Aspects of in situ angular scattering measurements in contrasting waters

Abstract

Rapid changes are observed in oceanic and coastal environments around the world due to global temperatures increases, ocean acidification and changing weather patterns - anthropogenic climate change. These changes have large effects on the ecosystems of the ocean. In order to understand the effects and possibly mitigate their consequences, it is necessary to increase and improve the environmental monitoring of the ocean. Optical properties of natural waters within the visible spectrum is closely linked to properties of phytoplankton, the foundation of oceanic ecosystems, as well as other particles on the micrometer and sub-micrometer scale in the water mass. Optical measurements can thus give us valuable information about the particle content of the water and the state of the ecosystem.

The volume scattering function (VSF) is a fundamental optical property describing how much light is scattered by a medium and in what direction the light is scattered. In natural waters, by far most of the light is scattered in the very forward direction, which makes it technically challenging to measure the VSF. The LISST-VSF is the first commercially available instrument for field measurements of the VSF over a large angular domain. To trust the measurements, it is important to validate the performance of instrument and identify any error sources, in particular the valid range of the instrument, given that scattering coefficients of natural waters can span three orders of magnitude.

In this thesis, I have characterised LISST-VSF measurements using both in situ measurements of highly contrasting water types, controlled laboratory measurements, and Monte Carlo simulations of instrument geometry. Similar aspects have been investigated for the LISST-200X, which measures the VSF at angles 0.04-13˚ at 670 nm.

In Paper I, these two instruments are calibrated and validated using polymer beads and in situ measurements spanning from clear waters on the North Pole to highly turbid glacial meltwater. The measurements demonstrated that due to instrument design, the LISST-200X only gives valid scattering measurements in moderate-to-turbid waters. The LISST-VSF gives valid measurements also in clear waters (however with a loss in precision), but is limited by multiple scattering errors in more turbid waters.

Multiple scattering effects on LISST-VSF measurements are investigated in detail in in Paper II and III. For this purpose, a Monte Carlo simulation was developed and validated with experimental data, and subsequently used to simulate LISST-VSF measurements with Fournier-Forand and Henyey-Greenstein phase functions. We demonstrated that the multiple scattering can yield uncertainties of 10% when the scattering coefficient is 1 m-1, which significantly restricts the accurate measurement range of LISST-VSF. LISST-200X is less affected by this error due to its shorter path length.

Scattering can be an error source for other optical measurements as well. In Paper IV, we attempt to correct in situ depth profiles of absorption coefficients measured with the ac-s instrument using VSF measurements collected with the LISST-VSF in coastal waters. We show that this method does not show a clear and consistent improvement over existing methods, which are simpler to use but make strong assumptions about absorption and scattering properties. The discrepancies in the VSF correction can be attributed to several different confounding factors, such as spatial variability and multiple scattering, which are exceedingly propagated to the corrected absorption values. Nevertheless, VSF measurements are found useful to analyze the scattering error. We show that the VSF between 5-90˚ can contribute significantly to the scattering error, depending on the phase function and the reflectance efficiency of the reflective tube. Moreover, by simulating the VSF wavelength-dependency using Mie theory, we show that particle sub-populations with diameters close to the wavelength can explain why scaling the scattering error with the scattering coefficient sometimes fails.