CAJ | 학술논문

为了优化生物光学模型中的悬浮物后向散射模型系数（bbx）和双向反射因子（f/Q），基于珠江口、韩江河口、徐闻珊瑚礁保护区实测数据（遥感反射率、叶绿素a质量浓度、悬浮物质量浓度、CDOM吸收系数），采用模拟退火算法同时优化了bbx以及f/Q（N＝43），并开发了生物光学正演模型（N＝22）。文章优化的f/Q为0.1049，悬浮物在参考波长531 nm处后向散射系数bbx为0.268×[TSS]0.295。通过输入实测的水质数据模拟出531 nm处的遥感反射率，并外推到其他波段（412、443、490、551、667 nm）。模拟得到的531 nm处遥感反射率与实测值的均方根误差RMSE＝0.0008，N＝22，除了模拟的667 nm处遥感反射率误差较大外（RMSE＝0.0036，N＝22），其他波段都具有较好的精度（RMSE＜0.0023）。结果表明，该研究区与其他研究区（如黄、东海）的bbx和f/Q不一致，这2个参数需要根据研究区的数据进行优化。验证结果说明迭代优化的bbx以及f/Q能够适用于广东近岸河口生物光学参数模拟。
위료우화생물광학모형중적현부물후향산사모형계수（bbx）화쌍향반사인자（f/Q），기우주강구、한강하구、서문산호초보호구실측수거（요감반사솔、협록소a질량농도、현부물질량농도、CDOM흡수계수），채용모의퇴화산법동시우화료bbx이급f/Q（N＝43），병개발료생물광학정연모형（N＝22）。문장우화적f/Q위0.1049，현부물재삼고파장531 nm처후향산사계수bbx위0.268×[TSS]0.295。통과수입실측적수질수거모의출531 nm처적요감반사솔，병외추도기타파단（412、443、490、551、667 nm）。모의득도적531 nm처요감반사솔여실측치적균방근오차RMSE＝0.0008，N＝22，제료모의적667 nm처요감반사솔오차교대외（RMSE＝0.0036，N＝22），기타파단도구유교호적정도（RMSE＜0.0023）。결과표명，해연구구여기타연구구（여황、동해）적bbx화f/Q불일치，저2개삼수수요근거연구구적수거진행우화。험증결과설명질대우화적bbx이급f/Q능구괄용우엄동근안하구생물광학삼수모의。
The optimization of bio-optical algorithm parameters in remote sensing of CASE-II water is a necessary step before they can be applied, because these parameters vary with the changes of study area. Biochemical parameters (Chlorophyll a, Total suspended solids, CDOM absorptioncoefficient)and above-water surface reflectance collected inthePearl River Estuary, Hanjiang Estuary and Xuwen Coral Reef Zone were used to iterate the back scattering coefficient model of total suspended solids (bbx)and the bidirectional reflectance (f/Q) by simulated annealing algorithm (N=43). The calibration results indicated that the f/Q was 0.1049, and thebbx at the wavelength of 531nm was 0.268×[TSS]0.295. Then, the forward model was developed to test the optimized results, i.e. the remote sensing reflectance was simulated by inputting water quality parameters (N=22). The validation results showed that reflectance obtained from the optimized algorithm agreed well with field observations(RMSE=0.0008,N=22). The remote sensing reflectance at the wavelength of 667nm extrapolated from the Bio-optical Algorithm had poor accuracy (RMSE=0.0036,N=22)compared with the field data , while all the other band (412,443,490,551 nm)had good precision (RMSE<0.0023,N=22). It showed thatbbx andf/Q varied with different study areas, and they should be optimized according to the measured data. Thebbx and f/Q values obtained from the optimized algorithm can be applied to bio-optical algorithm.