Remote Sensing Inversion Of Phytoplankton Particle Size Classes And Its Spatial-temporal Characteristics In The Pearl River Estuary

Phytoplankton are the most important primary producers in the ocean,and their biomass and distribution characteristics are one of the most fundamental scientific questions in biological oceanography.Phytoplankton size class(PSC)is an important parameter to characterize the functional type of phytoplankton,and its variation can directly affect the energy transfer and carbon cycle of the whole ecosystem.In the Pearl River Estuary,the existing phytoplankton particle size studies are mainly based on the interpolation of the measured data to obtain the overall spatial distribution,and the inversion of phytoplankton particle size using remote sensing data is rare,which cannot obtain the phytoplankton particle size distribution in a large scale,high timeliness and strong continuity.In this paper,the phytoplankton particle size in the Pearl River Estuary is studied by combining multi-source data such as ground hyperspectral data,public cruise data and remote sensing image data.Firstly,the continuous wavelet transform is introduced to achieve the highprecision inversion of chlorophyll a concentration;secondly,the phytoplankton particle size class inversion model is improved regionally to obtain the phytoplankton particle size distribution products in the Pearl River Estuary for four seasons from 2019 to 2020,and its spatial and temporal distribution characteristics are analyzed.The main research contents and results are as follows:(1)High-precision chlorophyll a concentration inversion.The existing abundance PSC inversion algorithm requires input chlorophyll a as the independent variable.This paper compares and analyzes various conventional methods for chlorophyll a concentration inversion based on the measured hyperspectral signals and simultaneous water quality detection data from several water bodies(rivers and reservoirs)within the Pearl River Estuary and its upstream waters of Class II.Then,the continuous wavelet transform was introduced to achieve highprecision inversion of chlorophyll a concentration.The experimental results showed that the quantitative inversion model based on the sym6 wavelet basis function transform performed best,=0.732,=6.457,=2.600.Compared with the conventional threeband inversion method,increased by 0.052 and decreased by 2.797.This method can effectively extract the hidden chlorophyll information from the hyperspectral signal and is suitable for estimating chlorophyll a concentration in ground hyperspectral data.(2)Regionalized PSC model improvement.Regionalized PSC model improvement based on in situ observation data from two Pearl River Estuary cruises.The finalized model parameters ? ? and were 0.1371,0.2684,0.9308,and 2.909,respectively,which improved the mean value of RPD by 0.389,reduced the mean value of RMSE by 0.051,and reduced the mean value of MAPE by 8.962% compared with the existing model.(3)The spatial and temporal distribution of PSC in the Pearl River Estuary was characterized.Based on the above research results,the spatial and temporal distribution of PSC in the Pearl River Estuary was analyzed,and the results of PSC inversion showed that the dominant population in the nearshore area of the northern part of the Pearl River Estuary was mainly Micro-class,and the proportion was decreasing from north to south;the phytoplankton in the southern offshore area was mainly Nano-class;the proportion of Pico was less than 0.2overall,and the proportion increased from north to south,with some differences in different seasons.There are some differences between seasons.The proportion of Micro level is more distributed in the middle and high values in spring,autumn and winter,and the proportion in summer is mainly in the middle and low values;Nano level is mainly in the middle and low values(0.2-0.6)in spring,autumn and winter,and the proportion in summer is mainly in the middle and high values;Pico level is lower than 0.2 in the whole year,and its proportion in spring,autumn and winter is mainly in the range of 0.1-0.6.It is mainly in the range of 0.1-0.2,and the main percentage value in summer is 0-0.1.