Developing a two-step retrieval method for estimating total suspended solid concentration in Chinese turbid inland lakes using Geostationary Ocean Colour Imager (GOCI) imagery

Abstract

Total suspended solid (TSS) concentration is an important water quality parameter. Mapping its varying distribution using satellite images with high temporal resolution is valuable for studying suspended sediment transportation and diffusion patterns in inland lakes. A total of 255 sites were used to make remote-sensing reflectance measurements and surface water sampling at four Chinese inland lakes, i.e. Taihu Lake, Chaohu Lake, Dianchi Lake, and the Three Gorges Reservoir, at different seasons. A two-step retrieval method was then developed to estimate TSS concentration for contrasting Chinese inland lakes, which is described in this article. In the first step, a cluster method was applied for water classification using eight Geostationary Ocean Colour Imager (GOCI) channel reflectance spectra simulated by spectral reflectance measured by an Analytical Spectral Devices (ASD) Inc. spectrometer. This led to the classification of the water into three classes (1, 2, and 3), each with distinct optical characteristics. Based on the water quality, spectral absorption, and reflectance, the optical features in Class 1 were dominated by TSS, while Class 3 was dominated by chl-a and the optical characteristics of Class 2 were dominated jointly by TSS and chl-a. In the second step, class-specific TSS concentration retrieval algorithms were built. We found that the band ratio Band 8/Band 4 was suitable for Class 1, while the band ratio of Band 7/Band 4 was suitable for both Class 2 and Class 3. A comprehensive determination value, combining the spectral angle mapper and Euclidean distance, was adopted to identify the classes of image pixels when the method was applied to a GOCI image. Then, based on the pixel's class, the class-specific retrieval algorithm was selected for each pixel. The accuracy analysis showed that the performance of this two-step method was improved significantly compared to the unclassed method: the mean absolute percentage error decreased from 38.9% to 24.3% and the root mean square error decreased from 22.1 to 16.5 mg l -1 . Finally, the GOCI image acquired on 13 May 2013 was used as a demonstration to map the TSS concentration in Taihu Lake with a reasonably good accuracy and highly resolved spatial structure pattern.

Evaluation of chlorophyll-a retrieval algorithms based on MERIS bands for optically varying eutrophic inland lakes

“…However, it does not work for differentiating water types II and III because of their similar spectral shapes. As a result, a comprehensive parameter, G Lyu et al, 2015), which combines the SAM and Euclidean Distance, was adopted to determine the water type of MERIS image pixels, as given in Eq. (8):…”

Developments in Earth observation for the assessment and monitoring of inland, transitional, coastal and shelf-sea waters

“…SEVIRI was not designed primarily for applications over water but several studies have shown the feasibility of retrieving TSM in turbid waters (Neukermans et al, 2009;Neukermans et al, 2012). The GOCI sensor, however, is specifically designed for ocean colour applications and thus can be used to estimate Chl-a concentrations and at a spatial resolution that also allows data to be acquired for systems ranging in scale from shelf-seas to large lakes (Lyu et al, 2015).…”

High Temporal Resolution Monitoring of Suspended Matter Changes from GOCI Measurements in Lake Taihu

“…Although GOCI was designed for ocean waters [20,21], many studies have evidenced that it also can be applied to inland waters. For example, some scholars developed a new method for estimating the total suspended solid concentration in Chinese turbid inland lakes using GOCI data [22]. Others used GOCI data to estimate the chlorophyll concentrations in turbid inland lakes [23,24].…”