Extraction And Dynamic Analysis Of Coastal Wetland Type Information Based On The Seasonal Rhythm Of Vegetation

The coastal wetland is an important part of the ecosystem and it has great significance in regulating regional ecological balance and maintaining regional climate change,Realizing the extraction of coastal wetland type information is conducive to the sustainable development of protected areas.This paper takes the core area of yancheng national nature reserve of Jiangsu Province as the research object and explore the seasonal rhythm of wetland vegetation and the multiple characteristics of typical wetland types based on the remote sensing image data of 2005,2010 and 2015.The using CART Decision tree classification model to focus on the study of different wetland type information extraction methods based on pixel and object-oriented and analyzes the dynamic changes of various wetland types in the study area in the past 10 years.The main research results obtained in this paper are as follows:(1)The NDVI time series curve of spartina,phragmites and suaeda in the study area is generally in line with the seasonal growth of vegetation.Based on Landsat remote sensing image data from January to December of 2014 in the study area,explore and analyze the seasonal growth regularity of different vegetations in one year through extracting the NDVI average values of spartina,phragmites,suaeda and establishing the time series curve of NDVI.It is concluded that the growth period of 3 plants is about 2-11 months,and the NDVI value reaches the maximum in November,the analysis of the seasonal rhythm of vegetation can effectively provide a theoretical basis for the selection of subsequent remote sensing images.(2)The object-oriented classification method is superior to the pixel-based classification method,the accuracy increases from 82.9%to 87.5%,and the Kappa coefficient increases from0.77 to 0.84.Based on the multivariate analysis of seasonal rhythms,spectral features,vegetation indices,and texture features of vegetation,combined with the CART decision tree model,using the single-level classification methods based on pixel and object-oriented to extract wetland type information in the study area.The results show that the object-oriented method can effectively improve the precision of information extraction of coastal wetland types.(3)The object-oriented multi-hierarchical classification method is superior to the object-oriented single-hierarchy classification method,the accuracy increases from 87.5%to92.3%,and the Kappa coefficient increases from 0.84 to 0.90.Using the object-oriented hierarchical classification method based on the CART decision tree model to extract information of various wetland types in the study area,based on the characteristics of each wetland type,various optimal feature quantities such as spectrum,vegetation index,and texture.The result show that the object-oriented multi-hierarchical classification method can realize fine classification of wetland types in the study area.(4)In the past 10 years,the phragmites has the largest change in area,it increased from3005.56 hm~2 to 5882.40 hm~2;the water has the smallest change in area,it decreased from10986.93 hm~2 to 10216.46 hm~2.Through the analysis of change amplitude,dynamic degree and transition matrix of various wetland types in the study area from 2005 to 2015,the change amplitude of phragmites is 95.7%and its dynamic degree is 19.1%;the change amplitude of water is-7.0%and its dynamic degree is-1.4%;the change amplitude of spartins is 25.6%and its dynamic degree is 5.1%;the change amplitude of suaeda is-33.6%and its dynamic degree is-6.7%;the change amplitude of fishpond is-70.1%and its dynamic degree is-14.1%.(5)In the past 10 years,suaeda showed a decreasing trend in the study area,while spartina and phragmites showed an increasing trend.Based on the spatial superposition analysis of the main vegetation types in the study area in the past 10 years,it is concluded that the spatial pattern change of suaeda in the study area is mainly in the central part of the core area,showing a continuously decreasing trend;the spatial pattern of the spartina is mainly in the east and southwest of the core area,showing a trend of continuous expansion in both east and west direction;the spatial pattern of phragmites is mainly in the northwest and southwest regions,showing a trend of continuous expansion to the west.