Algorithm to derive inherent optics properties from remote sensing reflectance in turbid and eutrophic lakes

Xue Kun; Boss Emmanuel; Ma Ronghua; Shen Ming

摘要

Inherent optical properties play an important role in understanding the biogeochemical processes of lakes by providing proxies for a variety of biogeochemical quantities, including phytoplankton pigments. However, to date, it has been difficult to accurately derive the absorption coefficient of phytoplankton [a(ph)(lambda)] in turbid and eutrophic waters from remote sensing. A large dataset of remote sensing of reflectance [R-rs(lambda)] and absorption coefficients was measured for samples collected from lakes in the middle and lower reaches of the Yangtze River and Huai River basin (MLYHR), China. In the process of scattering correction of spectrophotometric measurements, the particulate absorption coefficients [a(p)(lambda)] were first assumed to have no absorption in the near-infrared (NIR) wavelength. This assumption was corrected by estimating the particulate absorption coefficients at 750 nm [a(p)(750)] from the concentrations of chlorophyll-a (Chla) and suspended particulate matter, which was added to the a(p)(lambda) as a baseline. The resulting mean spectral mass-specific absorption coefficient of the nonalgal particles (NAPs) was consistent with previous work. A novel iterative IOP inversion model was then designed to retrieve the total nonwater absorption coefficients [a(nw)(lambda)] and backscattering coefficients of particulates [b(b)(p)(lambda)], a(ph)(lambda), and a(dg)(lambda) [absorption coefficients of NAP and colored dissolved organic matter (CDOM)] from R-rs(lambda) in turbid inland lakes. The proposed algorithm performed better than previously published models in deriving a(nw)(lambda) and b(b)(p)(lambda) in this region. The proposed algorithm performed well in estimating the a(ph)(lambda) for wavelengths > 500 nm for the calibration dataset [N = 285, unbiased absolute percentage difference (UAPD) = 55.22%, root mean square error (RMSE) = 0.44 m(-1)] and for the validation dataset (N = 57, UAPD = 56.17%, RMSE = 0.71 m(-1)). This algorithm was then applied to Sentinel-3A Ocean and Land Color Instrument (OLCI) satellite data, and was validated with field data. This study provides an example of how to use local data to devise an algorithm to obtain IOPs, and in particular, a(ph)(lambda), using satellite R-rs(lambda) data in turbid inland waters. (C) 2019 Optical Society of America

机译：固有的光学特性在理解湖泊的生物地球化学过程中，通过提供各种生物地球化学量，包括浮游植物颜料，起着重要作用。然而，迄今为止，难以准确地从遥感中获得浑浊和富营养化水中的浮游植物[A（pH）（Lambda）的吸收系数。测量了从长江和淮河流域（Mlyhr）中下游的湖泊收集的样品测量反射率[R-RS（Lambda）]和吸收系数的大型数据集。在分光光度测量的散射校正的过程中，首先假设颗粒状吸收系数[a（p）（λ）]在近红外（nir）波长中没有吸收。通过将750nm [A（P）（750）的颗粒吸收系数从叶绿素-A（CHLA）和悬浮颗粒物质的浓度估计，将该假设校正，并将其加入到A（P）（Lambda）中作为基线。所得的扁平颗粒（NAP）的平均光谱分子特异性吸收系数与先前的工作一致。然后设计一种新型迭代IOP反转模型，用于检测颗粒的总非水吸收系数[A（NW）（Lambda）]和反向散射系数[B（b）（p）（lambda）]，a（pH）（lambda）和浑浊的内陆湖中R-RS（Lambda）的A（DG）（Lambda）[吸收系数] [午睡和有色溶解有机物（CDOM）]。所提出的算法比以前公布的模型更好地在该区域中衍生出（NW）（Lambda）和B（B）（B）（P）（Lambda）。在估计波长的A（pH）（Lambda）> 500nm的校准数据集的算法[n = 285，无偏见的绝对百分比差（UAPD）= 55.22％，根均线误差（RMSE）= 0.44米（-1）]和验证数据集（n = 57，UAPD = 56.17％，RMSE = 0.71 m（-1））。然后将该算法应用于Sentinel-3a海洋和土地彩色仪器（OLCI）卫星数据，并用现场数据验证。本研究提供了如何使用本地数据来设计算法以获得IOPS，特别是（pH）（Lambda），使用浑浊内陆水域中的数据。（c）2019年光学学会

Algorithm to derive inherent optics properties from remote sensing reflectance in turbid and eutrophic lakes

摘要

著录项

相关主题

期刊订阅