Extraction Of Wetlands Information Using Multi-temporal Remote Sensing Images In Shuangtai Estuary Area Of Liaoning Province

Wetland is one of the world's three major ecosystems, it is widely distributed in the world. The wetland can be used as a water supply, and it will reduce environmental pollution, and it also can be used as the habitat of birds, which plays an important role in quality protection of coastal water, the wetland is one of the most important living environment of mankind. China is rich in wetland resources, the total area of its wetland is the fourth largest in the world, the Shuangtai Estuary area of Liaoning Province has a area of the largest reed swamp in the high latitude area. In recent years, with the increasing human activities, such as reclaim land from the sea, oil production and other factors had caused serious and irreversible damage to wetland, so the research with wetland is more and more important. In order to research the wetland resources protection of the Liaoning Shuangtai Estuary national nature reserve as the study area, the classification and extraction of wetland information were proceeded in this study. The main content and conclusions of this paper are as follows.Using 2015 Landsat8/OLI image data, selecting Liaoning Shuangtai Estuary national nature reserve as the studing region, and the wetland cover information was extracted by using supervised machine learning method,such as(SVM)support vector machine, BP neural network method, and non supervised classification,such as(ISODATA) iterative self organizing data analysis technique, and object oriented method with rule set. According to the multi-temporal images, the accuracy evaluation had been proceeded to the extraction results, and the accuracy of extraction information by using object-oriented with rule set classification method was proved to be highest, and the overall accuracy was 85.22%.Consequently, the object-oriented classification for wetland information extraction is the best classification method.In order to improve the accuracy of the object-oriented classification method, according to the growth rhythm of wetland vegetation, selecting Landsat8/OLI images on May, June, September 2015 as multi-temporal data sets, and image segmentation with these multi-temporal data was conducted by software e Cognition. The rule set is constructed by using the spectrum, texture, space and shape of the object in multitemporal image, and the information of the reserve is extracted by the selected membership function and threshold value.The classification accuracy of wetland information extraction by using multi-temporal remote sensing data can be effectively improved, and the overall accuracy of the classification result is more than 90%,which is 92.17%.By using wetland information extraction results of multi-temporal remote sensing images, the landscape pattern characteristics was analysed by the landscape index via FRAGSTATS software. The results show that in the Shuangtai Estuary national nature reserve, the natural wetlands were main types, and they accounted for 73.21%. The Artificial wetland categories include breed surface, paddy fields and reservoir are less only 18.53%, while the proportion of the less frequent human activities in the area is only 8.27%. Therefore, in the Shuangtai Estuary reserve, the natural wetland is main types, the non wetland area in which the human activities is more frenquent is lower proportion relatively. The results of the statistics showed that in the area of natural wetlands, reeds and natural waters were the main types. In the Artifical wetland, the paddy field and the breed surface were the main types. In the non wetland type, the construction land is the main type. The main land cover types of Shuangtai Estuary reserve were two natural wetlands types, which are reeds and natural waters.The dominant patches in the landscape in Shuangtai Estuary reserve exist higher connectivity and landscape with higher polymerization degree, according to a variety of landscape index, the landscape heterogeneity and landscape crushing degree were concluded. According to the classification results, Shuangtai Estuary reserve wetland containing a lot of corridor landscape unit, and it was segmented by same types of landscape, also showed that Shuangtai Estuary national reserve has been suffered a certain extent impact of human activities interference. Therefore, the conservation of wetland resources is needed to be taken seriously. This study provides a basis for the analysis of the landscape ecological pattern of the Shuangtai Estuary reserve, the future development of the wetland ecosystem and the research of resource conservation.