Research On The Complexity Of Optical Property Of Main Vegetation In The Wetland Of The Yangtze River Estuary Based On Three-dimensional Radiative Transfer Model

Wetland is a special ecosystem which is affected by both water and land,and it is an important natural resource.The Yangtze River Estuary wetland represented by Chongming Island is an important tidal flat wetland in our country,with complex hydrology,soil and vegetation structure,but its native species such as Phragmites australis are deeply affected by the invasion of alien species S.alterniflora,and biodiversity is facing severe challenge.The use of remote sensing technology with a wide monitoring range,large amount of information,and high timeliness to carry out wetland vegetation monitoring has become a powerful means for rapid acquisition of large-area wetland vegetation information,but it is often affected by the variety of wetland vegetation,complex vegetation structures,and changes in the underlying surface background.Due to the influence of factors,its optical properties are more complex than other ground features,and there are greater uncertainties in the identification of wetland vegetation types and the quantitative inversion of their physiological and biochemical parameters.At the same time,the accessibility of tidal flat wetlands is poor,and it is difficult to obtain field observation data.Also,the underlying surface is significantly affected by tidal action and micro-topography.Researches about radiation transfer processes and mechanisms need to be pushed,and traditional canopy radiation transfer models are difficult to consider the optical properties of wetland vegetation complexity.In response to the above problems,this study took the Liuyao north of Chongming Island in Shanghai which belongs to Yangtze River estuary salt marsh region as the research area,and took the typical native species of P.australis and the invasive species S.alterniflora as the research objects.The three-dimensional radiation transfer model can be used to model complex scenarios in detail,and to carry out the advantages of high-precision radiation transfer process simulation to explore the origin of the complexity of the canopy spectral.The DART(Discrete Anisotropic Radiative Transfer)model was selected to carry out multiple scene modeling and radiation transfer simulation of two types of wetland species in the study area to reveal the complexity of wetland canopy optical properties caused by plant structure characteristics,vegetation population characteristics and mixed community characteristics.The main research content and research results are as follows:1)Facing the 3-D fine modeling of wetland vegetation scenes.We designed and conducted field campaign experiments for P.australis and S.alterniflora.The 3-D morphological data and population distribution characteristics of the two plants in different periods were obtained,and the characteristics of different leaf positions were observed.Using the spectra of leaves,canopy and underlying surface to establish a measured database.The experimental results show that the growth morphology of the two plants at different periods is quite different.The population distribution is significantly affected by the tidal effect of the underlying surface and the micro-topography.The plant density changes significantly in different regions and the leaf spectra of different in height layers are significantly various.While,there is a lack of scientific statistics and accurate description of the average spectrra of canopy leaves in present studies.After a large number of comparisons with canopy spectra at different locations in the study area,it is believed that the above-mentioned differences,which are in plant morphology,canopy structure and population scenarios may be important factors that result the complexity of wetland canopy spectra.2)We proposed a modeling method and parameter selection index that both take into account the radiation transfer simulation efficiency and the simulation accuracy in the process of detailed scenarios modeling when combined the characteristics of the plant morphology and the actual situation of the study area.It indicated that the simulation accuracy loss is controllable under the premise through appropriate simplification of the plant's 3-D model and appropriate adjustment of the cell dimension size in exchange for a higher computing efficiency.Then we provided the optimal parameter combination.Taking a 1m*1m square high-density(120 plants/m~2)scene as an example,the optimal parameter combination of the P.australis scene is:the simplified 4 times model(P4)with the 0.03m cell size;the optimal parameter combination of S.alterniflora is:the most Simplified model(S5)with 0.05m cell dimension size.3)Experiments revealed that the accurate description of the difference in leaf spectrra at various heights of the P.australis canopy and the accurate description of the average leaf spectrum are significant factors affecting the simulation of near-infrared reflectance,however this factor is often ignored in many studies.The influence mechanism mentioned before is analyzed from the perspective of radiation transmission,then use the k-means method to scientifically distinguish and effectively classify the individual P.australis leaves in the vertical space attribution layer.After this,the model simulated near-infrared reflectance can reach a significant consistency improve compared with the measured reflectance.Field measurements have shown that the improved simulated spectra of the P.australis canopy overlaps with the measured spectra almost completely,and the correlation can reach 99.50%.The full-band RMSE is reduced by 83.66%and the near-infrared band RMSE is reduced by 25.99%compared with that before the improvement.4)Focusing on the three main influencing factors of plant growth density,mixed growth conditions,and underlying surface differences in the study area,we design 5 density gradients,5 mixing ratios,and 2 underlying surface backgrounds,to sum up to a total of 50 simulated quadrats.Thus analyzing the response of quadra canopy reflectance to various factors with every simulated sample.The results of the study screened out the sensitive band and spectral index suitable for distinguishing the population density of P.australis and S.alterniflora population,the density of P.australis-S.alterniflora mixed community and their mixing ratio.It provides a research foundation for solving the phenomenon of homogeneity and heterogeneity caused by underlying surface,population density differences and mixed growth.5)Based on 50 simulated sample plots,we try to establish a simulated pixel reflectance database corresponding to the real environment of the study area.Then innovatively reverse the establishment of a 3D scenarios of the salt marsh wetland in the study area through the consistency check with the satellite image pixel SAM and Euclidean distance.This method provides theoretical methods and technical support for the efficient 3D reconstruction of the real environment of the salt marsh wetland in the Yangtze River Estuary,and provides a new idea for the 3-D radiation transfer model to obtain remote sensing simulation data of multiple angles,arbitrary bands and spatial resolutions in this environment.