The primary focus of the Ocean and Land Colour Instruments (OLCI), on-board the Sentinel 3-A and Sentinel 3-B platforms, is the remote sensing of Ocean constituents. Essentially for this purpose ist the atmospheric correction, since the largest part of the signal measured by OLCI originates from the atmosphere and not from the Ocean. A valid atmospheric correction requires an accurate characterization of the atmospheric aerosol.
We will present results of the validation of OLCI’s Aerosol product (aerosol optical thickness and Angstroem coefficient at 865nm). We continuously compare it with data from AERONET (Holben et al 1998), AERONET-OC (Zibordi et al 2009) and maritime AERONET (MAN: Smirnow et al 2009). This is an ongoing process, where co-located data are collected and analysed. Only quality assured L2 AERONET data is used. We concentrate on the reprocessed non-time-critical full-resolution data (FR NT). Up to now, several 10000 co-locations of OLCI-A with AERONET stations or available maritime AERONET cruises have been identified for the time period from 2016-04. For a matchup, the temporal distance between the satellite overpass and the AERONET acquisition had to be less than 30 minutes. Only OLCI measurements are used for the validation which are cloud-free (according to the standard cloud flags: cloud, cloud margin and cloud ambiguous) in an area of about 10x10 km2 around the AERONET acquisition. Further, all recommended flags from Sentinel-3 OLCI Marine User Handbook (EUM/OPS-SEN3/MAN/17/907205) have been applied. Eventually, to reduce the influence of undetected (sub pixel or sub visual) clouds, only matchups have been used, where the standard deviation of the aerosol optical thickness within the 10x10 km2 area was low (< 0.2). This strict filtering reduces the number of valid co-locations to few hundreds.
In addition to OLCI-A results we plan to present first results for OLCI-B.

Area D - Board 417

Geostationary Ocean Color Imager-II (GOCI-II), which will be launched in early 2020, has high temporal and spatial resolution to monitor oceanographic phenomena occurred by various biogeochemical and/or physical interactions. The data amount of GOCI-II has increased hugely but its timeliness requirement is very short like near real-time. We have been developing new ground segment (G2GS) and user support plug-in called GOCI-II Toolbox (GTBX). G2GS is composed of Algorithm Development Environment (ADE), System Operation Environment (SOE), Data Service Environment (DSE) for the researcher/engineer to develop new algorithm for GOCI/GOCI-II, for system operator in Korea Ocean Satellite Center (KOSC), for data user to ask the GOCI/GOCI-II data, respectively. To get high performance in GOCI/GOCI-II Level 2 data processing, we have to adopted parallel and distributed processing techniques using heterogeneous computing system (Multi-core CPU, GPU) on linux machines. GPU version of data processing, for example, showed the performance improvements 40 times better than a sequential version and nearly twice as high as a CPU version in parallel processing of GOCI atmospheric correction algorithm. The GTBX is developed for easy searching and easy accessing the GOCI/GOCI-II data based on SNAP (Sentinel Application Platform). It has three functions; Access the GOCI/GOCI-II data remotely, Translate geostationary observed satellite data internally using data model for data visualization and data analysis using SNAP functionalities, Process GOCI/GOCI-II Level 1B data to Level 2 data in SNAP. If you have a GPU in your data processing machine, you can enjoy the good performance of GOCI parallel data processing. The GTBX is able to support for expanding GOCI/GOCI-II applications like to data fusion/compound researches among geo/leo OC satellite data. We can easily get the comparison results for water quality parameters between GOCI and Sentinel-3 Ocean and Land Colour Instrument (OLCI) using GTBX/SNAP. For example, we can find similar patch for chlorophyll concentration, suspended sediment concentration, etc. at once in Korean South Sea both GOCI image and Sentinel-3 OLCI image thanks to GTBX on SNAP. We can make a comparison plot between two sensor’s data. This toolbox will be distributed using dedicated repository after In-Orbit Test of GOCI-II, 2020.

Area D - Board 418

Estimation of ocean surface salinity from space has widely been tried mainly by passive microwave sensors such as Soil Moisture and Ocean Salinity (SMOS), Aquarius. But, the results are not satisfactory in the coastal waters due to their rough spatial resolutions. Recent enhanced technologies of ocean colour satellite sensors in terms of spatial and temporal resolution can be employed effectively to the mapping of sea surface salinity (SSS), in particular in the coastal waters, although it is still challenging. As SSS can be a good indicator of short- or long-term changes in ocean environment and consequently the climate changes, many attempts have been made to mapping SSS in the wide area of ocean surface, although little has been successful particularly in the coastal area. Although it is well-known that coloured dissolved organic matter (CDOM) absorption property has a strong correlation with SSS in the area that influenced by freshwater inflow, it is not an easy task to retrieve accurately the CDOM absorption in highly turbid coastal waters due to the non-algal particle from land. In this study, remote sensing reflectance (Rrs) from the Geostationary Ocean Colour Imager (GOCI) centred at 490 nm (band 3), 555 nm (band 4), 660 nm (band 5) have been employed and validated to map SSS in the mid-western coastal area of Korean peninsula. Empirical relationships between Rrs at three bands and SSS are derived from in-situ measurement collected at the Gyeonggi-bay located in the middle of the west coast of Korean peninsula. We also applied the method to the South China Sea and validate the results in the period when the low salinity waters from Yangtze River affect the region. GOCI-derived SSS maps were compared with the ones from Sentinel-3. The results were also compared with those from HYCOM-modelled along with a model developed by KIOST.

Area D - Board 419

It is well known that Earth Observation (EO) satellites are able to monitor changes of ecological parameters in the environment; yet the radionuclide dispersion cannot be directly detected using satellite remote sensing data and techniques. However, the levels of radionuclides in the marine environment, especially for the soluble ones, are associated with other physical, chemical and biological parameters of the natural environment (e.g. temperature/evaporation, water density, biota exchange processes, salinity, water mass translocation etc.) (Buesseler et al., 1987; Buesseler et al., 1990; Buesseler et al., 1991; Florou, 1992; Florou and Kritidis, 1994; Florou, 1996; Schnoor, 1996; Ligero et al., 2001; Van Der Stricht and Kirchmann, 2001; Florou et al., 2003; Povinec et al., 2003; Evangeliou et al., 2009; Ashraf et al., 2014; Delfanti et al., 2014). The marine dynamic processes, like horizontal advection and vertical mixing, control the distribution of radionuclides in waters (Buesseler et al., 1991; Schnoor, 1996; Van Der Stricht and Kirchmann, 2001; Povinec et al., 2003). The water density controls the transport of water masses among basins, which results to trans-boundary radioactive contamination (Bueseller et al., 1991; Florou et al., 2003; Povinec et al., 2003). Moreover, physical parameters such as salinity, which is dependent on temperature, weathering processes and pollutant charge, controls water density. Sea temperature influences the biological retention time of the pollutants and consequently the elimination rate of the pollutants from the organisms (Florou, 1992; Florou, 1996). Therefore salinity and temperature are major parameters for the marine radiological assessment, as soluble forms of the artificial radionuclides like ¹³⁷Cs contribute in them.
Thus a concept is explored in terms of radiological remote control, based on the above attributes. Considering these features, marine parameters issued from Copernicus Marine Environment Monitoring Service (CMEMS) database are utilized, for the investigation of relations with ¹³⁷Cs activity concentrations. Such relations will lead to a tool based on remote sensing data and in situ ¹³⁷Cs measurements for the remote radioactivity detection of the Greek marine ecosystem both for routine and emergency recordings. Hence the main objective is the investigation of the relationship between ¹³⁷Cs activity concentrations, issued from the Environmental Radioactivity Laboratory, NCSR “Demokritos” sea water samplings and CMEMS database thematic maps. The parameters used in this study are Salinity (SSS), Sea Surface Temperature (SST), Mixed Layer Thickness (MLD), Nitrate Concentration (NO3), Phosphate Concentration (PO4), Phytoplankton Carbon Biomass (PCB), Chlorophyll (CHL), Dissolved Oxygen (DOX), pressure of CO₂ (PCO) and pH from CMEMS forecast, reanalysis and observation products. The ¹³⁷Cs activity concentration measurements have been derived from the database of Environmental Radioactivity Laboratory, whereas updated measurements have been carried out by analyzing seawater samples using a radioanalytical pre-concentration method (Ammonium Phosphomolybdate method and ¹³⁴Cs additive as carrier and yield tracer) and in-situ adsorption using filters impregnated with Copper Ferrocyanide (Cu₂[Fe(CN)₆]) (Folsom and Sreekmaran, 1970; Lento and Harjula, 1987; Wong et al., 1994; Nilchi et al., 2003; Evangeliou et al., 2009; Florou et al., 2010). The treated samples have been measured using gamma spectrometry.
The presented results show the relations for the area of Aegean Sea (Greece) and three sub-regions (Souda Bay, Corinth Bay and North Aegean Sea) for the years 2000-2018. The relations have been studied for different types of regressions (linear, second degree polynomial, exponential) both temporal and spatial for two distinctive seasonal periods (warm and cold period) with the use of statistical and GIS tools. The correlation was evident in several cases (r²> 0.5) with different parameters, especially for the cold period. The parameters that showed the best results are SST, MLD, NO3, PO4, CHL and pH. This is an ongoing study and additional information is being retrieved in order to improve the relation between marine environmental monitoring and radionuclide activity monitoring of the marine environment.

References:
Ashraf, M. A., Akib, S., Maah, M. J., Yusoff, I., & Balkhair, K. S. (2014). Cesium-137: radio-chemistry, fate, and transport, remediation, and future concerns. Critical reviews in environmental science and technology, 44(15): 1740-1793.
Buesseler, K.O., (1987), Chernobyl: Oceanographic studies in the Black Sea. Oceanus, 30(3): 23-30.
Buesseler, K. O., Livingston, H. D., Honjo, S., Hay, B. J., Konuk, T., & Kempe, S. (1990). Scavenging and particle deposition in the southwestern black sea—evidence from Chernobyl radiotracers. Deep Sea Research Part A. Oceanographic Research Papers, 37(3): 413-430.
Buesseler, K. O., Livingston, H. D., & Casso, S. A. (1991). Mixing between oxic and anoxic waters of the Black Sea as traced by Chernobyl cesium isotopes. Deep Sea Research Part A. Oceanographic Research Papers, 38: S725-S745.
Delfanti, R., Özsoy, E., Kaberi, H., Schirone, A., Salvi, S., Conte, F., Tsabaris, C., & Papucci, C. (2014). Evolution and fluxes of 137Cs in the Black Sea/Turkish straits system/North Aegean Sea. Journal of Marine Systems, 135: 117-123.
Evangeliou, N., Florou, H., Bokoros, P., & Scoullos, M. (2009). Temporal and Spatial distribution of 137Cs in Eastern Mediterranean Sea. Horizontal and vertical dispersion in two regions. Journal of Environmental Radioactivity, (100):626-636.
Florou, H. (1992). Dispersion and Behaviour of Long Lived Radionuclides in Marine Ecosystems. In Greek. Ph.D. thesis. Athens, National and Kapodistrian University of Athens.
Florou, H., & Kritidis, P. (1994). The dispersion of 137Cs in the Aegean Sea. Radiochimica Acta, (66/67): 415-417.
Florou, H. (1996). 137Cs Inventory in Abiotic Component and Biota From the Aegean and Ionian Sea-Greece. Chemistry and Ecology, 12(4), 253-258.
Florou, H., Kritidis, P., Vosniakos, F., Trindafyllis, J., Delfanti, R., Papucci, C., ... & Patrascu, V. (2003). Caesium-137 in the Eastern Mediterranean- impact sources and marine pathways. Fresenius Environmental Bulletin, 12(1):3-9.
Florou, H., Nicolaou, G., & Evangeliou, N. (2010). The concentration of 137Cs in the surface of the Greek marine environment. Journal of Environmental Radioactivity, 101: 654-657.
Folsom, T., & Sreekumaran, C. (1970). Some reference methods for determining radioactive and natural caesium for marine studies. Reference Methods for Marine Radioactive Studies. IAEA Technical Report Series, 118: 129-149.
Lento, J., & Harjula, R. (1987). Separation of cesium from nuclear waste solutions with hexacyanoferrate (II) s and ammonium phosphomolybdate. Solvent Extraction and Ion Exchange, 5(2):343-352.
Ligero, R. A., Ramos-Lerate, I., Barrera, M., & Casas-Ruiz, M. (2001). Relationships between sea-bed radionuclide activities and some sedimentological variables. Journal of environmental radioactivity, 57(1): 7-19.
Nilchi, A., Malek, B., Maragheh, M. G., & Khanchi, A. (2003). Exchange properties of cyanide complexes. Journal of Radioanalytical and Nuclear chemistry, 258(3): 457-462.
Povinec, P. P., du Bois, P. B., Kershaw, P. J., Nies, H., & Scotto, P. (2003). Temporal and spatial trends in the distribution of 137Cs in surface waters of Northern European Seas—a record of 40 years of investigations. Deep Sea Research Part II: Topical Studies in Oceanography, 50(17-21): 2785-2801.
Schnoor, J. L. (Eds.). (1996). Environmental modeling: fate and transport of pollutants in water, air, and soil. John Wiley and Sons. New York. pp. 682.
Van Der Stricht, E., & Kirchmann, R. (Eds.). (2001). Radioecology: Radioactivity & Ecosystems. Liege, Belgium, Imprimerie Fortemps Drukerij.
Wong, M. K., Jokela, T. A., & Noshkin, V. E. (1994). Radiochemical procedures for analysis of Pu, Am, Cs and Sr in seawater, soil, sediments and biota samples.Technical Report UCRL-ID-116497. Lawrence Livermore National Laboratory. Livermore, CA, USA

Area D - Board 420

Different types of phytoplankton dominate the ocean biomes and by that drive differently the marine food web, biogeochemical cycling in the ocean and exchange mechanisms with the atmosphere. During the Polarstern cruise PS113 (May-June 2018) we transected the Atlantic Ocean from Patagonian Shelf to the English Channel and measured within different Longhurst provinces (Longhurst 2010) phytoplankton composition and the optical properties continuously via inline spectrophotometry calibrated with measurements on discrete water samples. By processing the underway spectrophotometry measurements following Liu et al (2018) we derived continuous particulate absorption (ap) data set which was used to get a high resolved chl-a data set using the absorption line height method. The ap data were further processed to obtain highly resolved phytoplankton absorption and then major phytoplankton groups: For this we modified the Empirical Orthogonal Function method developed by Bracher et al. (2015) by using ap data for predicting phytoplankton groups. Here, we present results on biooptical properties (absorption of phytoplankton, other particulates and coloured dissolved organic matter (CDOM) and remote sensing reflectance spectra) and phytoplankton composition derived from the underway spectrophotometry, and from discrete station results of water sample analysis by HPLC and underwater radiometry. We also show the validation to matchups of OLCI remote sensing reflectance, chlorophyll, and absorption of the different water constituents. Figure 1 shows the comparisons of the average chlorophyll concentration (Chl-a) from OLCI-OC4 versus the HPLC Chl-a measured at discrete station. The agreement of the OLCI-FR (S3A) product fits very well, results based on daily 3x3 matchups show an R^2 of 0.89, a bias of -0.27 and an RMSE of 0.33. Comparable results are obtained using the underway spectrophotometry Chl-a data set. The OLCI data will be further used together with our in-line particulate absorption data to upscale the phytoplankton diversity information over the entire ocean.
References:
Bracher A., Taylor B.B., Taylor M., Dinter T., Röttgers R., Steinmetz F. (2015) Using empirical orthogonal functions derived from remote sensing reflectance for the prediction of concentrations of phytoplankton pigments. Ocean Science 11: 139-158
Liu. Y., Roettgers R., Ramírez-Pérez M., Dinter T., Steinmetz F., Noethig E.-M., Hellmann S., Wiegmann S., Bracher A. (2018) Underway spectrophotometry in the Fram Strait (European Arctic Ocean): a highly resolved chlorophyll a data source for complementing satellite ocean color. Optics Express 26(14): A678-A698; https://doi.org/10.1364/OE.26.00A678
Longhurst, A. R. 2010. Ecological geography of the sea. Academic Press.

Area D - Board 421

The impact of jellyfish blooms occurrences on human activities has increased over the past years causing problems on marine ecosystems and human activities. Most of the jellyfish species found in the Mediterranean Sea are mainly distributed near the coastal zone. A large number of reports on problems caused by jellyfish presence near the shore confirms their negative impact on coastal zone activities, such as tourism, fisheries, near-shore platforms, and aquaculture. In particular, aquaculture is one of the sectors affected the most, as jellyfish blooms are causing important damages to fish farms and facilities, and are responsible for a great loss in fish stocks [1]. Therefore, understanding the biological processes, ecological interactions and biomass dynamics associated with the occurrence of jellyfish is critical. Despite the importance of the phenomenon, few studies addressed the monitoring of distribution and abundance of jellyfish blooms [1]-[4]. These studies have limitations either because they are applied in a small area, thus giving only local information which is difficult to transfer in a different region, or because they are focusing on the direct detection of jellyfish blooms using Unmanned Aerial Vehicles (UAVs) and not on the forecasting of the blooms. Our work aims to integrate EO satellite data, along with model based products and in situ observations using machine learning techniques in order to innovate how jellyfish blooms can be detected and monitored on large scale across the Mediterranean Sea.
Most recent EO sensors, such as OLCI, MODIS, MERIS, offer a wide range of data products useful for estimating water quality parameters and understanding ocean dynamics at large scale temporally, however they are not suitable alone to be adopted as a standalone solution in real case scenarios. Copernicus Marine Environment Monitoring Service (CMEMS) provides regular and systematic biogeochemical and physical information on the marine environment. In more details, the Mediterranean Sea Biogeochemistry Analysis and Forecast is produced by means of MedBFM model system and provides daily mean estimations of biochemical parameters. The Mediterranean Sea Physics Analysis and Forecast is a hydrodynamic-wave model implemented over the Mediterranean Basin provides hourly, daily and monthly mean estimation of physical parameters. In addition to the models, we exploit Level-4, daily, gap-free satellite observations for chlorophyll-a concentration and sea surface temperature from multi-platform observations. In this context, we use CMEMS data to describe environmental conditions and studying their correlation to the ecological processes that leads to an outbreak of jellyfish. Knowing this relation, in-situ records of jellyfish occurrence are used to train a machine learning method.
For the purposes of this work, we have used a Support Vector Machine (SVM) classifier to estimate the jellyfish abundance level. The robustness of SVM has been demonstrated in handwriting recognition, text classification, object recognition and classification of remote sensing images [5]. An approach of jellyfish forecasting based on SVM was presented in [1]. This work shows an accurate prediction of jellyfish bloom occurrence, however the results focused only on a restricted region in Catalonia, Spain. Thanks to the scalability of pattern recognition methods and to the robustness of SVM, our work will extend the detection of jellyfish bloom to a larger area across the Mediterranean Sea. Our training dataset is made of in situ observations of jellyfish occurrence for two classes of species (both the benthic polyp stage and the adult medusa stage), used as target values, distributed throughout the Central-East Mediterranean coast and collected from January 2016 to September 2018. The 80% of the samples were used to train our system and the remaining 20% were used as test samples. For every set of coordinates an array was created including all the features. Using SVM we classified the abundance of jellyfish taking into account three separate classes “0: absence”, “1: presence”, “2: bloom”. In preliminary results, the proposed methodology leads to an average estimation accuracy of 75% for jellyfish distribution and abundance estimation.
The present work demonstrates the feasibility of using EO data and machine learning techniques for monitoring jellyfish distribution. Our results indicate the appropriateness of CMEMS data to model species distribution and dynamics when combined with in-situ observations. This work has been selected as operational service for the Mediterranean Sea (Lot 8) in the context of the project “User Uptake Activities Copernicus Marine Environment Monitoring Service (CMEMS)” by Mercator Océan. Future developments will focus on the extension of the proposed approach to more species, to develop a self-learning procedures, as well as geographical expansion.

[1] Albajes-Eizagirre, A., Romero, L., Soria-Frisch, A., & Vanhellemont, Q., "Jellyfish prediction of occurrence from remote sensing data and a non-linear pattern recognition approach," in Proceedings of SPIE, 2011. [2] Aznar, F., Pujol, M., & Rizo, R., "A swarm behaviour for jellyfish bloom detection," in Ocean Engineering, pp., 24–34, 2015. [3] Barrado, et al., "Jellyfish monitoring on coastlines using remote piloted aircraft," in IOP Conference Series: Earth and Environmental Science, 2014. [4] Canepa, A., et al., Forecasting jellyfish blooms along the Catalan Coast: a citizen science-based model and an on-line app to pay it forward," in Aquatic Science Meeting, 2016. [5] Melgani, F. and Bruzzone, L., "Classification of Hyperspectral Remote Sensing Images with Support Vector Machines," in IEEE Transactions on Geoscience and Remote Sensing, 42, 1778-1790, 2004.

Area D - Board 422

Difficulties to ensure compliance of fishing activities in high seas became evident since the first years after the implementation of the Law of the Seas. The lack of a responsible state in high seas adds to the difficulties of surveying vast oceanic regions and the high cost this implies for in situ supervision of ongoing activities. Alternatives to in situ vigilance includes the recent potential to harvest through satellites the Automatic Identification System (S-AIS) from vessels involved in fishing operations, even when they are offshore. The Global Fishing Program has made use of this potential and it is presently delivering data to research institutions and public bodies to support in this aim. S-AIS is a very valuable tool but it is not the only source of information that satellites can provide for fishing activities in high seas. Thus, orbiting radiometers such as VIIRS have proven useful in this task. VIIRS, a scanning radiometer, collects visible and infrared imagery and radiometric measurements of the land, atmosphere, cryosphere, and oceans. Its high sensitivity allows detecting anthropogenic light present in the ocean’s surface resulting from certain fishing operation during the night. Despite its limitations involving coarse spatial resolution as well as low bit quantization and saturation under bright lights, it is a valuable tool to able to provide information on human activities at high seas. Both S-AIS and VIIRS data have a worldwide coverage so that these anthropogenic operations can be assessed with both of them at high seas to look for synergy and complementarity. In this contribution, we perform such exercise for the high sea fisheries in the southwest Atlantic. We show here the coherent and complementary nature of both sources of information and, therefore, validate the role of VIIRS as a complement to AIS. This added information can be particularly useful if high seas activities reacts in response to the growing awareness that their activities are being recorded by AIS even when they are not close to the coast.

Area D - Board 423

Extracting information on phytoplankton community structure (e.g., phytoplankton functional types (PFTs), size classes (PSCs) and composition) has become a research topic of priority in ocean colour (OC) remote sensing. Indeed, PFTs and PSCs are playing an important role in understanding the marine food web and provide aids in the modelling associated with climate change impacts on biogeochemical and ecological cycling of oceans (IOCCG, 2014; Bracher et al., 2017). Satellite OC data enables observation of PFTs over large areas or even at global scale. With previous (e.g., SeaWiFs and MERIS) and current available OC satellites, especially the newly launched Sentinel 3A and 3B, a vast amount of quality-controlled OC data allow us to contribute in improving/developing methods and their applications to satellite data for identifying and estimating PFTs with respect to global observation. In the present study we focus on PFT retrieval algorithms that are then applied to Sentinel-3A (S3) OLCI data and merged OC products from CMEMS GlobColour archive. The main retrieved PFTs include diatoms, haptophytes, dinoflagellates, green algae and prokaryotic phytoplankton (cyanobacteria). Previously investigated retrieval methods, empirical orthogonal functions (EOF) for pigment concentrations estimation (Bracher et al. 2015) and generalized OC inversion model for inherent optical properties (GIOP) (Werdell et al. 2013, 2014) for PFT discrimination, are tested and adapted potentially with the use of extensive in situ measurements, matchups between in situ and satellite data, and satellite OC products. Performances of both EOF- and GIOP-based approaches are assessed statistically and cross-validated. The retrieved PFTs based on in situ data and matchups are compared with the results of diagnostic pigment analysis (DPA) from HPLC measurements, with results showing that both approaches could well predict chlorophyll-a concentrations for most of the mentioned PFTs but less accurate for prokaryotes due to their general low concentration resulting in weak signal in the reflectance spectra. The two algorithms are then adapted and applied to the GlobColour merged OC products and OLCI data. This is to eventually develop the global satellite PFT products for long-term observation, updated timely with more available OLCI data in the future, and intercompared to the results with other existing PFT products (e.g. PhytoDOAS, OC-PFT, SynSenPFT, PHYSAT).

Area D - Board 424

In this study, we evaluate the performance of empirical models and products from MODIS and Sentinel-3 for estimating important parameters in Southeastern Mediterranean sea waters. In particular, based on multitemporal Aqua-MODIS and Sentinel-3 satellite data, chlorophyll-a global prediction algorithms were applied and ocean-color products were derived and quantitatively evaluated against available in-situ data. Three algorithms i.e OCI, MedOC3 and OC3 were evaluated and new regression models were developed based on A-MODIS imagery and in-situ data. The OCI algorithm has been adapted to very oligotrophic waters, while bio-optical OC3M global algorithm was adapted to MODIS covering concentrations ranging from 0.012 to 72.12 mg m-3. MedOC3 is a regional bio-optical algorithm specific developed for Mediterranean Sea. For the whole study period 2013-2016, 689 sea surface temperature measurements and 365 chla data were used from SeaDataNet infrastrusture. Experimental results indicated that there is a good correlation between Aqua-MODIS products and in-situ data (R² ≈ 0.66 for chlorophyll-a and R² ≈ 0.89 for sea surface temperature). The OCI algorithm led to better correlation with in-situ data than OC3 and MedOC3 when linear (R² ≈ 0.62) and log-linear (R² ≈ 0.69) regression models were fitted. MedOC3 led to better results (R² ≈ 0.66) than OC3 (R² ≈ 0.61) when log-linear model was applied. New Prediction models led to better correlation when b10 (469nm, 488nm), b11 (531nm) an b13 (667nm) bands of MODIS where involved. The new log-linear model led to better correlations (R² ˂ 0.78, p < 0.0001) than the linear model (R² ˂ 0.65, p < 0.0001). A smooth comparison between OC4Me algorithm from Sentinel-3 OLCI sensor and OCI algorithm from Aqua-MODIS indicated that these two algorithms presented in general the same trend. Estimated chlorophyll-a from OCI algorithm was higher than Sentinel-3 OLCI products (OC4Me) for almost the whole dataset. Especially in the area of North Aegean Sea OC4Me presented much higher values than OCI algorithm, due to the presence of organic matter (CDOM) and specific phytoplankton species in the area. A further evaluation based on the multitemporal data and time-series in-situ observations should be conducted in order to assess the robustness and stability of the considered models.

Area D - Board 425

Chromophoric dissolved organic matter (CDOM) is compounded by a complex mixture of water-soluble organic substances which mainly include humic and fulvic acids. Because of its absorption spectral shape, CDOM performs a major control of the UV radiation penetration into the ocean, playing a crucial role for a variety of marine biogeochemical processes. CDOM can also influence the accuracy of global satellite-based measurements of ocean chlorophyll and primary productivity.
CDOM’s absorption and spectral signature can change over time and space, depending on its composition and origin. In coastal waters is mainly originated from rivers and land washing, but can also be locally produced, like in open ocean, by phytoplankton as well as through the degradation processes of this marine organic matter. While numerous algorithms have been documented for estimating CDOM from ocean color remote sensing (OCR) in coastal areas, studies dedicated to open ocean waters are still relatively scarce. The main issue for a proper estimation of CDOM absorption, acdom(λ), from OCR stands in our ability to distinguish it from the non-algal particles (NAP) absorption. This represents a real challenge considering the similarity of the CDOM and NAP absorption spectral shape. As a matter of fact, CDOM and NAP are usually described in oceanic waters by a single component acdm (λ) which refers to the absorption of colored detrital matter. CDOM is usually considered as the major contributor to acdm (λ) in case 1 waters, however the spatio-temporal variability of the relative contribution of acdom (λ) to acdm (λ) in response to the impact of the diverse source and sink controlling processes needs to be further investigated.
In this context, the present study is dedicated to the assessment of acdom (λ) over open ocean waters. Recent existing acdom (λ) inversion algorithms have been tested over synthetic and in situ data sets, while a new approach more suitable for global data has been proposed. This algorithm will be applied to the Globcolour ocean colour archive (including OLCI/Sentinel3 data) for match-up analysis as well as for characterizing the spatio-temporal variability of acdom (λ) and its relative contribution to acdm (λ) (i.e. acdom/acdm).

Area D - Board 426

Half of the Earth’s biological carbon fixation is carried out in the marine environment by oceanic micro-algae. In the climate change context, estimate the amount of carbon fixation is a particularly hot topic. Indeed, part of the organic matter produced through photosynthesis will sink out of the productive layer and be exported toward the sediment and sequestrated for centuries. Global ocean is therefore a carbon sink for the atmosphere.
In addition of the role of the marine surface productivity for climate, one could also notice that primary production is at the base of the marine food-web, providing energy to higher trophic level. As fish is the main source of proteins for many coastal populations, the intensity of the marine primary production is also important for economic reasons.
However due to anthropic perturbations, modulations of the biological carbon pump are expected. Study the primary production at large spatio-temporal timescales is important but not allowed by in situ measurements, due to their scarcity. As they present a relatively wide temporal archive and a high frequency, satellite estimates of primary production is a strong tool.
In the literature, several algorithms have been developed to estimate primary production. Here, we propose to present an inter-comparison of usually used primary production algorithms. Algorithms are compared with in situ measurement but also with biogeochemical models and using new metrics (spatial metrics, bloom occurrences, dispersion metrics…). The most accurate algorithm will be refined in order to propose a product with coastal regions well-resolved. In fine, the algorithm will be implemented on the Copernicus Marine Environment Monitoring Service which offers an unprecedented capability to observe, understand and anticipate marine environment events. The main purpose of this project is to supply information on marine primary production to different community users, as primary production involved impacts on climate and socio-economics resources.

Area D - Board 427

Primary production in the tropical open ocean is dominated by the smallest phytoplankton, with the contribution of picocyanobacteria and picoeukaryotes alternating as a function of nutrient availability. As nutrients increase in surface waters, towards the equatorial upwelling region and the edges of subtropical gyres, the dominance of picocyanobacteria is suppressed by the contribution of pico- and nano-eukaryotes, until the concentration of nutrients becomes high enough to support the growth of large (micro-) phytoplankton such as diatoms and dinoflagellates. As colored dissolved organic matter (CDOM) and inorganic suspended particles occur in low concentrations in oligotrophic subtropical waters, changes in the spectral shape of remote-sensing reflectance (Rrs) – the normalized spectra of light exiting the water column - are predominantly attributed to varying concentrations of living cells (phytoplankton and bacteria). Furthermore, considering the smooth exponential spectral shape of the absorption by CDOM and the low absorption by seawater in shorter wavelengths of the visible spectrum (400-550 nm), such variations in the shape of Rrs are mostly attributed to the varying absorption and scattering properties of phytoplankton cells in the surface waters of subtropical regions. As part of the NASA PACE mission (Plankton, Aerosol, Cloud, ocean Ecosystem), we investigate mechanisms to use hyperspectral Rrs information to retrieve major phytoplankton groups in the world’s oceans. Here we use changes in the Rrs spectral shape at the 400-550 nm region to discriminate areas where phytoplankton communities are dominated by picocyanobacteria or co-dominated by cyanobacteria and picoeukaryotes in subtropical oceanic waters. The second derivative of Rrs hyperspectral signatures was used to emphasize changes in the shape of standardized Rrs curves that are strongly correlated with the contribution of picocyanobacteria or picoeukaryotes in oligotrophic waters. This allowed the detection of a tilt in the Rrs curve at ~ 455 nm that possibly varies as a function of the availability of CDOM (including amino-acids used by picocyanobacteria) and the absorption of photosynthetic pigments by picoeukaryotes, when these are present. As nanoeukaryotes become more abundant, the absorption of light by their pigments increases in a way that causes a flattening of the Rrs in this spectral region, allowing the discrimination of areas dominated by larger phytoplankton cells (i.e. not picoeukaryotes). This method provides a useful, light-processing tool to discriminate two major phytoplankton groups that present severely different nutrient requirements, thus influencing element paths on biogeochemical cycles. This discrimination can potentially improve biogeochemical models in tropical and subtropical areas. The results provide a broad view of the distribution of these two major phytoplankton groups in oligotrophic waters, which can potentially lead to an improvement of our understanding about the influence of phytoplankton on biogeochemical cycles and the food web dynamics in the subtropical open ocean.

Area D - Board 428

One primary product of the Ocean and Land Colour Instrument (OLCI) on board the Sentinel-3 satellites is the remote-sensing or marine reflectance at the sea surface. It is the basis for determining optical properties of the upper mixed water layer and thereby source for the estimation of water constituents and biogeochemical properties. The marine reflectance is computed from the top-of-atmosphere signal by a pre-correction, such as gaseous and glint correction, and the subsequent atmospheric correction (AC). The Ocean Colour Bright Pixel Correction (OC-BPC) is part of the AC and corrects the signal in the near-infrared (NIR) spectral range for any residual marine reflectance (in particular due to sediment backscattering) so that the downstream clear water atmospheric correction can be applied under the “black pixel assumption”. OC-BPC is applied on all water pixels, not only over turbid waters, and must be seen as one of the numerous intermediate steps of the Level-2 processor. It means that OC-BPC contributes crucially to the quality of the OLCI Level-2 products.
The OC-BPC module has been completely revised within the framework of a EUMETSAT/Copernicus study. We document the changes in the OC-BPC module (modelling, choice of spectral bands, inverse method) and characterize the responses of OLCI Level-2 products due this update in various water types and atmospheric conditions.

Area D - Board 429

The in-water algorithm ONNS (OLCI Neural Network Swarm) is specially designed to retrieve various water quality parameters from data of the Ocean and Land Colour Instrument on board the Sentinel-3 satellites [Hieronymi et al., 2017]. The aim of the development is to provide one algorithm that is suitable to all natural waters, from clearest oceanic waters to very turbid coastal or highly absorbing inland waters. For this purpose, a fuzzy logic optical water type classification is applied in conjunction with a set of specific neural networks. Input to the algorithm are normalized remote sensing reflectances at 11 OLCI bands, i.e. atmospheric corrected satellite data. Through further development of the algorithm, it is also possible to process data from other ocean colour sensors such as SeaWiFS, MERIS, VIIRS, and others. The processing chain has changed over time and new features and ocean colour products are provided, e.g. IOP-based dissolved and particulate organic carbon, DOC and POC. The updated version has become accessible via free and open source software. Results of ONNS are water-type-wise compared to different atmospheric correction algorithms, to standard ocean colour products (e.g. Chl_OC4ME and Chl_NN) and set into context with in situ observations.

Area D - Board 430

The Marine Strategy Framework Directive (MSFD) was adopted in 2008 by the European Union (EU) with the aim to protect more effectively the marine environment across Europe. The MSFD aims to achieve Good Environmental Status (GES) of the EU's marine waters by 2020 and to protect the resource base upon which marine-related economic and social activities depend. In total eleven qualitative descriptors describe what the environment will look like when GES has been achieved, of which the fifth is the eutrophication status in marine waters.
Eutrophication is the result of excessive enrichment of water with nutrients which may cause an accelerated growth of phytoplankton biomass in the water column resulting in a range of undesirable disturbances in the marine ecosystem. This process can be caused due to natural variabilities or due to external perturbations (anthropogenic effects). Human-induced eutrophication is one of the criteria for assessing the extent to which GES is being achieved and is established by monitoring of chlorophyll-a (Chl-a) concentration as a proxy of phytoplankton biomass.
Rivers serve as the main channel for the delivery of significant amounts of nutrients from terrestrial environments to the ocean. Consequently, monitoring the spatio-temporal distribution of Chl-a in coastal waters is important to evaluate the impacts of human activities and the discharges generated from inland sources which may change the ecosystem around. While in situ data acquisition is still considered as the main monitoring tool, there is a growing tendency to use optical remote sensing as a supporting tool to achieve a coherent assessment of Chl-a in the Belgian Continental Shelf (BCS) and its coastal waters.
The launch of Sentinel-3 in 2016 equipped with the Ocean and Land Colour Instrument (OLCI) made possible a continuous, high frequency monitoring of marine environments. The OLCI sensor has a 300 meters spatial resolution combined with a high-temporal resolution providing valuable information that allow the monitoring of Chl-a near BCS coast and offshore zones. Additionally, the OLCI sensor has a spectral resolution covering the visible and near-infrared (NIR) wavelengths which allow the use of blue-green band ratios algorithms for low-medium Chl-a concentrations in clear waters and red-edge algorithms for high Chl-a concentrations in turbid waters, respectively.
The objective of this study is to apply the multi-algorithm Chl-a product developed by the Remote Sensing and Ecosystem Modelling (REMSEM) group supported by the Royal Belgian Institute of Natural Sciences (RBINS) for the OLCI dataset in the BCS. The Chl-a algorithms were determined using the Coast Colour Round Robin (CCRR) dataset, designed to test algorithms and assess their accuracy for retrieving water quality parameters. The CCRR quality controlled Chl-a algorithms were applied to the OLCI data and subsequently merged together into a single OLCI Chl-a product.
This study presents the spatio-temporal growing of phytoplankton biomass along the BCS and determines the validity of this approach to enable a MSFD eutrophication monitoring based on Sentinel-3 images. The results of this study are expected to be useful for the governmental management of Belgian sea waters and marine service operationality.

Area D - Board 431

Earth system and climate research is driven primarily by numerical models, which allow us to quantify and understand key global systems and processes. Earth observation (EO) is an increasingly important source of input data for assimilation into these models, with its synoptic scale coverage at fine spatial resolution with a high repeat rate. However, the parameters derived from EO are often the result of a complex multi-stage processing sequence, each stage of which incurs error, and the need for rigorously quality-controlled assimilation data with well-characterized uncertainty has led to the concept of Essential Climate Variables (ECVs), which now include satellite chlorophyll.
We are developing a formalism to model the uncertainties in satellite products such as chlorophyll as a combination of three components, each of which is a function of sources of error that can vary from pixel to pixel. The components are: a bias that can be corrected to improve the accuracy of the product; the standard error of the bias, and a residual standard deviation. The latter two serve as measures of the remaining uncertainty after bias correction, or all three can be used as measures of the uncertainty in uncorrected data. These values are calculated uniquely for each satellite measurement. The error sources can be anything quantifiable that contributes to or is correlated with the uncertainty, such as solar zenith angle, aerosol optical depth or spectral remote sensing reflectance. The overarching goal of this research is to provide a consistent approach to uncertainty across a wide range of disciplines.
Here we apply this formalism to the Ocean Colour Climate Change Initiative (OC-CCI) global database of satellite-in situ chlorophyll matchups to create a model of chlorophyll bias and residual uncertainty as functions of spectral remote sensing reflectance and optical water type. We apply this model to global OC-CCI data and compare results with the uncertainty analysis performed by OC-CCI, which assigns a bias and standard deviation to each optical water type and weights these at each pixel.

Area D - Board 432

Chlorophyll concentration is the universal biomass proxy for phytoplankton and, hence, it is a key variable in the global carbon cycle, recognised as an Essential Climate Variable (ECV) by the Global Climate Observation System (GCOS). Ocean colour remote sensing enables regular global observations of chlorophyll-a but calculation of trends requires a long and consistent time-series derived from multiple missions.
The ESA Ocean Colour Climate Change Initiative (OC-CCI) is producing a merged product from the various ocean colour sensors, which may be used in climate change research and assimilated into models. OC-CCI focuses on water-leaving radiance in the visible domain which is merged from ESA’s Envisat MERIS and Sentinel-3 OLCI, NASA’s SeaWiFS and MODIS-Aqua and NOAA’s Suomi VIIRS sensors, and then used to derive chlorophyll-a, inherent optical properties and diffuse attenuation coefficient.
The need for high-quality fiducial reference measurements and in situ data to develop and validate ocean-colour satellite products is well acknowledged by the scientific community and space agencies. This work presents latest developments in the OC-CCI in-situ database including validation and uncertainty characterization of the OC-CCI products. The database comprises in-situ measurements of remote-sensing reflectance, chlorophyll-a concentration, inherent optical properties, diffuse attenuation coefficient and total suspended matter, from 1997, at the global scale, gathered from several sources (see Valente et al, 2016, ESSD). Inclusion of in situ data requires conversion of “raw” data into a consistent format, quality control (including identification/removal of duplicates), reduction of data in time/space, metadata propagation and merging of all data into one unique table. The database is designed to be easily accessed and used by the OC community, and version 2 (Valente et al 2016) is available on https://doi.pangaea.de/10.1594/PANGAEA.854832
This presentation describes work to extend the database to October 2017, and highlights the differences between previous and current versions. In particular, the performance of Sentinel-3 OLCI is shown using data from the UK annual Atlantic Meridional Transect (AMT), highlighting its importance for the coverage of the Atlantic Ocean and gathering fiducial reference measurements for satellite validation.

Area D - Board 433

This talk will present satellite ocean colour techniques for discriminating harmful algal blooms (HABs) to provide early warning for the finfish and bivalve shellfish aquaculture industry; and to discuss their potential and limitations for improving human health. Global demand is growing for seafood, though one obstacle to the expansion of the aquaculture industry is the risk of algal toxins that can kill finfish or contaminate shellfish, causing human poisoning when consumed. HAB events are also costly in prolonged harvest closures, product recalls and loss of consumer confidence, threatening the sustainability of farms. These risks have instigated costly monitoring programmes to ensure the safety of seafood, for example by the Foods Standards Agency for the UK; though such measures may not be sufficiently implemented in developing nations.

We have undertaken research to develop classifiers for certain high-biomass HAB species (e.g. Pseudo-nitzschia and Karenia mikimotoi), for example in the UK ShellEye project we have adapted the techniques for ESA Sentinel-3 OLCI data at 300m resolution. Real-time trials were carried out to provide regular bulletins to shellfish farms, and to evaluate the benefits to farmers of early warning of such blooms. We will discuss limitations such spectral sensitivity of current ocean colour sensors, cloud cover, and the unfeasibility of satellite detection of many toxic blooms that are harmful even in low concentrations. Ongoing projects are seeking to address such limitations through combining satellite and in situ monitoring with short-term forecasts from hydrodynamic and ecosystem models.

Area D - Board 434

The System Vicarious Calibration (SVC) is the process used to adjust the spectral calibration coefficients of ocean color (OC) space borne sensors in order to meet the desired accuracy in retrieving the marine leaving radiance. Simplifying, OC-SVC consists in comparing contemporary in situ and satellite retrievals of fully normalized water leaving radiance in the best possible observation conditions to minimize data uncertainties, and then adjust satellite measurements onto field measurement.

Current OC-SVC processing schemes requires knowledge of Total Chlorophyll-a (TChla) at several stages of the calculation of the adjustment coefficients (so called g-factors). TChla is a widely used indicator of the bio-optical state of the ocean and in OC-SVC is needed, for example, as input for the BRDF correction look up tables. High Performance Liquid Chromatography (HPLC) analysis of discrete samples is the most accurate technique used for the determination of TChla. To extend our capability to measure TChla at temporal and spatial scales not compatible with HPLC analyses, numerous methods have been developed to estimate TChla with in situ or remotely deployed instrumentation based on optical proxies (e.g.: fluorescence, reflectance, etc.).

Here we evaluate different methods and algorithms for the estimate of TChla using a long-term time series of optical properties and HPLC data collected at the BOUSSOLE station (NW Mediterranean Sea) and their impact on the determination of Sentiel3A-OLCI g-factors.

Area D - Board 435

Fuzzy-logic can be used for optical classification of oceanic and coastal waters. The resultant classes can then be used for blending ocean-colour algorithms across optical water types (OWT) (Moore et al. 2009, 2014). An OWT set is computed using clustering techniques applied to a training dataset and the use of objective functions to assess set performance. The application of optical water type membership to algorithm blending by requires the creation of a globally representative set of optical types that captures the observed diversity in spectra across all ocean waters. This method was updated and utilised within the OC-CCI project (Jackson et al. 2017) to maximise product quality across differing optical water types, but was limited to the six SeaWiFS wavebands.

The application of these techniques to the state-of-the-art OLCI sensors on board Sentinel 3A and 3B requires a new OWT set that will make use of the full OLCI waveband set. Here we present an initial OWT set for OLCI that covers the global oceans. We then investigate the geographic distribution and variability of OWT memberships and of the dominant OWT to gain an insight into the optical diversity of the global oceans and to demonstrate the application of algorithm blending on OLCI data.

The new OWT set is compared with that used within earlier versions of the OC-CCI products and the impact of the additional information provided by the OLCI wavebands is discussed. We also discuss the possibility of increasing the OWT set to include inland waters and the additional requirements this might involve.
Lastly, by comparing the global distribution of the OWT memberships to a comprehensive in-situ database (Valente et al. 2016) allows us to highlight regions where targeted in-situ data collection would allow a better understanding of the characteristics such as algorithm performance or environmental significance of the OWTs.

Area D - Board 436

Recently, Jackson et al. (2017) showed that the effect of light on the chlorophyll-to-carbon ratio of phytoplankton is a function of the dimensionless irradiance (that is, the irradiance normalised to the photoadaptation parameter I_k) and the maximum chlorophyll-to-carbon ratio attained when available light tends to zero. Using a province-based database of the photo-adaptation parameter, in combination with the OC-CCI chlorophyll products and NASA data on photosynthetic available light at the sea surface, we demonstrate a method to calculate phytoplankton carbon in the oceanic mixed layer, at the global scale. The importance of the maximum chlorophyll-to-carbon ratio in these calculations is examined, and field-based estimates of it are compared with reported values from laboratory cultures.

Area D - Board 437

Suspended sediment concentration (SSC) is one of the most important parameters of ocean water, especially the case II water. It an important factor to affect the coastal zone development decision-making, the coast of the exploitation and utilization of resources, geomorphological evolution, environmental change, economic develop-ment and its important role in the study of more and more countries along the atten-tion of suspended sediments. Remote sensing, a common method to study the spec-tral characteristics of water and the variation of suspended sediments, is an im-portant method for large-scale investigation of suspended sediments in water. Re-mote sensing can effectively overcome the shortcomings of field measurement and has the advantages of wide range and high efficiency. After the serious study by Ritchie , Aranuvachapun , Dekker , Doxaran , D'Sa , Wass , Han , Hellweger , it becomes the main technical means for the study of suspended sedi-ment in marine waters.
The absorption and scattering properties of a medium such as sea water are de-scribed by its inherent optical properties (IOPs). IOPs are properties of the medium and do not depend on the ambient light field. That is, a volume of water has well defined absorption and scattering properties, whether there is any light to be absorbed or scattered. It is important for characteristics of suspended particles , primary productivity , chlorophyll concentration , euphotic zone depth , and dis-solved and particulate organic carbon concentrations . The Quasi-Analytical Algorithm (QAA) was developed by Lee et al. to derive inherent optical properties of optically deep waters in 2002. The International Ocean-Colour Coordinating Group present a report of algorithms for IOPs in 2006 . In this report, results sug-gest that the retrieval performance of a quasi-analytical algorithm is very high for both clear and coastal waters. Until today, the QAA has undergone 5 updates, the latest formally published version is QAAv6 . These updates have made the QAA more consummate, and the performance has improved greatly. After Lee, many scholars have improved the algorithm. Chen, based on MODIS data, proposed an improved algorithm, QAA-RGR.
The samples collection was performed along the Shuantaizi River and Daliaohe River, the two rivers belong to the tributaries of Liao River, on June 7 to 9. 104 water samples were collected and taken back to the laboratory, and the coordinate posi-tioning information including latitude and longitude was acquired using a GPS receiver.
Standards for determination of suspended sediment concentration based on the Ministry of Ecology and Environment of the People’s Republic of China, Water qual-ity - Determination of suspended substance - Gravimetric method (GB/T11901-1989). The standard points out that the suspended substance in water refers to the material that the sample of water passes through the filter membrane with a pore diameter of 0.45 micron and is intercepted on the filter membrane and dried to con-stant weight at 103-105 degrees Celsius.
Tiangong-2, a Chinese space laboratory and part of the Project 921-2 space station program, was launched on 15 September 2016. Tiangong-2 contains 3 sensors, Wide-band Imaging Spectrometer, Three-dimensional Imaging Microwave Altimeter, Mul-ti-band Ultraviolet Edge Imaging Spectrometer.
The TG-2 MWI Level-2 data were download from the MSADC website via ac-count registration and applying an application for the data products free of charge (http://www.msadc.cn/en), which has been geometrically corrected and radiation corrected. Before the atmospheric correction, the meta data has been modified to add the band information. The MWI image from June 1,2018 over Liaohe estuary from June 1,2018 was used to build the model. The Level-2 data contained calibrat-ed data for the apparent reflectance, and the surface reflectance of images was gen-erated by Quick Atmospheric Correction (QUAC), an atmospheric correction module, implemented in ENVI.
To remove the influence of water refraction on reflectivity, the Above-Surface Remote-Sensing Reflectance Rrs (λ) measured by the satellite are converted to their subsurface values, Below-Surface Remote-Sensing Reflectance rrs (λ).
rrs (λ)=(Rrs (λ))/((0.52+1.7Rrs))
The relationship between Ratio Of Backscattering Coefficient To The Sum Of Absorption And Backscattering Coefficients u(λ) and Absorption Coefficient Of The Total a(λ), Backscattering Coefficient Of The Total bb (λ) is shown in formula, where g₀=0.089 and g₁=0.1245.
u(λ)=(bb (λ))/(a(λ)+bb (λ) )=(-g₀+(g₀^2+4g₁*rrs (λ))^0.5)/(2g₁)
According to the characteristics of suspended sediment reflection, the absorption coefficient at 700nm decreases rapidly. We designated 682.5nm as λ₀, which the central wavelength of VNI Band 7. Detailed Quasi-Analytical Algorithm calculation steps are shown in next.
Step 0: rrs (λ)=(Rrs (λ))/((0.52+1.7Rrs))
Step 1: u(λ)=(-g₀+(g₀^2+4g₁*rrs (λ))^0.5)/(2g₁)
Step 2: a(λ₀)=a(670)=aw (670)+0.39((Rrs (670))/(Rrs (443)+Rrs (490) ))^1.14
Step 3: bbp (λ₀ )=bbp (670)=(u(λ₀)*a(λ₀))/(1-u(λ₀))-bbw (670)
Step 4: η=2.0×(1-1.2 exp(-0.9 (rrs (443))/(rrs (55x))))
Step 5: bbp (λ)=bbp (λ₀) *(λ₀/λ)^η
Step 6: a(λ)=[1-u(λ)][bbw (λ)+bbp (λ)]/u(λ)
After step 6 of Quasi-Analytical Algorithm model, we got the absorption coefficient of the total samples. There is a polymerization point, which we choose the 682.5nm as λ₀.
We use Root-Mean-Square Error (RMSE), Average Relative Error (ARE), and R-square (R²) to evaluate the inversion effect of the total absorption coefficient and analyze the accuracy of the QAA model.
Combined with the reflectivity characteristics of suspended sediment, we selected Band 9 and Band 10 built a SSC model.
SSC=75.31*(Band10-Band9)/(Band10+Band9)+41.43
The accuracy of the SSC model are: 1) R-square is 0.6859; 2) RMSE is 2.8148; 3) ARE is 9.68%.
In this study, using the TG-2 MWI data of June 1, 2018, based on the Quasi-Analytical Algorithm, selected the reference wavelength 682.5nm as λ₀, established the inverse model of suspended sediment concentration. The model was applied to TG-2 MWI data, the SSC result is shown in Figure.
The Quasi-Analytical Algorithm can establish the relationship between optical and biochemical characteristics step by step according to the radiance of water. Its calculation steps are easier to transfer and accumulate the error. The second error mainly comes from the process of remote sensing reflectance above the water surface, which needs to consider the wind speed, shadow and so on. Below-surface remote-sensing reflectance rrs (λ) cannot be directly obtained by measurement. There is no function can be accurately modeled for different measurement angles. Especially for shallow water, lakes and turbid water with complex optical characteristics, it is often calculated by the empirical parameters of above-surface remote-sensing reflectance Rrs (λ). Even if the lake bottom reflectance contribution is excluded and assumed to be zero error, there are still many uncertainties in the calculation of Quasi-Analytical Algorithm. For example, in Step 1 of the Quasi-Analytical Algorithm, the formula is derived from the radiative transfer equation, and g₀ and g₁ are not constants. They are affected by the zenith angle of the sun and the optical characteristics of the water, and vary with the scattering phase function and the scattering characteristics. The absorption coefficient at the reference wavelength is also one of the main sources of error in the Quasi-Analytical Algorithm.
In addition, most of the sampling sites are close to the shore where the SSC value is higher than other parts of study area and it causes the results of the inversion relatively concentrated on high value.

Area D - Board 438

The Copernicus program brings a wealth of satellite and environmental data at high spatial, spectral, radiometric and temporal resolution with free and open data access policy and user support service. This is a real opportunity to marine environment monitoring and high-scale mapping of water quality parameters in regions with lack of regular sampling.
The MultiSpectral Imager (MSI) on the Sentinel-2 series of satellites provide images with unprecedented characteristics in open data category; wide swath (290 km), high spatial resolution (20 m) and short revisit time (2-3 days). At this point, we utilize both Sentinel-2 images and surveyed sampling sites in summer of 2018 (July, September) to assess and map water quality parameters (Turbidity and Chlorophyll-a concentration) over Bousmaïl Bay, before called "Castiglione", and considered with its area of 509 km2 among the most important in Algeria. Despite its agricultural and tourist nature, studies revealed the presence of strong anthropic pressure in some areas where industrial activities have recently emerged. The proximity of this study area from ENSSMAL is an advantage so that it is studied regularly, which facilitate the verification and interpretation of results.
We use Sentinels Application Platform (SNAP) for processing data; experimenting Dark object subtraction method, computing turbidity based on red band and Chlorophyll-a concentration based on local calibration of MODIS data. The two maps (from satellite data and those of sea campaigns) seem to show a similar distribution for the 2 parameters. A clear and precise distribution of maximum chlorophyll-a values is present in the central part of BouIsmail Bay. The interpolation also shows us that the distribution of Chlorophyll-a tends to decrease while moving towards the sea which is a characteristic of this region of the Mediterranean Sea. The spatial distribution of turbidity shows strong values at the mouth of Oued (small river) Mazafran. Quantitative comparison will be presented.

Area D - Board 439

Suspended particulate matter (SPM) is one of the main water constituents in estuaries and along with turbidity (T), which is highly correlated with SPM concentration, are key parameters to evaluate water quality.
Ocean color remote sensing has been shown to be a useful tool to map turbidity and SPM concentration over coastal waters, and different algorithms to retrieve these parameters from water reflectance have already been developed. However, SPM algorithms usually require regional calibrations to take into account local differences in particle type (size and composition).
In the present work, in situ data (including T, SPM and radiometry measurements) were collected at a pontoon in Tagus estuary (Portugal), between the 27th August to the 7th September 2018, to test different algorithms (e.g. Nechad et al. 2016, Dogliotti et al. 2015) and evaluate coincident Sentinel 2 (S-2) and Sentinel-3 (S-3) imagery estimates using two different processors [i.e. ACOLITE (only S-2) and C2RCC (S-2 and S-3)].
Turbidity and SPM values varied between 1.49 – 6.88 NTU and 1.3 - 8.3 mg/L, respectively, variability mainly associated with the tidal conditions. The low number of cloud-free contemporary images (S-2, n=2 and S-3, n=1) does not allow any robust statistics, but provides a first preliminary validation of S-2 and S-3 data for the Tagus area.
For S-2, ACOLITE processor retrieved higher reflectance values at 665, ρw(665), wavelength used for SPM and T retrieval, while C2RCC processor underestimated it in comparison to in situ radiometric data. For the S-3 match-up, only C2RCC processor was compared and also underestimated in situ ρw(665).
Regarding SPM retrievals, using ACOLITE (v20180925.0) and Nechad et al. (2016) algorithms overestimated in situ values, particularly at low SPM concentrations. C2RCC also provided higher SPM concentration values, for both S-3 and S-2 match-ups, except for the conditions of lower in situ SPM values, where a slight underestimation was observed.
Considering S-2 derived turbidity, ACOLITE processor using both Nechad et al. 2016 and Dogliotti et al. 2015 algorithms were tested, showing the latter better results. Applying the same algorithms to C2RCC-derived reflectances, results were underestimated for both sensors data.
This work is a first Sentinel validation effort for the area, further and regular sampling should be pursued in the future in order to increase the database, essential for algorithm development and validation activities of satellite products for the region.
This activity was conducted in the framework of PLATAGUS, a bilateral cooperation between Portugal and Argentina (FCT-MinCyT).

Area D - Board 440

Ebro Delta bays are two estuarine environments where phytoplankton (i.e. chlorophyll-a, chl-a), has a major importance because of shellfish aquaculture, which has a large socioeconomic impact in the region. The implementation of phytoplankton monitoring programs has become mandatory at European production sites and the use of space-borne remote sensing for this purpose has been increasing along the years. Quantitative remote sensing of chl allows the transformation of satellite images into chl-a distribution maps. Regular monitoring through space-borne sensors requires to estimate water-leaving radiance by removing atmospheric contributions from the signal received at the top of atmosphere (TOA) and to obtain a robust relationship between water quality and satellite-based parameters.

With the purpose of finding a suitable methodology for chl-a monitoring using space-borne remote sensing, this study aims to explore the performance of the Sentinel-2 multispectral instrument (S2/MSI) spectral bands and the derived spectral indices for chl-a retrieval at Ebro Delta bays. The proposed approach uses both TOA and bottom of atmosphere (BOA) images together with several pre-processing technics including different atmospheric processors, pixels outlier detection, principal component analysis (PCA) and spectral separability study. Thus, the objectives are;

1) Assessing the spectral differences between atmospherically uncorrected and corrected images comparing different atmosphere correction processors; 2) Finding the maximum or optimum size of the pixel window that could be used for model calibration (spatially average the reflectance over a number of pixels centered at an in situ measurement site for reducing environmental, sensor and positioning noises) and 3) Determinate the sensivity of the different bands and spectral indices to quantify chl-a at different concentrations (for band selection and algorithm building).

Similar approaches have been proposed for inland waters, but the use of Sentinel-2 is still scarce when assessing coastal areas. In addition, the use of space-borne remote sensing is commonly used to derive site-and-time specific empirical algorithms (regional tuning). Here, the proposed methodology has been tested with 42 Sentinel-2 images covering all the year 2017, and thus including several environmental scenarios. The S2 imagery were atmospherically corrected with both ACOLITE and POLYMER processors. On this time-series, several spectral indices were computed and S2/MSI full spectrum was considered. Pixel intra-stability was evaluated by the use of the coefficient of variation (Cv) and pixel inter-variability was analysed by means of spectral separability. For all indices two different models were developed by using two in-situ measures of chl-a (Nfluorimeter = 56 and Nspectrophotomete = 43), all S2/MSI bands were considered in a PCA for new algorithm development.

The results of our study show that despite not being designed for aquatic purposes, Sentinel-2 satellites provide a new opportunity for moderate spatial resolution of chlorophyll-a (chl-a) retrieval, mapping and monitoring at coastal areas. The preliminary results, in agreement with the ones obtained in other coastal areas, suggest the need for atmospheric correction processing. The CV served as an outlier detector and the PCA and spectral separability analyses provided for the best configuration of the methodology (bands used, spectral indices and average pixel window size).

Area D - Board 441

The North Adriatic Sea (NAS) is a semi-enclosed basin, located in the North-Eastern part of the Mediterranean Sea. In this basin river inputs and meteo-marine conditions strongly affect optical properties, dynamics, and ecology of coastal and shelf waters.
To adequately capture these processes, accurate satellite estimates of particulate backscattering (bbp) and dissolved and particulate absorption (adg) at high spatial and temporal resolution are needed.
The daily observation provided by most of the polar Ocean Colour (OC) satellites, may result insufficient to describe the short time scale dynamic of the bio-optical properties. The only geo-stationary satellite available over the NAD (SEVIRI-MSG) provides multiple observations of the area during the same day but, due to its coarse spatial resolution over this basin (6kmx6km), it cannot detect the small spatial variability, typical of the coastal areas. Moreover, retrieval of OC products apart from turbidity/Suspended Particulate Matter (SPM) is not feasible, with this sensor being designed mainly for meteorological and weather applications.
The aim of this work is to emulate a virtual geostationary OC satellite (VGOCS), based on the constellation of OC polar satellite, in order to obtain multiple observations of the investigated area during the same day. The VGOCS will approach the temporal resolution of a geostationary satellite, while maintaining the spatial, spectral, and radiometric resolutions of the polar orbiters. This will enable to capture both short time and small spatial variability, typical of the NAS coastal and shelf areas.
The sensors that will be included in this constellation are MODIS-AQUA, VIIRS-NPP, OLCI-Sentinel 3A, OLCI-Sentinel 3B, and VIIRS-NOOA20. In this work L2 Remote sensing reflectance (Rrs) data, provided by the different space agencies (e.g. ESA, NASA, EUMETSAT), will be used. Data from different sensors can have residual differences between each other due to different processing and calibration procedures. Moreover, during the same day the sensors can have different observation geometries, possibly amplifying the inter-sensor biases.
In this work, the inter-sensor biases between the constellation sensors over NAS will be reduced by using in situ Rrs AERONET-OC data, acquired at the Acqua Alta Oceanographic Tower (AAOT) in the gulf of Venice. The Rrs adjustment will be performed using a multi-linear regression scheme that minimizes the systematic differences between satellite and in situ Rrs data. Being the AAOT representative of the optical properties of the NAS, the adjustment applies to all Rrs spectra observed in the basin. This adjustment will also reduce the uncertainties associated to observations with high sensor zenith angles, making available in the analysis those data normally masked in standard processing. The adjusted VGOCS Rrs data will then be used as input to consistently calculate the optical properties of interest (adg and bbp), using the Quasi Analytical Algorithm (QAA v6).
To showcase the feasibility of the proposed VGOCS in adequately capturing short term variability of optical properties in the NAS, case studies will include imagery of high river discharge regime and strong wind driven resuspension. The use of VGOCS multiple observations (4-6 acquisitions in ~3 hours), provided by the constellation, enables to observe the optical variability driven by those forcings.

Area D - Board 442

High spatial (10 m) Sentinel-2 MSI data are adequate to monitor water masses of small dimensions such as those of lakes and reservoirs. Reflectance data used as input for the in-water algorithms has to be as accurate as possible, especially when working in complex waters, like the hypertrophic Albufera Lagoon near Valencia (Spain). Since 2017 the Ecological Status of AQuatic Systems with Sentinels Satellites (ESAQS) project has been evaluating the performance of five different MSI atmospheric correction processor over different water types: Sen2Cor, Acolite, C2RCC, iCor and Polymer. Results obtained with the statistical analysis made from the correlation between the S2-MSI reflectance matched-up with in situ reflectance showed than Polymer and C2RCC provided the minimum errors in most of the cases. However, the Albufera Lagoon turner out to be a very special case, with many pixels rejected as valid pixels and consequently not corrected by the processors. Adjacency and bottom reflectance contamination add up uncertainties to the already complex water in the lagoon. In order to quantify properly the uncertainties and figure out the best solution for this case, we keep researching the water of the lagoon to understand better its dynamics, which can help obtaining more accurate reflectance values from satellite sensors. Adjustments to the parameters of the AC processors will be made, and sun glint, adjacency effects and bottom reflection will be considered. The methodology will include the analysis of the optical water types observed in the lagoon, and the analysis of the cyanobacteria blooms since MSI is operational (using the standard L2 products), besides the already mention determination of the areas and dates when bottom reflectance can be observed. Precise water quality monitoring of the Albufera Lagoon is urgently needed, since it constitutes a natural reservoir of great environmental, social and economic value for the Valencian region.

Area D - Board 443

Three environmental studies in the Western Mediterranean basin are being developed by students at the Department of Geography of the University of Valencia. The studies use ocean colour data as main input, and they stand out because they have a strong experimental side, with some risk of failure.
The three studies use Sentinel-2 MSI or Sentinel-3 OLCI, applying the Case 2 Regional Coast Colour (C2RCC) neural net algorithm, to derive in-water parameters: 1) the first has the title "Changes in the optical types of the Valencian coastal waters with the OLCI sensor data (Sentinel-3)" and will use OLCI C2RCC reflectance data to classify and analyze optical water type changes in the Gulf of Valencia during the year 2017. This project is related to a more ambitious study to be developed from next year (2019), in which machine learning classification algorithms will be used to derive new OWT schemes in inland and coastal waters (AQUACLASS project); 2) the second work is a bit more experimental, looking for wildfires and inland/coastal water quality relationships in the Valencian region. The title is "Effects of forest fires on water quality: soil loss and changes in water color with Sentinel-2 MSI and Landsat-8". Forest fires that have been followed by torrential rains in the Mediterranean area are affecting the water quality of lakes, reservoirs and coastal areas nearby. To check the capabilities of the MSI sensor, together with Landsat-8, to monitor water quality changes related to post-fire runoff is the main goal of this project; 3) the last study with the title "Spatio-temporal study of dissolved organic matter at the Albufera Lagoon (Valencia) with data from Sentinel-2 MSI” will use 3 years of Sentinel-2 MSI data to derive coloured dissolved organic matter (CDOM) products on the Albufera Lagoon, and study is intra- and inter-annual evolution. The Albufera Lagoon is a freshwater reservoir and estuary on the Gulf of Valencia, it is part of the Natural Park and it is key in the economic, social and environmental life of the region.
The three works and closely related, since patterns of CDOM in the Albufera Lagoon will probably affect the optical water types of the estuary area; as well as the changes in the colour of the water produced by the runoff after wildfires. Knowledge of the water quality of reservoirs, lakes and coastal water in this high populated area, which is also a hub and tourist destination for worldwide visitors, it is more than relevant. The results of these three small projects could help to put the seed for bigger project developments in the near future.

Area D - Board 444

Recently, ocean subsurface particle distributions have been successfully derived using CALIOP (Cloud-Aerosol Lidar with orthogonal Polarization) lidar measurements on the CALIPSO (Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations) satellite. This new application on CALIOP products suggested that space-based lidars could provide information about the depth distribution of optical scattering and significant scientific results were obtained at global scale and in Polar Regions. Assessing the oceanic surface layer's optical properties through CALIOP is one of the reasons of the extension of the CALIOP mission for another 3 years (2018-2020). Despite these results, this new application for ocean retrievals has limitations and uncertainties that need to be further assessed and estimated. The objective of this work is the evaluation of the potential CALIOP ocean products at regional scale. Within this framework, the Mediterranean and Black seas have been considered as case studies and three years of CALIOP (2015-2017) dataset have been considered. The ocean column-integrated depolarization ratio measurements at 532 nm were firstly derived through CALIOP measurements. This parameter allowed deriving the lidar ocean sub-surface column-integrated particulate backscatter. This latter parameter has been compared to the surface hemispheric backscattering coefficients bbp at 443 nm derived by the multi-sensor L3 product (MODIS-AQUA and NPP-VIIRS data) in the Mediterranean and Black seas provided by the Copernicus Marine Environment Monitoring Systems (CMEMS). Analyses of seasonal distributions of CALIOP and CMEMS bbp will be presented for both basins. For the Mediterranean case, the analysis will be extended to the different bio-optical regions following the different “trophic regimes”. Sub-surface backscatter presents good correlation with the bbp in optically complex waters (coastal and/or dominated by terrigenous load) and it demonstrated to be as efficient as the ocean color parameters for the phytoplankton dynamics at seasonal and basin scales. CALIOP measurements allowed also analyzing the day versus night difference in terms of sub-surface backscatter.

Area D - Board 445

Current research continues the investigations [1 and references in], which are aimed to better account of downwelling short wave radiation in hydrodynamic and ecological models of the Black Sea [2].
Statistical relationship between downwelling shortwave radiation at sea level, calculated using the regional meteorological model MM5 [3], and the standard satellite product – photosynthetically active radiation, obtained from measurements of SeaWiFS [4] and MODIS-Aqua/Terra [5, 6] instruments has been found in different regions of the Black Sea in different seasons. The average value of the correlation coefficient lies in the range of 0.7 to 0.9. RAS funded this research [grant numbers АААА-А18-118020890112-1, АААА-А18-118020790229-7 and АААА-А18- 118012690119-7]. This work was partly supported by the Russian Foundation for Basic Research, [projects numbers 17-05-00113 and 18-45-920070].
References
1.Suslin V.V., Korolev S.N., Kucheryaviy A.A., Churilova T.Ya., Krivenko O.V.
Photosynthetically Available Radiation On Surface Of The Black Sea Based On Ocean Color Data // Proceedings of SPIE - The International Society for Optical Engineering 21, Atmospheric Physics. Сер. "21st International Symposium on Atmospheric and Ocean Optics: Atmospheric Physics" 2015. С. 96800T.
2. Korotaev G.K., Dorofeev V.L., Motyzhev S.V., Belokopytov V.N., Stanichny S.V., Ratner Y.B., Suetin V.S., Suslin V.V., Palazov A., Malciu V., Zatsepin A., Nesterov E. Contribution Of The Black Sea Observing System To Ecoop // Ocean Science Discussions. 2011. Т. 8. № 4. С. 1695-1722.
3. Shokurov, M.V., “Numerical modeling of the atmospheric circulation over the Black Sea”, Ecological safety of coastal and shelf zones and comprehensive use of shelf resources: Collected scientific papers, 25(2), 91-113 (2011).
4. NASA Goddard Space Flight Center, Ocean Ecology Laboratory, Ocean Biology Processing Group. Sea-viewing Wide Field-of-view Sensor (SeaWiFS) Ocean Color Data; 2018 Reprocessing. NASA OB.DAAC, Greenbelt, MD, USA. doi: 10.5067/ORBVIEW-2/SEAWIFS/L2/OC/2018. Accessed on 11/08/2018
5. NASA Goddard Space Flight Center, Ocean Ecology Laboratory, Ocean Biology Processing Group. Moderateresolution Imaging Spectroradiometer (MODIS) Aqua Ocean Color Data; 2018 Reprocessing. NASA OB.DAAC, Greenbelt, MD, USA. doi: 10.5067/AQUA/MODIS/L2/OC/2018. Accessed on 11/08/2018
6. NASA Goddard Space Flight Center, Ocean Ecology Laboratory, Ocean Biology Processing Group. Moderateresolution Imaging Spectroradiometer (MODIS) Terra Ocean Color Data; 2018 Reprocessing. NASA OB.DAAC, Greenbelt, MD, USA. doi: 10.5067/TERRA/MODIS/L2/OC/2018. Accessed on 11/08/2018

Area D - Board 446

The Black Sea belongs to Case 2 waters, where colored dissolved organic matter (CDOM) is main optically active component. CDOM domination in light absorption makes assessment of chlorophyll a concentration (Chl-a) based on remote sensing data is complicated. Therefore quantitative descriptions of the optical properties of CDOM is crucial for correct retrieving of Chl-a by regional three-band algorithm based on remote sensing data (Suslin, Churilova, 2016).
For this analysis we used bio-optical data measured in 2014–2018 in different regions of the Black Sea (open and coastal waters). The spectral distribution of CDOM absorption coefficients (a_CDOM (λ)) was described with exponential function and two parameters were determined: a_CDOM (λ) coefficients at reference wavelength (a_CDOM (440)) and exponential slope coefficients fitted to 350- 500 nm (S_CDOM). The variability in aCDOM(440) and SCDOM was assessed for open and coastal waters. For the open waters seasonal differences in aCDOM(440) and SCDOM were revealed. This seasonality was caused by effect of high solar radiance in summer and shallow upper mixed layer.
The variability ranges of these parameters measured in open and coastal waters were overlapped, which allowed to analyzed data of open and coastal waters jointly. For the joint dataset it was obtained that aCDOM(440) varied from 0.015 m-1 to 0.32 m-1 and SCDOM was in a range from 0.014 nm-1 to 0.034 nm-1. The Chl-a (in sum with phaeopigments) varied from 0.15 mg m-3 to 3.5 mg m-3. Under this diapason Chl-a variability the values of aCDOM(440) were not correlated with Chl-a. The analysis showed that the spectrum slope coefficient was inversely and nonlinearly related to the CDOM absorption coefficient. The obtained relationship was applied for regional algorithm (Suslin, Churilova, 2016). Comparison of the measured CDOM absorption coefficients with colored detrital matter absorption coefficients retrieved by the regional algorithm showed high accuracy of the refined regional algorithm.
RAS funded this research [grant numbers АААА-А18-118020890112-1, АААА-А18-118020790229-7 and АААА-А18-118012690119-7]. The research was partly supported by the Russian Foundation for Basic Research, [projects numbers 17-05-00113 and 18-45-920070].

References
Suslin, V., & Churilova, T. 2016. The Black Sea regional algorithm of separation of light absorption by phytoplankton and colored detrital matter using ocean color scanner’s bands from 480-560 nm. International Journal of Remote Sensing 37 (18): 4380-4400.

Area D - Board 447

The aquatic science products from the Earth Observation sensors developed mainly for land applications require vicarious calibration coefficients to increase the performance of the sensor in the case of highly absorbing surface such as coastal water which could be outside the range of validity of the radiometric coefficients provided by the mission.
Within BELHARMONY project (funded by BELSPO, 2018-2020) the gains retrieval for the MSI/S2, OLI/L8 and Proba-V data acquired over coastal environment is performed by using satellite data acquired over the AERONET-OC stations (Zibordi et al., 2009) when in-situ atmospheric and water products are available. All the in situ measurements considered for the simulation of the sensor signal satisfy the quality assurance criteria, which means that the AERONET products are at level 2.0.
The sensor signal is simulated by using the 6SV radiative transfer model (Kotchenova et al., 2007; Vermote et al., 1997). In order to minimize the aerosol effect on the remotely sensed signal as reported in Bassani et al., 2015, only images with contemporary in-situ AERONET products with low aerosol loading represented by the aerosol optical thickness at 550nm are considered suitable for this study. Regarding the micro-physical properties of the aerosol the inverse AERONET products are often lacking at level 2.0, to overcome this condition the size distribution is assumed bi-modal dominated by fine or coarse aerosol mode depending on the Angstrom exponent, which is the AERONET direct product sensitive to the aerosol particle size. Furthermore, the AERONET-OC normalized water-leaving radiance is also input for simulation of sensor signal by modifying the source code of the 6SV model. A new method for the spectral interpolation of the water AERONET product was proposed by mixing a cubic spline and an analytic function to obtain a hyperspectral water-leaving reflectance over the entire VNIR (visible and near-infrared) domain starting with values at the AERONET channels.
In order to achieve an accurate simulated signal, the specular reflection of the water surface is considered by introducing as input the speed and surface direction of the wind provided by AERONET station and the approximate value obtained by the online NMMB/BSC-Dust model, respectively. The dataset was screened for the wind speed following the results presented in Moore et al., 2000; the threshold is fixed to 5m/s to limit the increasing of the water-leaving reflectance due to the white-caps over the water surface.
The last screening is represented by the conditions about the geometrical configuration of the acquisition and illumination reported in Pahlevan et al., 2014 to minimize the sun glint effect by restricting the relative azimuth and the solar zenith angle.
The 6SV model was run for each selected images. The simulated TOA (top of atmosphere) radiance continuous within the discrete-spectral domain VNIR and sampled every 2.5nm has to be convolved to the spectral response function (SRF) of the sensor to obtain the simulated sensor signal (REF). Finally, the gain is calculated as the ratio of REF/MEAS, where MEAS stands for the observed signal.
The removal of available data allows to the reliability and validity of the input data and the outcome of this study. To this aim, this study requires to work over a sufficient number of observations in order to reduce the uncertainty in the final gain retrieval. New data in the next months are expected to be available over the AERONET stations improving the quality of the retrieved gains.
References:
- Zibordi,G.; Melin, F.; Berthon, J.-F.; Holben, B.; Slutsker, I.; Giles, D., D’Alimonte, D.; Vandemark, D.; Feng, H.; Schuster, G.; Fabbri, B.E.; Kaitala, S.; Seppala, J.: AERONET-OC: a network for the validation of ocean color primary products, Journal of Atmospheric and Oceanic Technology, 26(8), 1634–1651, doi:10.1175/2009JTECHO654.1, 2009.
- Vermote, E.F.; Tanré, D.; Deuzé.; Herman, M.; Morcrette, J.J.; Kotchenova, S.Y.: Second Simulation of a Satellite Signal in the Solar Spectrum-Vector (6SV). Available online: http://6s.ltdri.org (accessed on 23 August 2016)
- Kotchenova, S. Y. and Vermote, E. F.: Validation of a vector version of the 6S radiative transfer code for atmospheric correction of satellite data. Part II. Homogeneous Lambertian and anisotropic surface, Appl. Optics, 46, 4455-4464,doi:10.1364/AO.46.004455, 2007.
- Bassani, C.; Manzo, C.; Braga, F.; Bresciani, M.; Giardino, C.; and Alberotanza, L.: The impact of the microphysical properties of aerosol on the atmospheric correction of hyperspectral data in coastal waters, Atmos. Meas. Tech., 8, 1593-1604, doi:10.5194/amt-8-1593-2015, 2015.
- Pahlevan, N., Lee, Z., Wei, J., Schaaf, C. B., Schott, J. R., Berk, A.: On-orbit radiometric characterization of OLI (Landsat-8) for applications in aquatic remote sensing, Remote Sensing of Environment, 154, 272-284, doi: 10.1016/j.rse.2014.08.001.

Area D - Board 448

On the basis of inherent optical properties (IOPs), optically active substances (OAS: chlorophyll a and total suspended matter) and vertical profiles of temperature and salinity, the optical properties in the coastal waters of northwestern Estonia have been investigated. The analysis of thermohaline structure reviled three different situations in terms of vertical stratification: 1) mixed profiles up to 25 meters, 2) intermediate situation and 3) profiles with distinct thermocline. Comparison between the vertical profiles of temperature and IOPs showed similar patterns in vertical variability. In the situation of mixed water column the optical parameters did not change significantly through the entire water column. In the stratified water column the concentrations of OAS changed rapidly and the vertical variability of the optical parameters were significantly higher compared to the mixed layer. The results showed that the stratification of the water column regulates the variability of the optical parameters through distribution of the optically active substances (OAS).
In addition, the spectral remote sensing reflectance Rrs(λ) were calculated from the measured IOPs. The values of Rrs(λ) were calculated from IOPs measured at three different depths: surface layer, 10 meters and 20 meters. Performed simulation allows to assume how surface Rrs(λ) values would change if the upwelling events occur. The results showed that in the upwelling region the signal measured by the remote sensor may decrease up to 89% as a result of deep water uptake. Also the Rrs(λ) spectra would be considerably smoother.
Furthermore, the obtained simulation results of Rrs(λ) for different water masses were confirmed by measurements from an independent dataset of radiance (Lw(λ)) and irradiance(Ed(λ)) measurements (TriOS RAMSES) collected during the upwelling event. The above water irradiance and radiance measurements were used for the calculation of the Rrs(λ). Two different types of water masses were distinguishable from the Rrs(λ) spectra: 1) typical coastal water which strongly reflects in the green part of the spectrum (550-580 nm) 2) upwelled water with considerably smoother Rrs(λ) spectra. The difference in Rrs(λ) spectrum shapes allowed to use the ratios between blue (443 nm) and green (560 nm) values as an indicator for the detection of upwelled water (Rrs(443)/Rrs(560)>0.7). Results obtained from the research should be considered when developing remote sensing algorithms for coastal waters of the Baltic Sea.

Area D - Board 449

Inland and coastal waters contribute to various ecosystem services. As they provide resources for human use and important and diverse habitats for aquatic life, they also support services that depend on the status of local ecosystem. The assessment of the state of these ecosystems is important from regional to global scales. Sentinel series satellites Sentinel-2/MSI and Sentinel-3/OLCI allow to derive optical properties of these waters with frequent time frame and sufficient spatial resolution to monitor small lakes as well. This allows to include satellite-based estimates for environmental monitoring and management as an additional source of information. As the accuracy of the satellite products is prone to errors in case of small water bodies and in addition to added complexity by highly absorbing or chlorophyll a rich waters, the satellite derived products must be analysed beforehand based on the field measured dataset. Here, the performance of the S3/OLCI and S2/MSI data is analysed over optically complex Estonian lakes and coastal waters (aCDOM(442) 1.3 – 14 m-1; Chlorophyll a 9.5 – 71.8 mg m-3; TSM 4.3 – 25.7 mg/l). In situ remote sensing reflectance (Rrs) measured by a set of three hyperspectral radiometers is assessed. Major sources for uncertainty budget result from sensor calibration, sensor instability, surface reflectance and environmental variability. The relative combined standard uncertainty for in situ Rrs is calculated by combining contributions from each input. The uncertainties are accounted for the validation of S2/MSI and S3/OLCI radiometric products over different conditions and also compared with the uncertainty estimates derived from satellite data processing chain. This allows improved comparability of radiometric in situ data with S3/OLCI and S2/MSI data. The validation of radiometric products indicates higher uncertainties in blue wavelengths which are often overestimated by satellite derived products, whereas better precision is obtained for green and red wavelengths. From the water quality parameters, the best agreement with in situ data was achieved with the sum of the detritus and dissolved organic matter absorption coefficients. In case of in—water products (IOPs, chl-a, TSM), the satellite-based uncertainty estimates were sensitive only to strong biases in TSM product. The uncertainty estimates for in situ or satellite data allow to estimate the data quality for various applications in coastal and inland waters.

Area D - Board 450

Overall, the Baltic Sea is optically dominated by coloured dissolved organic matter (CDOM). However, in near-coastal areas and in very inner coastal bays total suspended matter (TSM) also has a large optical influence. The influence decreases with distance from the shore.
TSM consist both of an organic and an inorganic fraction and is usually used as an indicator for coastal influence. In natural waters, the concentration of suspended particles is directly related to the scatter which is an inherent optical property (IOP). Particle scatter can be measured from space as the remote sensing reflectance is a function of absorption and scatter of all optical substances in the water. The reflectance generally decreases with absorption while it tends to increase with scatter.

Usually, TSM, Chl-a, CDOM and diffuse attenuation are the main water quality products derived from remote sensing data. The C2RCC processor that can be applied to OLCI (Sentinel-3) data from ESA's Copernicus mission, however, also provides absorption and scattering products. The concentration of suspended particles, e.g., can then be derived from particle scatter if the specific scattering properties of the respective water body is known. We use here the Baltic Sea-specific scatter properties measured in situ in order to retrieve both total and inorganic suspended matter from Sentinel-3 data.

The generated Sentinel-3 images show that most of the inorganic suspended matter falls out rather close to the shore, whereas TSM is carried much further off-shore. The images clearly highlight the areas that are strongly influenced by terrestrial matter. Differences between the north-eastern Baltic and the south-eastern Baltic proper can be explained by the difference in hydrology and bathymetry and the influence of wind-wave stirring.

The method is of interest for coastal zone management as well as for assessing the effect of seasonal changes in SPM loads. It can also be used to evaluate the effect of climate change which has lead to an increase of extreme storm and flooding events that are usually accompanied by increased erosion and run-off from land.

Area D - Board 453

The Baltic Sea and the Gulf of Riga are relatively complex and dynamic environments, therefore their condition should be regularly monitored in order to avoid ecological problems and management. However marine monitoring is financially costly process and therefore a great attention is paid to the creation of cost-efficient methods and tools to make environmental assessment as effective as possible.
The Ocean and Land Color Instrument (OLCI) on board of ESA Sentinel-3A satellite provide constant measurements and thus an opportunity to see changes in the marine environment on daily basis. Yet to fully include OLCI products in monitoring system, they have to be validated with in situ measurements.
In 2018, from May to November, we conducted three sampling campaigns in Latvian territorial waters, where we collected chlorophyll a samples and tested the standard Case-2 Regional / Coast Colour (C2RCC) atmosphere model performance, to look at possible matching with the chlorophyll a and total suspended matter field data. In situ chlorophyll a values were determined using different methods: 1) CTD Chlorophyll a measurement at 1 m depth; 2) Chlorophyll a top layer sample (0 - 1m); 3) sampling of the integrated hose (0-10 m) using HELCOME recommended ethanol extraction method.
The linear regression model showed the strongest relationship between OLCI chl a values and field measurements in open Baltic, where the determination coefficient explained up to 54% of the CTD and upper layer (0 - 10m) chlorophyll a concentration variation during the testing period. In contrast the correlation slightly weakened in the Gulf of Riga, explaining only 33% of data variance.

Keywords: OLCI; Sentinel-3; Baltic Sea; Gulf of Riga; C2RCC

Area D - Board 454

The Copernicus Atlantic Meridional Transect for Sentinel Fiducial Reference Measurements (AMT4SentinelFRM) collected high FRMs for the validation of a range of Sentinel-1, -2A, -2B, -3A and -3B products during annual voyages between the UK and the South Atlantic during September to November 2016 and 2017. AMT4SentinelFRM builds on the Atlantic Meridional Transect programme which has been running for 28 years and was established in 1995 in collaboration with NASA, as an independent platform to validate SeaWiFS Ocean Colour data. It not only provided vital FRMs for the duration of the SeaWiFS mission, but also served as a developmental and inter-comparison platform for selecting the most accurate ocean-colour algorithm for SeaWiFS. AMT4SentinelFRM has now developed the programme into a multi-sensor, multi-mission platform for satellite validation. In this paper we validate Sentinel-3A OLCI, VIIRS and MODIS-Aqua algorithms in both the Atlantic Ocean and a range of Case 2 waters, including the Baltic Sea, North Sea and Western English Channel. The Baltic Sea is optically dominated by coloured dissolved organic material (CDOM; >1.0 m-1) which makes accurate ocean colour remote sensing challenging in these waters. The Western English Channel (WEC) oscillates between case 1 and 2 type waters and has medium levels of CDOM (0.06 – 0.25 m-1) and Total Suspended Material (TSM; 0.5 – 1.2 g m-3). We firstly quantify differences between sensors used on the different campaigns, then present an uncertainty budget for the radiometric measurements. Finally we evaluate a range of different atmospheric correction processors for Sentinel-3A OLCI to identify the most accurate in different regions. For the Atlantic Ocean, traditional analysis of water samples for HPLC Chlorophyll-a are complemented with an underway measurement system that collects along track particulate absorption measurements to estimate surface chlorophyll concentration with high accuracy (± 10% relative error). These quasi-continuous measurements maximise the spatial and temporal frequency of data acquisition to 40,000 underway minute-binned optical measurements, which in turn enhances the number of potential in situ match-ups with satellite data to approximately 300 match-ups. This allows us to assess problems related to the spatial-resolution of the sensor and provides data for a comprehensive comparison with NASA VIIRS and MODIS-Aqua satellites.

Area D - Board 455

The Canadian coastal waters are optically complex characterized by high freshwater inputs, high productivity and physical features highly dynamics (Fuentes-Yaco et al. 1997, Loos et al. 2017, Laliberté et al. 2018). These coastal ecosystems are subject to a variety of environmental changes from both natural and anthropogenic origins. To quantify these changes, the Department of Fisheries and Oceans Canada (DFO) as developed an extensive monitoring program of the coastal waters. This includes a network of autonomous scientific buoys that measure physical, bio-optical and radiometric quantities (Bélanger et al., 2017). In order to validate the performance of the new Sentinel 3 Ocean and Land Instrument (OLCI) over the Canadian coastal environment, we performed a matchup exercise between OLCI data and in situ radiometric measurements. First, we will test two atmospheric correction algorithms available for OLCI: the standard ESA (Moore et al. 1999) for Case 2 water and an alternative neural network algorithm adapted for coastal environment (C2RCC, Brockman et al., 2016). Secondly, we will make an inter-comparison of the radiometric measurement between OLCI and contemporary ocean color sensors (MODIS/VIIRS). For the ESA atmospheric correction, we found that the overall absolute relative percent difference (RPD) in the Canadian coastal water is 43% root mean square error (RMSE) is 0.003, with RPD found in the blue region reaching as much as 88% (RMSE = 0.006) and 20% (RMSE = 0.0018) in the green/red (510, 560, 665 nm) channels. In the blue band, up to 22% of the water leaving reflectance retrieval were negative at 412 nm. Canadian coastal waters are weakly to moderately turbid showing large concentration of both phytoplankton and CDOM resulting in a typical reflectance spectrum with very low values in the blue and weak but significant signal in the red/NIR. This often leads to an underestimation of the reflectance in the short wavelength for standard atmospheric correction. The C2RCC first results seem promising with a good estimation in the blue and the red/NIR channel. The synergy uses of OLCI, MODIS and VIIRS will help the detection of changes in the Canadian coastal water.

Area D - Board 456

The Arctic is subject to substantial changes due to the greenhouse gas induced climate change. While impacts on lateral transport pathways such as rivers have been extensively studied yet, there is little knowledge about ecological and geological reactions of nearshore environments, even though those are of high importance for native communities. In this study, we use the extensive Landsat archive with comparable data from 1982 on to investigate sediment dispersal and sea surface temperatures under changing seasonal wind conditions in the nearshore zone of Herschel Island in the western Canadian Arctic. Herschel Island lies approximately 120 km west of the Mackenzie Delta and experiences varying input from the Mackenzie River plume depending on the prevailing wind direction. Even in the absence of an extensive in-situ dataset, we reveal clear differences between the two prevailing wind conditions (E and NW). During E wind conditions, the Mackenzie River Plume gets distributed over large parts of the Canadian Beaufort Shelf and is the main influencing factor for nearshore sediment dispersal and sea surface temperature dynamics. Contrary, the nearshore dynamics during NW wind conditions are not affected by the Mackenzie River plume, revealing the local nature of the nearshore environment. We’ve applied several recently published turbidity models to our dataset, revealing a large range of modelled values, with maximums between 30 and 200 FTU. This highlights issues to calculate accurate bio-optical models in dynamic nearshore environments such as in the Arctic. These issues contain the challenging atmospheric correction, a seasonal sea ice cover and fast changing weather conditions, together with the remote location and thus difficult accessibility of most Arctic research sites. In order to overcome these issues, we plan to validate our dataset with field measurements collected in summer 2018. This validation will be the first step to assess the changes of Arctic nearshore environments under changing climate conditions.

Area D - Board 457

I studied spatial and temporal variations in MERIS (Medium Resolution Imaging Spectrometer) satellite fluorescence (F_sat) and fluorescence efficiency (Phi_sat) over the Arctic (60-90N) in order to research the relationship between chlorophyll-a (Chl-a) and F_sat and its influence factors. Analyzing 10 years of MERIS data I focused on the months May to September due to satellite data availability being limited by ice extent, viewing geometry and lack of sunlight during the rest of the year. Due to the heterogeneity of the Arctic I also separated the domain into subdomains (e.g. the Greenland-Norwegian Sea, Barents Sea, etc.).

F_sat and Phi_sat show strong seasonal variability with overall lower values in summer (June to August), whereas the seasonal cycle differed highly from region to region. On the one hand temporal variations align with seasonal changes in stratification and nutrient availability, on the other hand spatial variations correspond with ocean currents in the Arctic such as the North Atlantic Current (NAC) and Trans Polar Drift (TPD). This feature can be seen in figure 1 showing the Phi_sat climatology for July.
The relationship between F_sat and Chl-a varies significantly from domain to domain. This is displayed in figure 2, which depicts climatological F_sat vs. Chl-a in July for the North Atlantic (2a) and Bering Strait (2b). In the majority of regions F_sat and Chl-a are strongly correlated. In some domains and months, however, there is only weak correlation or a much lower slope between the two variables, as seen in fig. 2b. The variations in the relationship might be associated with species, nutrients, stratification or other factors.
To test certain hypotheses on the influence factors on Phi_sat, I compared the MERIS data with climatologies of nutrients, salinity and mixed layer depth. Depending on month and area the major factors affecting Phi_sat varied. Results show, that in some cases nutrient stress is linked to a reduction of fluorescence efficiency leading to a lower F_sat signal. Another major contribution, however, is the stratification and creation of subsurface Chl-a maxima by a shallowing of the mixed layer depth. Under these conditions the MERIS still detects some Chl-a but the F_sat signal is strongly decreased due to the high absorption by water at the fluorescence wavelength 683 nm [1].

The seasonal and regional changes I observed in Phi_sat correspond with the current understanding of phytoplankton dynamics in the Arctic such as cycles of stratification [2] and advection of nutrient rich (e.g. Arctic) or depleted (e.g. Atlantic) water. This study highlights the capabilities of satellite fluorescence from MERIS as well as its successor OLCI on Sentinel 3a and b in the fields of ocean colour, ocean biology and ocean ecology.


Captions:
Fig. 1: summer climatology of Phi_sat from MERIS. High Phi_sat in red, low Phi_sat in green. Grey areas show invalid pixels at any given time in the climatology (May-Sep. 2002-2012).


References:
1. J. Fischer, U. Kronfeld (1990). Sun-stimulated chlorophyll fluorescence 1: Influence of oceanic properties. International Journal of Plankton Research, Vol 11:12, 2125-2147, DOI: 10.1080/01431169008955166

2. S. Chiswell, P. Calil, P. Boyd (2014). Spring blooms and annual cycles of phytoplankton: A unified perspective. Journal of Plankton Research, Vol 37:3, 500-508, DOI:10.1093/plankt/fbv021

Area D - Board 458