Remotely estimating total suspended solids concentration in clear to extremely turbid waters using a novel semi-analytical method
Remotely estimating total suspended solids concentration in clear to extremely turbid waters using a novel semi-analytical method
复制标题
DOI：
10.1016/j.rse.2021.112386
复制
发表时间：
2021-03-14
影响因子：
13.5
通讯作者：
O 'Donnell, David
中科院分区：
文献类型：
--
作者：
Jiang, Dalin;Matsushita, Bunkei;O 'Donnell, David
总悬浮物（TSS）浓度是水质管理和泥沙输移研究的重要地球化学参数。在这项研究中，我们提出了一种新的半分析方法估计TSS在明确的极端浑浊的沃茨从遥感反射率（R-rs）。该方法包括三个子算法顺序使用。首先，通过比较490、560、620和754 nm处的R-rs值，将遥感沃茨分为清澈（I型）、中度浑浊（II型）、高度浑浊（III型）和极度浑浊（IV型）水体类型。其次，使用特定于每种水类型的半分析模型来确定在相应的单个波长（即，I型为560 nm，II型为665 nm，III型为754 nm，IV型为865 nm）。第三，在每种水类型中使用TSS和相应波长处的B（bp）之间的特定关系。与其他现有的方法不同，该方法是严格的半分析，其子算法仅使用合成数据集开发。使用模拟的（N = 1000，TSS范围从0.01到1100 g/m（3））和现场测量的（N = 3421，TSS范围从0.09到2627 g/m（3））对R-rs和TSS，将所提出的方法的性能与其他三种最先进的方法的性能进行比较。结果显示，模拟数据的中位绝对百分比误差（MAPE）为16.0%与30.2-90.3%，原位数据的中位绝对百分比误差（MAPE）为39.7%与45.9-58.1%。新方法随后被应用于175中分辨率成像光谱仪（MERIS）和498海洋和陆地颜色仪器（OLCI）在2003-2020年的时间框架内获得的图像，以产生长期的TSS时间序列的日本的Suwa湖和霞浦湖。使用MERIS和OLCI配对进行的性能评估显示与现场TSS测量结果具有良好的一致性。
Total suspended solids (TSS) concentration is an important biogeochemical parameter for water quality management and sediment-transport studies. In this study, we propose a novel semi-analytical method for estimating TSS in clear to extremely turbid waters from remote-sensing reflectance (R-rs). The proposed method includes three sub-algorithms used sequentially. First, the remotely sensed waters are classified into clear (Type I), moderately turbid (Type II), highly turbid (Type III), and extremely turbid (Type IV) water types by comparing the values of R-rs at 490, 560, 620, and 754 nm. Second, semi-analytical models specific to each water type are used to determine the particulate backscattering coefficients (b(bp)) at a corresponding single wavelength (i.e., 560 nm for Type I, 665 nm for Type II, 754 nm for Type III, and 865 nm for Type IV). Third, a specific relationship between TSS and b(bp) at the corresponding wavelength is used in each water type. Unlike other existing approaches, this method is strictly semi-analytical and its sub-algorithms were developed using synthetic datasets only. The performance of the proposed method was compared to that of three other state-of-the-art methods using simulated (N = 1000, TSS ranging from 0.01 to 1100 g/m(3)) and in situ measured (N = 3421, TSS ranging from 0.09 to 2627 g/m(3)) pairs of R-rs and TSS. Results showed a significant improvement with a Median Absolute Percentage Error (MAPE) of 16.0% versus 30.2-90.3% for simulated data and 39.7% versus 45.9-58.1% for in situ data, respectively. The new method was subsequently applied to 175 MEdium Resolution Imaging Spectrometer (MERIS) and 498 Ocean and Land Colour Instrument (OLCI) images acquired in the 2003-2020 timeframe to produce long-term TSS time-series for Lake Suwa and Lake Kasumigaura, Japan. Performance assessments using MERIS and OLCI matchups showed good agreements with in situ TSS measurements.