CAJ | 학술논문

浮游藻类的后向散射是水体光谱构成的重要组成部分，作为水体辐射传输模型中的重要参数，高精度的藻类后向散射系数对水体叶绿素a浓度的遥感反演精度至关重要。本文以简化的辐射传输模型-生物光学模型为基础，尝试性分离了太湖浮游藻类的后向散射系数。通过藻类后向散射规律分析，建立了浮游藻类吸收、后向散射特征的叶绿素a反演模型，改善了叶绿素a浓度的遥感反演精度。分析表明：藻类颗粒物的后向散射系数与吸收系数之间存在反比关系，且在560 nm、700 nm附近存在明显的散射峰，与叶绿素a浓度之间相关性显著；低密度藻类水体总悬浮颗粒的后向散射以非色素颗粒为主导，适合采用经典的指数模型模拟后向散射系数，而藻类密度较高的富营养化水体，水体总悬浮颗粒的后向散射以藻类颗粒为主导，传统的指数模型已不适用；采用分离藻类后向散射系数的方法，使得叶绿素a浓度的反演值与真实值相关系数从0.66提高到0.98，相对误差从55%降低到38%，均方根误差（RMSE）也由60.95μg/L降低至13.98μg/L。其真实性检验表明，与经典指数模型方法相比，利用藻类颗粒后向散射分离方法反演叶绿素a浓度，能够显著改善反演精度。
부유조류적후향산사시수체광보구성적중요조성부분，작위수체복사전수모형중적중요삼수，고정도적조류후향산사계수대수체협록소a농도적요감반연정도지관중요。본문이간화적복사전수모형-생물광학모형위기출，상시성분리료태호부유조류적후향산사계수。통과조류후향산사규률분석，건립료부유조류흡수、후향산사특정적협록소a반연모형，개선료협록소a농도적요감반연정도。분석표명：조류과립물적후향산사계수여흡수계수지간존재반비관계，차재560 nm、700 nm부근존재명현적산사봉，여협록소a농도지간상관성현저；저밀도조류수체총현부과립적후향산사이비색소과립위주도，괄합채용경전적지수모형모의후향산사계수，이조류밀도교고적부영양화수체，수체총현부과립적후향산사이조류과립위주도，전통적지수모형이불괄용；채용분리조류후향산사계수적방법，사득협록소a농도적반연치여진실치상관계수종0.66제고도0.98，상대오차종55%강저도38%，균방근오차（RMSE）야유60.95μg/L강저지13.98μg/L。기진실성검험표명，여경전지수모형방법상비，이용조류과립후향산사분리방법반연협록소a농도，능구현저개선반연정도。
As key parameters in bio-optical model, the backscattering coefficients of phytoplankton plays an im-portant role in modelling the reflectance spectra and retrieving chlorophyll-a (CHL-a) concentrations from eutro-phic water. An exponential model is usually used to simulate the total backscattering coefficients by omitting the phytoplankton backscattering in inland water characterized by lower concentrations of CHL-a. However, the ex-ponential model is not valid for inland water with high CHL-a concentrations, and high relative errors and resi-dues may exist in retrieving the CHL-a concentrations in the algae blooming area, due to the errors made by omitting or introducing inaccurate backscattering coefficients of phytoplankton. Therefore, a precise determina-tion of the phytoplankton backscattering coefficients is of great importance in retrieving chlorophyll-a concentra-tions. Based on the classical bio-optical model, we proposed a method to partition the phytoplankton backscatter-ing coefficients. The variations of the backscattering coefficients of phytoplankton particles with wavelengths in 400~700 nm and the chlorophyll-a concentrations are illustrated and discussed in details. According to the results mentioned above, following conclusions are drawn:(a) it is appropriate to model the total backscattering coeffi-cients by using exponential function in most Case 2 waters with lower concentrations of Chlorophyll-a, where the non-algal suspended sediments dominated the optical properties. However, it is not applicable in eutrophic waters with higher concentrations of chlorophyll-a, where the algal particles dominated the optical properties;(b) phytoplankton backscattering coefficients vary inversely to their absorption coefficients, and two backscatter-ing peaks emerge in the wavelengths of 560nm and 700nm, which are significantly correlated with CHL-a con-centrations; (c) compare to the exponential model, the accuracy of the bio-optical model using the partitioned phytoplankton backscattering coefficients has improved greatly:the correlation coefficient between the retrieved and the measured CHL-a is increased from 0.66 to 0.98, the average relative error decreases from 55%to 38%, and the RMSE decreases from 60.95 to 13.98 in estimating CHL-a concentrations.