Search

The Gulf of Aden, situated in the northwest Arabian Sea and linked to the Red Sea, is a relatively unexplored ecosystem. Understanding of large-scale biological dynamics is limited by the lack of adequate datasets. In this study, we analyse 15 years of remotely-sensed chlorophyll-a data (Chl- a , an index of phytoplankton biomass) acquired from the Ocean Colour Climate Change Initiative (OC-CCI) of the European Space Agency (ESA). The improved spatial coverage of OC-CCI data in the Gulf of Aden allows, for the first time, an investigation into the full seasonal succession of phytoplankton biomass. Analysis of indices of phytoplankton phenology (bloom timing) reveals distinct phytoplankton growth periods in different parts of the gulf: a large peak during August (mid-summer) in the western part of the gulf, and a smaller peak during November (mid-autumn) in the lower central gulf and along the southern coastline. The summer bloom develops rapidly at the beginning of July, and its peak is approximately three times higher than that of the autumnal bloom. Remotely-sensed sea-surface temperature (SST), wind-stress curl, vertical nutrient profiles and geostrophic currents inferred from the sea-level anomaly, were analysed to examine the underlying physical mechanisms that control phytoplankton growth. During summer, the prevailing southwesterlies cause upwelling along the northern coastline of the gulf (Yemen), leading to an increase in nutrient availability and enhancing phytoplankton growth along the coastline and in the western part of the gulf. In contrast, in the central region of the gulf, lowest concentrations of Chl- a are observed during summer, due to strong downwelling caused by a mesoscale anticyclonic eddy. During autumn, the prevailing northeasterlies enable upwelling along the southern coastline (Somalia) causing local nutrient enrichment in the euphotic zone, leading to higher levels of phytoplankton biomass along the coastline and in the lower central gulf. The monsoon wind reversal is shown to play a key role in controlling phytoplankton growth in different regions of the Gulf of Aden. [ABSTRACT FROM AUTHOR]

Abstract: Coastal areas of the North Sea are commercially important for fishing and tourism, and are subject to the increasingly adverse effects of harmful algal blooms, eutrophication and climate change. Monitoring phytoplankton in these areas using Ocean Colour Remote Sensing is hampered by the high spatial and temporal variations in absorption and scattering properties. In this paper we demonstrate a clustering method based on specific-absorption properties that gives accurate water quality products from the Medium Resolution Imaging Spectrometer (MERIS). A total of 468 measurements of Chlorophyll a (Chla), Total Suspended Material (TSM), specific- (sIOP) and inherent optical properties (IOP) were measured in the North Sea between April 1999 and September 2004. Chla varied from 0.2 to 35mgm−3, TSM from 0.2 to 75gm−3 and absorption properties of coloured dissolved organic material at 442nm (aCDOM(442)) was 0.02 to 0.26m−1. The variation in absorption properties of phytoplankton (aph) and non-algal particles (aNAP) were an order of magnitude greater than that for aph normalized to Chla (aph*) and aNAP normalized to TSM (aNAP*). Hierarchical cluster analysis on aph*, aNAP* and aCDOM reduced this large data set to three groups of high aNAP*–aCDOM, low aph* situated close to the coast, medium values further offshore and low aNAP*–aCDOM, high aph* in open ocean and Dutch coastal waters. The median sIOP of each cluster were used to parameterize a semi-analytical algorithm to retrieve concentrations of Chla, TSM and aCDOM(442) from MERIS data. A further 60 measurements of normalized water leaving radiance (nLw), Chla, TSM, aCDOM(442) and aNAP(442) collected between 2003 and 2006 were used to assess the accuracy of the satellite products. The regionalized MERIS algorithm showed improved performance in Chla and aCDOM(442) estimates with relative percentage differences of 29 and 8% compared to 34 and 134% for standard MERIS Chla and adg(442) products, and similar retrieval for TSM at concentrations >1g−3. [Copyright &y& Elsevier]

In recent years, ultraviolet (UV) bands have received increasing attention from the ocean colour remote sensing community, as they may contribute to improving atmospheric correction and inherent optical properties (IOPs) retrieval. However, most ocean colour satellite sensors do not have UV bands, and the accurate retrieval of UV remote sensing reflectance (Rrs) from UV satellite data is still a challenge. In order to address this problem, this study proposes a hybrid approach for estimating UV Rrs from the visible bands. The approach was implemented with two popular ocean colour satellite sensors, i.e. GCOM-C SGLI and Sentinel-3 OLCI. In situ Rrs collected globally and simulated Rrs spectra were used to develop UV Rrs retrieval models, and UV Rrs values at 360, 380 and 400 nm were estimated from visible Rrs spectra. The performances of the established models were evaluated using in situ Rrs and satellite data, and applied to a semi-analytical algorithm for IOPs retrieval. The results showed that: (i) UV Rrs retrieval models had low uncertainties with mean absolute percentage differences (MAPD) less than 5%; (ii) the model assessment with in situ Rrs showed high accuracy (r = 0.92–1.00 and MAPD = 1.11%–10.95%) in both clear open ocean and optically complex waters; (iii) the model assessment with satellite data indicated that model-estimated UV Rrs were more consistent with in situ values than satellite-derived UV Rrs; and (iv) model-estimated UV Rrs may improve the decomposition accuracy of absorption coefficients in semi-analytical IOPs algorithm. Thus, the proposed method has great potentials for reconstructing UV Rrs data and improving IOPs retrieval for historical satellite sensors, and might also be useful for UV-based atmospheric correction algorithms. [Display omitted] • A scheme is developed to estimate UV Rrs from visible ocean colour bands. • The scheme works well for both clear oceanic and optically complex waters. • Model-estimated UV Rrs are even better than satellite-sensed values. • The method holds potentials for improving and extending satellite data. [ABSTRACT FROM AUTHOR]

Particulate organic carbon (POC) plays vital roles in marine carbon cycle, serving as a part of "biological pump" moving carbon to the deep ocean. The blue-to-green band ratio algorithm is applied operationally to derive POC concentrations in global oceans; it, however, tends to underestimate high values in optically complex waters. With an attempt to develop accurate and robust oceanic POC models, this study aimed to explore machine learning methods in satellite retrieval of POC concentrations. Three machine learning methods, i.e. extreme gradient boosting (XGBoost), support vector machine (SVM) and artificial neural network (ANN), were tested, and the recursive feature elimination (RFE) method was employed to identify sensitive features. Matchups of global in situ POC measurements and Ocean Colour Climate Change Initiative (OC-CCI) products were used to train and evaluate POC models. Results showed that machine learning methods produced obvious better performance than the blue-to-green band ratio algorithm, and XGBoost was the most robust among the tested three machine learning methods. However, the blue-to-green band ratio algorithm still worked well for clear open ocean waters with low POC, and ANN was more effective for optically complex waters with extremely high POC. This study provided globally applicable methods for satellite retrieval of POC concentrations, which should be helpful for studying POC dynamics in global oceans as well as in productive marginal seas. • Globally oceanic POC models are developed using machine learning methods. • Machine learning methods work well for oligotrophic to productive waters. • Machine learning methods perform better than blue-to-green band ratio algorithm. • Machine learning methods are more robust for retrieving of POC concentrations. [ABSTRACT FROM AUTHOR]