A SWIR based algorithm to retrieve total suspended matter in extremely turbid waters
Knaeps, E.; Ruddick, K.G.; Doxaran, D.; Dogliotti, A.I.; Nechad, B.; Raymaekers, D.; Sterckx, S. (2015). A SWIR based algorithm to retrieve total suspended matter in extremely turbid waters. Remote Sens. Environ. 168: 66-79. http://dx.doi.org/10.1016/j.rse.2015.06.022
In: Remote Sensing of Environment. Elsevier: New York,. ISSN 0034-4257; e-ISSN 1879-0704, meer
| |
Trefwoorden |
|
Author keywords |
Short Wave Infra Red; Water; Extremely turbid; Total Suspended Matter |
Auteurs | | Top |
- Knaeps, E., meer
- Ruddick, K.G., meer
- Doxaran, D.
- Dogliotti, A.I., meer
|
|
|
Abstract |
In ocean colour remote sensing, the use of Near Infra Red (NIR) spectral bands for the retrieval of Total Suspended Matter (TSM) concentration in turbid and highly turbid waters has proven to be successful. In extremely turbid waters (TSM > 100 mg L- 1) however, these bands are less sensitive to increases in TSM. Here it is proposed to use Short Wave Infra Red (SWIR) spectral bands between 1000 and 1300 nm for these extreme cases. This SWIR spectral region is subdivided into two regions, SWIR-I (1000 nm to 1200 nm) and SWIR-II (1200 nm to 1300 nm) which correspond to local minima in the pure water absorption spectrum. For both spectral regions the water reflectance signal was measured in situ with an ASD spectrometer in three different extremely turbid estuarine sites: Scheldt (Belgium), Gironde (France), and Río de la Plata (Argentina), along with the TSM concentration. A measurable water reflectance was observed for all sites in SWIR-I, while in the SWIR-II region the signal was not significant compared to the Signal-to-Noise Ratio (SNR) of current Ocean Colour (OC) sensors. For the spectral band at 1020 nm (present in Ocean and Land Colour Instrument — OLCI, onboard Sentinel-3) and at 1071 nm, an empirical single band TSM algorithm is defined which is valid for both the Gironde and Scheldt estuarine sites. This means that a single algorithm can be applied for both sites without expensive recalibration. The relationship between TSM and SWIR reflectance at 1020 and 1071 nm is linear and did not show any saturation for the concentrations measured here (up to 1400 mg L- 1), while saturation was observed for the NIR wavelengths, as expected. Hence, for extremely turbid waters it is advised to switch from NIR to SWIR-I wavelengths to estimate TSM concentration. This was demonstrated for an airborne hyperspectral dataset (Airborne Prism Experiment, APEX) from the Gironde estuary having several spectral bands in the SWIR-I. The empirical single band SWIR TSM algorithm was applied to the atmospherically corrected scene providing a TSM concentration map of the Gironde from mouth to more upstream with concentrations expected in this region ranging from a few to several hundreds mg L- 1. These results, i.e. the existence of a single relationship for the Scheldt and Gironde, not showing any decrease of sensitivity, highlights the importance of having SWIR bands in future ocean colour sensors for studying extremely turbid rivers, coastal areas and estuaries in the world. A further implication of these results is that there is a TSM limit for application of atmospheric correction algorithms which assume zero SWIR marine reflectance. That limit is defined here as function of wavelength and sensor noise level. |
|