, while heavy metal pollution became a local problem. Spatial regression showed that LULC changes were good predictors for changes of heavy metals. However, GDP and population density contributed to petroleum and ammonia dynamics. For dynamics of dissolved oxygen and total phosphorus, population and GDP were major determinants at sub-basin scale, while LULC changes were determinants at500m buffer scale.Watershed ecosystem composite indicator presented obvious spatial variations in2003. Relationship between watershed ecosystem composite indicator and urbanization exhibited significant spatial non-stationarity. Generally,(1) at administrative scale, cities or counties, with faster economic development and higher non-agriculture population (%), had lower ecological quality;(2) at ecoregional scale, regions, with faster economic development and higher population, had lower ecological quality.The original contributions of this study were summarized as follows:(1) Spatially varying relationships between agricultural landscape patterns, soil landscape patterns and urbanization at different scales were quantitatively analyzed.(2) This study developed an innovative management oriented source apportionment framework; and analyzed the spatial determinants of river water quality dynamics.(3) This study proposed a new composite indicator watershed ecosystem model based on catastrophe theory; and quantitatively analyzed the dynamic response of watershed ecosystem to urbanization at different scales. This study incorporated several limitations and further study should focus on the selection of scales, regression models, composite indicator models, division of watershed ecosystem, uncertainty analysis and development of predicting models.