Font Size: a A A

Study On A Method For Delineation Of Wetland Hydrology Boundary Using Multi-sources Remote Sensing Data

Posted on:2018-07-14Degree:DoctorType:Dissertation
Country:ChinaCandidate:B L FuFull Text:PDF
GTID:1310330539965089Subject:Cartography and Geographic Information System
Abstract/Summary:PDF Full Text Request
Definition of wetland boundary is fundamental research of wetland science,representing the understanding of wetland definition,character and function.Scientific definition of wetland boundary is of great importance for wetland identification,classification and mapping,investigation of wetland resources,evaluation of wetland function and value.Although the variousness of definition theory result in lacking unified standard for defining wetland boundary,researchers have agreed that:(1)wetland hydrology,wetland vegetation and hydric soil are the sole criterion for identifying wetland and delineating wetland boundary,wetland hydrology plays a decisive role in wetland vegetation development and hydric soil formation,and is a genetic factor of wetland formation;(2)wetland hydrological characteristics is the sore reliable standard for definition of wetland boundary.In addition,due to the tight coupling between wetland hydrology and wetland vegetation or wetland soil,wetland boundary delineated based on hydrological characteristics is impossible to exist independently without wetland vegetation boundary or wetland soil boundary.This paper focused on the wetland hydrological characteristics,presented the theory and method for determining thresholds of water level depth-duration-frequency of wetland hydrology,explored “water level-duration” process of Honghe National Nature Reserve wetland,inverted the thresholds of water level depth-duration-frequency,and utilized these thresholds to define the boundary of wetland hydrology.Main work is as follows:(1)this paper summarized the current problems in studying water level depth-duration-frequency threshold of wetland hydrology,and systematically expounded the water level depth-duration-frequency threshold theory and method by combination of soil science,plant ecology,system theory and inversion theory.The paper also filled up a theoretic gap for determining water level depth-duration-frequency threshold in the under the condition of incomplete wetland types,and presented the test method and standard for duration threshold.(2)In order to identify the type of wetland vegetation with high precision,this paper extracted different wavelength polarimetric decomposition parameters of study area based on polarimetric SAR target decomposition theory,and integrated the polarimetric parameters with multi-spectral bands and its texture features for multi-source remote sensing dataset.Object-oriented image analysis technology and Random Forest machine-learning algorithm were utilized to build remote sensing identification models to classify wetland vegetation,and further delineate the wetland vegetation boundary.The classification accuracies derived from different data combinations were compared.(3)This paper attempted to use Interferometric Synthetic Aperture Radar(InSAR)technique to construct Digital Elevation Model(DEM)of study area,and monitorsed water level change of wetland.This paper also evaluated the performance of InSAR using the 1:10000 topographic map,and comparatively analyzed precision difference in the generated DEM between different wavelength SAR interferometric data.In addition,the paper further explored coherence differences of wetland vegetation between different wavelength coherence coefficient maps,and within different growth stages.(4)According to the theory and method of water level depth-duration-frequency threshold,a model was built to delineate boundary of wetland hydrology.Water level depth-duration threshold was inverted using the water level depth-duration curve in the vegetation transects from 2007 to 2015,which delinated boundary of wetland hydrology.Finally,Spatialization of boundary of wetland hydrology was achieved using the DEM derived from InSAR technique.(5)This paper were that compared to the single polarization L-band PALSAR or C-band Radarsat-2 images,wetland vegetation classifications improved overall accuracy between 7%~22% by the integration of ZY-3,PALSAR and Radarsat-2 images,and achieved 94.15% overall accuracy at 95% confidence level;The DEM produced by the L-band PALSAR images using InSAR technique performed better than used C-band Sentinel-1A or X-band TerraSAR,and when compared to SRTM DEM data product.In 111 points randomly selected from 1:10000 topographic map,there are 85 points with a little differences in elevation value,which was less than 3 meters;the relative change of water level of wetland was consistent between filed measurements and remote sensing observation,and InSAR technique is available to monitor centimeter-level water level change of wetland;The elevation of hydrological boundary produced by water level depth-duration-frequency threshold was between 51.7 and 52.4 m.Delineation of wetland scopes covered all nests of Ciconia Boyciana,and was consistent in in spatial distribution.Fluctuate situation at the hydrological boundary was identical to hydrological environment formed wetland vegetation and soil types in the plots.
Keywords/Search Tags:Wetland hydrology, Water level depth-duration-frequency, Wetland vegetation, Polarimetric SAR target decomposition, Interferometric Synthetic Aperture Radar(InSAR), Random Forest algorithm
PDF Full Text Request
Related items