Study On Remotely Sensed Monitoring Of Wetland At Zhongwei Oasis And Its Adjacent Regions

As an indicative land use and land cover in arid and semi-arid region of northwest China, wetland plays an important role in preserving balance and stability of ecosystem at Zhongwei oasis and its adjacent regions. It also ensures sustainable development and virtuous circle of oasis and prevents sand dune movement outward the Tengger Desert. This article analyzed the categories of wetland, the rules of wetland changes and driving forces of wetland changes at Zhongwei oasis and its adjacent regions based on3S technology (RS, GIS, GPS),which had scientific and socio-economic values.In this paper, the author explored the classification and dynamic change methods of wetland in arid and semi-arid region of northwest china based on time series of Landsat images and object-base image analysis. And then the author analyzed the categories, area, distribution, dynamic changes and change rules of wetland at Zhongwei oasis and its adjacent regions. On this basis, the author also analyzed the reasons of wetland changes from both natural factors and human activities.The main conclusions are as follows:(1) Object-based image analysis method can extract wetland effectively in the study area, and avoid "salt-and-pepper" phenomina which commonly happens in classical pixel-based image analysis. Object-based image analysis method can apply not only spectral feature, but also texture, shape, context and other spatial feature to extract wetland; Furthermore, object-based image analysis method has powerful soft-classifier based on fuzzy system, which can provide high classification accuracies.(2) This article evaluated the image segmentation results using the discrepancy measures. We used three metric indices, Potential Segmentation Error (PSE), Number-of-Segments Ratio (NSR), and Euclidian Distance2, to select the optimal segmentation parameter combination. The three metric indices focus not only geometric relationships but also arithmetic relationships between reference polygons and corresponding segments. The article got a good segmentation result using the optimal segmentation parameters, which can ensure getting an accurate classification result.(3) This article took full advantage of spectral, textual, shape and contextual feature of image objects to classify wetland types using membership function method based on fuzzy rules. This article firstly selected many features, such as NDVI, NDWI, mean SWIR, mean NIR、GLCM contrast, Area, Roundness, Rel. border to and so on, as the parameters for the rule set. Then the author analyzed the distribution rule of wetland to establish the membership function. Finally the author classified the wetland using OBIA based on the corresponding classification levels. the classification results showed the method can extract and classify wetland effectively in the study area.(4) The types of wetland in the study area consisted of river wetland, lake wetland, marsh wetland and artificial wetland. The area of the river wetland was2306.61hm2in2010, which accounted for16.13%of total area of wetland in the study area. The area of lake and marsh wetland was smaller, which is1250.55hm2,778.59hm2respectively. The area of artificial wetland was9960.39hm2in2010, which accounted for69.67%of total area of wetland in the study area; the area of paddy land was9668.34hm2in2010, which accounted for97.07%of artificial wetland in the study area.(5) The wetland located in the hinterland of Tengger Desert had obvious changes from May to September in2007, which mainly because the wetland benefited from the seasonal rainfall supplied from the surrounding mountains, and with the amount of precipitation, the area of the wetlands had frequent changes. Whereas, the wetlands located in the Zhongwei oasis had small changes from May to September in2007, which mainly because the wetland had the stable water supply from Yellow River.(6) The area of the wetland located in the Zhongwei oasis increased firstly and then decreased during nearly20years. But the area of other wetlands excluding the paddy land showed an increase trend, which mainly because of human activities. The area of wetland showed growth trend from1992to1999, which because a large number of fish ponds were built in the southeast margin of the Tengger Desert. The area of wetland increased from2002to2007, which because the Longgong lake was built. The area of wetland increased from2007to2010, which because a number of man-made lake were built in the study area.(7) The driving forces which caused the change of wetland at Zhongwei oasis and its adjacent regions include natural factors such as climate and hydrology, and human factors such as urban expansion and policy interventions. However, Human factors play a more prominent role in affecting wetland in the study area.