Application of the Support Vector Regression Method for Turbidity Assessment with MODIS on a Shallow Coral Reef Lagoon (Voh-Koné-Pouembout, New Caledonia)

Abstract

Abstract:Particle transport by erosion from ultramafic lands in pristine tropical lagoons is a crucial problem, especially for the benthic and pelagic biodiversity associated with coral reefs. Satellite imagery is useful for assessing particle transport from land to sea. However, in the oligotrophic and shallow waters of tropical lagoons, the bottom reflection of downwelling light usually hampers the use of classical optical algorithms. In order to address this issue, a Support Vector Regression (SVR) model was developed and tested. The proposed application concerns the lagoon of New Caledonia-the second longest continuous coral reef in the world-which is frequently exposed to river plumes from ultramafic watersheds. The SVR model is based on a large training sample of in-situ turbidity values representative of the annual variability in the Voh-Koné-Pouembout lagoon (Western Coast of New Caledonia) during the 2014-2015 period and on coincident satellite reflectance values from MODerate Resolution Imaging Spectroradiometer (MODIS). It was trained with reflectance and two other explanatory parameters-bathymetry and bottom colour. This approach significantly improved the model's capacity for retrieving the in-situ turbidity range from MODIS images, as compared with algorithms dedicated to deep oligotrophic or turbid waters, which were shown to be inadequate. This SVR model is applicable to the whole shallow lagoon waters from the Western Coast of New Caledonia and it is now ready to be tested over other oligotrophic shallow lagoon waters worldwide.

Rain-derived particles and CDOM distribution along the east coast of New Caledonia

“…On this figure, MODIS turbidity values in coastal bays are high (they are overestimated by a factor of 20 due to bottom effect [7]). Such overestimated high values are observed on shallow bays such as North of the Bay of Canala on the East coast, or in the Bay South of Koumac on the Northern coral coast which are particularly shallow (< 10 meters) Apart from these enclosed shallow bays, as the lagoon is deeper than 10 meters, turbidity distribution does not present any overestimation and corresponds well to in situ measurements.…”

“…Recent studies based on in situ data [1] have shown that NASA algorithms overestimate Chla in shallow waters of lagoons and underestimate it offshore ( [2,3,4]). Thus Chla can be estimated by regional algorithms (for water depths > 20 meters) [5,6] and other methods for shallow waters (< 20 m) [7]. Turbidity can be estimated through various parameterizations and was measured and estimated along the Eastern coast of New Caledonia [8] or [10].…”

Why and How Do We Study Sediment Transport? Focus on Coastal Zones and Ongoing Methods

“…Finally, we note the increasing use of statistical methods of artificial intelligence or approaching them, adapted to process a growing amount of data: neural networks for the interpretation of concentration and turbulence profiles (e.g., to derive floc sizes and promote the use of artificial neural network to study flocculation, [306]) or in the analysis of sediment transport in water basins [307,308]; fuzzy logic; fusion methods and machine learning (e.g., in the concurrent analysis of forcing parameters, suspended sediment profiles and sediment concentration at the surface provided by remote sensing, [309]); Random Forest for classifications (e.g., to determine a mineralogical classification of suspended sediments in rivers after using scanning electron microscopy, [310,311]); support vector regression methods in satellite data analysis (e.g., [312], this Special Issue), etc.…”

“…In their study, Wattelez et al [312] introduced a Support Vector Regression (SVR) method and tested its capacity to map the turbidity distribution in a part of this coral reef lagoon. The model was trained with a large dataset of in situ turbidity, on coincident reflectance values from MODIS and on two other explanatory parameters: bathymetry and bottom colour.…”

Construction of Multi-Year Time-Series Profiles of Suspended Particulate Inorganic Matter Concentrations Using Machine Learning Approach

“…When time is a factor, these data-driven approaches can also use analog Kalman filters [5] and analog Hidden Markov Models (HMMs) [6]. In optical remote sensing, machine learning is used to develop accurate algorithms for specific regional surface waters observed with hyperspectral [7] and multi-spectral [8] sensors. In the ocean color community, fusion methods and machine learning are emerging topics.…”