Dynamic Change Monitoring Of Typical Vegetation In Coastal Wetlands On The South Shore Of Hangzhou Bay

As an important part of wetlands,wetland vegetation is with extremely important value.From the perspective of biological systems,it is the main habitat,breeding ground,and important wintering ground for numerous rare water birds;in terms of regional ecological environment,it protects the biodiversity within the area and maintains the balance of the regional ecosystem.At the same time,it is also with the functions of bank protection,flood control,sediment retention,and water purification.It is an ecological barrier to regional defense and mitigation of disasters.Under the premise that wetland vegetation is seriously degraded due to human activities such as wetland reclamation,it is particularly important to obtain information on wetland vegetation accurately and quickly.Two Quick Bird images in the 2008 and 2013,an GF-2 images in the 2017 and field survey data was used as data sources to analyze the typical vegetation of coastal wetlands on the southern coast of Hangzhou Bay in a dynamical way.First,based on previous research results,the optimal segmentation scale evaluation was performed on the multi-scale segmentation results of high-resolution remote sensing images in the study area.Then,the effects of different data fusion methods and different classification algorithms on the vegetation extraction results of typical coastal wetlands were explored in the optimal segmentation scale.Finally,a superposition analysis of the classification results for2008~2013~2017 is conducted to summarize the types of land use and the spatial distribution characteristics of typical wetland vegetation,temporal and spatial variation patterns,and indications of ecological environment protection.The study mainly gets the following conclusions:(1)The average global score index method is the most object and effective way to get optimal segmentation scale.It plays an important role in improving the image classification accuracy.(2)Different data fusion methods and machine learning method(random forest and C4.5 algorithm)on the basis of optimal segmentation scales are used while 2013 Quick Bird data and field survey data as data sources.we found that the combination ofobject-oriented and random forest algorithm is more applicable to the accurate extraction of vegetation information from coastal wetlands than the combination of object-oriented and C4.5 algorithms.It can effectively extract 5 coastal wetland vegetations and 6 types of land use types in the study area.The overall classification accuracy is 86.90%.The classification accuracy of the random forest algorithm is not particularly sensitive to the number of feature variables,and integration into the Landsat-8 band information does not significantly improve the classification effect.(3)From the results of the 2008~2013~2017 three-stage classification results chart,after the image classification and comparison analysis method,the spatial distribution of typical wetland vegetation was found from the coastal intertidal zone to the land shoreline in order: Scirpus mariqueter,Spartina alterniflora,reeds,and artificial vegetation.Scirpus mariqueter is concentrated in the intertidal zone farther from the land,showing a strip parallel to the coastline;the intertidal zone where Spartina alterniflora is close to the land is scattered.And the reed is mainly distributed in the early stage of reform of the tidal flat.In the culture ponds,artificial vegetation is mainly distributed in the area where the beach cofferdam has been transformed for a long time.(4)The spatial distribution of vegetation in coastal wetlands is greatly influenced by the expansion of tidal flats and the exploitation of beach development.The development of tidal flats along the coast of Hangzhou Bay has continued to expand toward the sea,and the land used in the coastal zone has changed dramatically.There are four main types of vegetation change pattern: muddy beach-Scirpus mariqueter-water-unused land,muddy beach-Scirpus mariqueter-reed-water,Scirpus mariqueter-Spartina alterniflora-reed-water,Spartina alterniflora-water-unused landm,four kinds of conversion of cultivated lands.Scirpus mariqueter grows mainly in the lower part of the tidal flat of the new beach,and Spartina alterniflora grows in the lower tidal level along the coast.At the beginning of the redevelopment period,the beach became a pond,the habitat changed,the seagrasses disappeared,the Spartina alterniflora gradually disappeared,and the vegetation was mainly reed.When the pond was further converted into arable land,the vegetation was mainly artificially planted.(5)From 2008 to 2017,the total area of the Scirpus mariqueter was increased first and then decreased.The area of increasing and decreasing was approximately flat,which isthe area of cofferdam was similar to natural expansion;the total area of reed was increasing from 2.57 km2 to 7.44 km2.The total area of Spartina alterniflora was continuously reduced due to human governance.The area where human activities cause the transformation of wetland vegetation is far greater than the area where the natural succession of wetland vegetation shows that the ecological environment of wetland vegetation is seriously affected by the activities of humans and the speed is also accelerating.Vigorous human activities affect the ecological balance of wetlands,cause the deterioration of the habitats of animals and plants and break the balance between the development and protection of wetlands.