Ocean colour products from geostationary platforms , opportunities with Meteosat Second and Third Generation

Introduction Conclusions References


Introduction
There is an established user need for a range of water quality and bio-geochemistry information services for marine, coastal, estuarine and lake environments.These needs are expressed in several user surveys conducted by European Commission projects such as GMES PURE (Albert et al., 2014), MARCOAST (Brockmann et al., 2008;Ruddick et al., 2008), CoBiOS (Kaas and Peters, 2012), and FRESHMON (Stelzer et al., 2011).A critical component of these requirements has been the national obligations of European Union (EU) Member States to report on water quality under the Water Framework Directive (WFD) (European Commission, 2000) and the Marine Strategy Framework Directive (MSFD) (European Commission, 2008).These requirements have guided the development of remotely sensed ocean colour products which provide synoptic coverage of a range of water quality and bio-geochemistry indicators.They have also motivated the European Commission's funding of Sentinel-3 satellites as part of the Copernicus Space Component in support of the Copernicus Marine Environment Monitoring Service (CMEMS).Globally, international space agencies, for example in the United States, Japan, Korea, China, India, Brazil, Russia, and Canada, are investing in ocean colour programmes with similar goals.
Ocean colour observations are commonly performed from polar orbiting satellite platforms which include the Copernicus Sentinel-3 series.Ocean colour coverage from polar observations is, however, significantly reduced due to cloudiness, as well as gaps between orbits and sun glint.For example, polar instruments with data aggregated to 4 km spatial resolution provide typically between 4 and 8 % coverage of the global ocean per day, depending on the swath-width and glint avoidance capabilities (Gregg, 2007).Merger of data from multiple polar missions increases the global coverage but is not straightforward because of differences between instruments (Kwiatkowska and McClain, 2009).High spatio-temporal marine processes thus cannot be adequately resolved by infrequent observations from polar platforms (Antoine et al., 2012;Ruddick et al., 2014).Introduction

Conclusions References
Tables Figures

Back Close
Full There is a user need to supplement global polar observations with geostationary ocean colour coverage (Antoine et al., 2012;Albert et al., 2014).The major benefit of geostationary ocean colour is improvement in spatial coverage because frequent daily observations are more likely to record cloudless conditions as shown in Fig. 1 (Ruddick et al., 2014).High-temporal monitoring also gives information on quickly varying processes in coastal ocean, estuarine zones, and lakes which are of special interest to many users (Mouw et al., 2015).The processes include tidal dynamic, eddies, fronts, sediment transport, coastal erosion, river plumes, and hazards such as harmful algal blooms (HAB) and oil spills.It can also provide the required input to coupled models to meet operational needs for marine and coastal nowcasts and forecasts and to improve model variability at intra-seasonal and inter-annual time scales.(Neukermans et al., 2009) as first demonstrated the potential of geostationary ocean colour remote sensing to capture the tidal variability of suspended sediments with the Spinning Enhanced Visible and InfraRed Imager (SEVIRI).Subsequently, the first ocean colour instrument on a geostationary platform has been successfully demonstrated by the Korea Institute of Ocean Science & Technology (KIOST) (Choi et al., 2012;Ryu et al., 2012).The Institute's Geostationary Ocean Color Imager (GOCI) provides 2500 km×2500 km coverage in hourly intervals centered around the Korean peninsula and the follow-on mission, GOCI-II, will also support Full Disk coverage (Park, 2015).It has been shown that recurrent daily observations from a geostationary platform significantly increase the coverage of water quality and bio-geochemical processes (Wang et al., 2013).GOCI provides a three-fold improvement in coverage based on its acquisition of 8 scenes per day in comparison with a single daily acquisition.GOCI applications include fishing ground index, HAB index, underwater visibility, surface current vectors, water quality index and primary productivity.
EUMETSAT operates a series of geostationary platforms called Meteosat Second Generation (MSG) which carry Spinning Enhanced Visible and InfraRed Imagers (SE-VIRI) with solar bands in the range of 635, 810 nm, and 1640 nm, as shown in Table 1 (Schmetz et al., 2002).SEVIRI has reduced spectral and spatial resolution and a de-Introduction

Conclusions References
Tables Figures

Back Close
Full graded signal-to-noise ratio compared to standard ocean colour missions (Govaerts and Clerici, 2004).Nevertheless, the bands in red and near infra-red (NIR) allow for a limited range of water quality and bio-geochemistry products, those associated with water turbidity and very high plankton biomass.The spatial resolution of 3 km at nadir results in coarser spatial resolution over Europe: for example, 6 km resolution in the southern North Sea.Nonetheless, SEVIRI's frequent imaging every 15 min has been demonstrated to improve temporal coverage of coastal water clarity, tidal effects and sediment transport (Ruddick et al., 2014;Neukermans et al., 2012).SEVIRI has also been shown to be capable of distinguishing specific bio-geochemical features, such as coccolithophore blooms as, for instance, demonstrated in the Bay of Biscay (Vanhellemont et al., 2013) and displayed in Fig. 1.EUMETSAT's Meteosat Third Generation (MTG) Imaging satellites, with the first of the series planned for launch in 2020, will carry Flexible Combined Imager (FCI) instruments which are the upgraded continuation of MSG SEVIRI (EUMETSAT, 2015).FCI instruments will operate additional spectral bands in the visible blue and green wavelengths, 444 and 510 nm, as shown in Table 2, that are potentially suitable for additional ocean colour products of which the most important are chlorophyll concentrations.Frequent imaging of the Full Disk every 10 min has the potential to further improve spatial and temporal coverage of marine, coastal, estuarine and lake bio-geochemical processes.The spatial resolution of 1 km at nadir is an enhancement on SEVIRI's 3 km resolution and it is suitable for global ocean observations as well as provides meaningful improvement for coastal and lake studies.
This paper describes the ongoing effort to develop operational ocean colour products and data services from EUMETSAT's geostationary missions.Current work focuses on user requirements and scientific constraints.Introduction

Conclusions References
Tables Figures

Back Close
Full This review of user requirements for geostationary operational ocean colour products includes outcomes of the European Commission projects, like GMES PURE, MAR-COAST, CoBiOS, and FRESHMON, as well as additional user questionnaires and interactions conducted during the current study.
SEVIRI's full disk coverage provides observations over Europe and Africa, therefore user requirements from both continents are compiled.In Europe, the requirements are largely driven by the EU directives, WFD and MSFD.The directives call for monitoring of coastal and marine water quality.The WFD also requests the monitoring of inland lakes as well as rivers, transitional waters, nearshore waters and groundwater bodies.
In Africa, monitoring of coastal water quality and quality of African lakes is of high priority but less formalized.Further applications include sediment transport, ecosystem modelling, offshore operations, and carbon burial by coccolithophores.

Water resource monitoring in response to European Union directives
For European coastal waters, only MSFD is of relevance to SEVIRI because WFD focuses on waters within "1 nautical mile" of the coast which are impossible to observe with the coarse SEVIRI spatial resolution.The main goal of the MSFD is to achieve Good Environmental Status of EU marine waters by 2020.The Directive defines Good Environmental Status (GES) as: "The environmental status of marine waters where these provide ecologically diverse and dynamic oceans and seas which are clean,  , 1996).For these indicators, SEVIRI's use in combination with data from dedicated polar orbiting missions shows most capabilities (Ruddick et al., 2014).SEVIRI data alone provides the transparency indicator in turbid waters via the Secchi depth and could potentially support identification of the most extreme high-biomass algal blooms.Chlorophyll a concentration could not be obtained from SEVIRI but may be supported by MTG FCI instruments.The hydrographical condition descriptor requires monitoring of turbidity or suspended particulate matter indicators which can be provided by both the SEVIRI and FCI instruments.SEVIRI and FCI can provide the spatial extent and the temporal resolution required by the directive.Product time series can be made available over many years, or even decades for trend identification.The spatial resolution of SEVIRI in European waters allows adequate monitoring of offshore zones, but still presents challenges in coastal waters.Coastal studies will benefit more from the FCI instrument.
For European lakes, WFD defines plankton and general hydromorphological condition parameters which are relevant to SEVIRI and FCI.Out of these parameters, water turbidity and/or transparency (measureable via Secchi depth) are already observable from SEVIRI while additional plankton concentrations can be supported by FCI.Massive blooms of cyanobacteria are also relevant for WFD and could possibly be flagged using SEVIRI.An obvious limitation of SEVIRI for inland water applications is the spatial resolution.Only the very largest European lakes are observable, including Lake Balaton (max.length 77 km, max.width 14 km), Lake Geneva (73, 14 km) and the Bodensee/Lake Constance (63, 14 km).The increased resolution of the FCI instrument will support coverage of additional lakes.

African water resource monitoring
In African coastal and inland waters the environmental problems are often more acute than in Europe in their affects on human health and food supply (e.g.fisheries, aquaculture).Rio de Janeiro's United Nations Conference on Environment and Development of 1992 has generated international interest in preservation of biodiversity in Africa (UN, Introduction

Conclusions References
Tables Figures

Back Close
Full 1992).The identified user needs are tracking of harmful algal blooms and monitoring of water quality.Further needs address eutrophication, water transparency, detecting extreme high-biomass or cyanobacteria blooms, scums and floating vegetation like intensive proliferation of water hyacinths (Eichhorna crassipes) in lakes.In coastal regions, it is also required to monitor sediment transport, underwater visibility for maritime operations, and high frequency physical/biological processes.Geostationary spatial resolution over Africa is better than over Europe, thus SEVIRI is useful for offshore, coastal and lake applications and FCI promises even better coverage and additional products.The lakes that can be monitored with SEVIRI include Lake Victoria/Nam Lolwe/Nalubaale (max.length 337 km, max.width 250 km), Lake Tanganyika (673, 72 km) and Lake Malawi/Nyassa (560, 75 km).SEVIRI can deliver over a decadal product time series for trend analysis.

Additional user requirements
User needs also address transport of sediments in coastal waters -something which is of the major interest to coastal zone managers because of changes in bathymetry.
Sediment transport is critical for waterway navigation, offshore construction, and for the understanding of coastal erosion and sedimentation that affect flooding defences, real estate, recreation and aquaculture.SEVIRI and FCI can provide relevant turbidity and/or suspended particulate matter concentration products.The spatial resolution required is highly dependent on the specific application.For example, sediment transport in the vicinity of offshore structures or ports may involve processes at the scales of metres or tens of metres (Vanhellemont and Ruddick, 2014).On the other hand, sediment transport models have typical resolutions of 100 m-10 km with the coarser resolution models being used for large scale transport, e.g. at the scale of the southern North Sea.The temporal resolution required for sediment transport applications in regions of tidal variation is typically hourly, which is suitable for both SEVIRI and FCI instruments.
Ecosystem modelling is another application.It has been driven by the need to manage eutrophication and water quality (Lenhart et al., 2010).The models use sunlight Introduction

Conclusions References
Tables Figures

Back Close
Full FCI can improve on this and also incorporate attenuation due to phytoplankton.SEVIRI and FCI can resolve high frequency tidal and diurnal processes such as the timing of tidal resuspension/advection with respect to the sunlight maximum.In an ecosystem model, these processes have been demonstrated to modify the net effect of light on photosynthesis compared to daily-averaged values (Desmit et al., 2005).Most ecosystem models have spatial resolutions similar to the SEVIRI resolution.
Diving operations in context of offshore constructions and environmental monitoring are identified as another user niche which can be supported by horizontal visibility or turbidity products.SEVIRI high frequency data are well-suited to define optimal time windows for diving, which typically last 1-3 h because of tidal variability; while FCI will improve on the spatial coverage.
Coccolithophores are a class of calcifying phytoplankton of particular interest to global climate change, both past (de Vargas et al., 2007) and future, because of their role in oceanic inorganic carbon chemistry, their sensitivity to oceanic acidity (Smith et al., 2012) and for their importance in the long-term sink of carbon via sedimentation and burial (Milliman, 1993).Coccolithophores are distinguished by a covering of calcium carbon plates known as coccoliths which are highly reflective, thus making this species easily visible from space during blooms (Groom & Holligan, 1987).SEVIRI can detect the strongest coccolithophore blooms, mark their extent as the blooms are generally large scale, and map them with frequent imagery.A coccolithophore bloom flag and red marine reflectance are feasible products from SEVIRI.The FCI instrument will be able to distinguish coccolith concentrations.
Users have also identified the SEVIRI red marine reflectance as a parameter defining diurnal variability of individual marine, coastal and lake waters.Knowledge of this Introduction

Conclusions References
Tables Figures

Back Close
Full variability can facilitate improved calibration and validation protocols, such as matchups between satellite and in situ measurements and satellite to satellite matchups.

Summary product requirements
Table 3 gives the range of ocean colour products feasible from the SEVIRI instruments which have been requested through user surveys.Concerning product requirements, most applications call for spatial resolutions better than SEVIRI's within a range of several hundreds of meters to a few meters.Accuracies for the products are difficult to obtain, although most users recognize validation and product confidence as being highly important.The accuracy label classified as "threshold" must pass certain threshold criteria, "absolute" must provide quantitative accuracy measures and "scientifically sound" has no accuracy requirements but the algorithm must be validated.A need for Near Real Time (NRT) product dissemination is not identified except for a few applications including extreme high-biomass HABs, planning of offshore diving activities, and certain short range ecosystem models.Important for most applications is however the availability of multi-year historical data.
User-requested products that could be additionally obtained from FCI instruments include chlorophyll a concentrations, algal pigment absorption coefficient at 443 nm, and diffuse attenuation coefficient spectrum.

Scientific constraints of ocean colour SEVIRI and FCI products
In addition to user requirements, geostationary ocean colour processing must meet operational constraints over the Earth disk coverage and must be capable of providing a stable product time-series.
Scientific development of these products and services has to account for a number of differences between ocean colour data acquired from polar and geostationary orbits.One of the most impacting factors is that geostationary observations provide Introduction

Conclusions References
Tables Figures

Back Close
Full Earth disk coverage in which spatial resolution and viewing geometries expand significantly towards the disk's edges, as displayed in Fig. 2. Figure 2a shows that viewing geometries exceed 60 • in large parts of Europe.This poses significant challenges for ocean radiometric retrievals, particularly for atmospheric correction and air-sea interface modelling (Ruddick et al., 2014).Furthermore, the fact that SEVIRI and FCI instruments are not designed to meet ocean colour requirements causes additional product limitations.An important part of the development is therefore the characterization of the limitations of operational ocean colour product quality.

Algorithmic and instrument limitations
For geostationary ocean colour products, atmospheric corrections exceed current processing specifications at high airmasses, typically beyond the factor of 4, and necessitate using spherical shell atmosphere models (Ding and Gordon, 1994).High sun zenith angles result in weak sunlight transmittance to the surface (Wang, 2006).High viewing zenith angles cause strong skylight reflection (Ruddick et al., 2014) and weak sea-air interface transmittance and, in turn, produce a weak water-leaving signal.Most ocean colour data processing algorithms are not designed to function for viewing zenith angle greater than 60 • , but such viewing angles become important for high latitude remote sensing from geostationary platforms -see Fig. 2. Slant geometries amplify uncertainties associated with Rayleigh-aerosol interactions and decoupling of atmospheric gas layers like ozone.Effects of bright targets such as land or clouds that are adjacent to the water surface are extended over larger distances.High wind speeds cause additional uncertainties, particularly for correction of skylight reflected at the airsea interface at high viewing zenith angles.SEVIRI and FCI characteristics put further constraints on ocean colour products.SEVIRI is hindered by its spectral resolution because the red and NIR bands only enable a restricted range of products that are mostly defined by water turbidity.FCI's additional bands in the blue and green will improve on this and may enable chlorophyllrelated products.Most ocean colour coastal and inland water applications require spa-Introduction

Conclusions References
Tables Figures

Back Close
Full tial resolution significantly higher than SEVIRI's, particularly in European waters.This is highlighted in Fig. 2b.SEVIRI signal-to-noise ratios (SNR) are low compared to the SNR requirements of sensors dedicated to ocean colour observations (Govaerts and Clerici, 2004;McClain and Meister, 2012).SEVIRI absolute calibration and characterization are also limited with biases estimated at −8, −6 and +3.5 % in bands at 635, 810 nm, and 1640 nm, respectively; with these estimates having uncertainties of 1 to 1.5 % (Meirink et al., 2013).An example of the impact of 1 % absolute uncertainty in band 0.6 µm on water-leaving reflectances is shown in Fig. 3. Due to these limitations, SEVIRI can only quantify strong marine optical signals, like high turbidity, and can observe only very high-biomass blooms.

Summary of the algorithmic approach
In SEVIRI ocean colour development, the algorithmic approach largely follows previous work by (Neukermans et al., 2009(Neukermans et al., , 2012) ) and focuses on operational processing capacity over the full SEVIRI disk coverage.The processing stages include top-of-theatmosphere re-calibration, dedicated atmospheric correction and application of basic in-water algorithms.The major modifications occur in the application of the spherical shell atmosphere for the modelling of Rayleigh molecular scattering, and in the automated extrapolation of aerosol properties derived for clear waters to adjacent turbid regions.The SWIR 1.640 nm band is also investigated for aerosol modelling in combination with a strategy to increase its SNR via temporal averaging.If the main aerosol approaches fail for a given pixel, the atmospheric correction resorts to using aerosol monthly climatology.The SEVIRI High-Resolution Visible (HRV) band is studied also for its capacity to increase the spatial resolution, again, in combination with temporal averaging to increase its SNR.
Major product limitations due to retrieval and instrument conditions are summarized in Table 4.The largest errors arise at high airmasses, in the sun-glint geometry and at high aerosol optical depth conditions.

OSD Introduction Conclusions References
Tables Figures

Back Close
Full

Conclusions
User requirement studies point to high interest in ocean colour products from geostationary platforms.There is a well-defined need to increase the spatial extent of coverage of aquatic processes and to capture diurnal cycles in coastal, estuarine and inland-water environments which are currently undersampled by polar orbiting sensors (Mouw et al., 2015).The ongoing effort to develop operational water turbidity products from EUMETSAT's SEVIRI instruments serves to meet these needs and to investigate geostationary capabilities in anticipation of future dedicated ocean colour missions.
The concept of geostationary ocean colour already has broad coverage in the literature with thorough descriptions of user needs, applications, and mission characteristics (Antoine et al., 2012;Albert et al., 2014;Ruddick et al., 2014;He et al., 2013).The first dedicated geostationary ocean colour mission is now in space, GOCI from Korea, while GOCI-II is in development.A number of other agencies have their respective missions in stages of preliminary analyses.Proof of concept water turbidity products have been successfully demonstrated from SEVIRI (Neukermans et al., 2012) and shown to significantly improve the coverage of aquatic processes and to enable tracking of high spatio-temporal events thanks to frequent observations during the day.
The current work extends the recommendations and the research and builds operational ocean colour capability from SEVIRI.The survey of user requirements suggests that despite many SEVIRI limitations, turbidity-related products will be useful for user reporting within the European MSFD and WFD, for monitoring of coastal water quality and quality of lakes in Africa, and for further applications including sediment transport, ecosystem modelling, offshore operations, and coccolithophore observations.The FCI instrument on MTG will improve on SEVIRI's capabilities and enable higher spatial resolution observations and additional products such as possible chlorophyll a quantification.
The goal of the SEVIRI's ocean colour processor is to provide NRT capabilities so that it can be extended to the operational SEVIRI processing chain.The goal is also Introduction

Conclusions References
Tables Figures

Back Close
Full to reprocess multi-year and multi-mission SEVIRI data time series to provide the longterm turbidity trends required by many applications.SEVIRI algorithm development follows the published research (Neukermans et al., 2012).Additionally, it extends the turbidity products over the Earth disk coverage and incorporates the product error budget due to modelling uncertainties at large solar and viewing zenith geometries and due to the actual spectral, spatial, and radiometric limitations of the sensor.User requirements have added new products to the development: water-leaving remote sensing reflectance, Secchi depth, horizontal visibility, and flagging of extreme high biomass algal blooms and of extreme cyanobacteria blooms, surface scums and vegetation.
The science algorithm extends the atmospheric correction by incorporating the spherical shell modelling of Rayleigh molecular scattering and automated procedures for aerosol modelling.The new products also require new algorithms, the development of which is in progress.Furthermore, the 1640 nm and HRV bands are investigated to improve atmospheric correction and spatial resolution of the measurements.Generation of Level-3 spatially and temporally binned SEVIRI products is anticipated to facilitate long-term turbidity trend analyses.Further opportunities lie in the combined use of SEVIRI with dedicated ocean colour polar observations.The current development stage of the SEVIRI water-turbidity study is supported by initial algorithm validations.Thorough product validation will complete the study.Validation will include generation and verification of intermediate internal products, as well as inter-comparisons with in situ measurements, simulated data (Nechad et al., 2015), and with products from ocean colour polar observing missions.Introduction

Conclusions References
Tables Figures

Back Close
Full  Full  Full Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | healthy and productive" (European Commission, 2008; Article 3).MSFD defines GES via 11 qualitative descriptors of which the most relevant to SEVIRI are three: eutrophication and associated high algal biomass; hydrographical conditions and the turbidity parameter related to the load of suspended particulate matter; and biodiversity.The eutrophication descriptor includes two indicators which are measurable by optical remote sensing, these are chlorophyll a concentrations over the algae growing season, and water transparency related to the increase in suspended algae (Sanden and Håkans-Discussion Paper | Discussion Paper | Discussion Paper | son Screen / Esc Printer-friendly Version Interactive Discussion Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |and nutrient load-up, where light is represented by scalar quantum Photosynthetically Available Radiation (PAR: 400-700 nm) and its vertical attenuation in the water column (KdPAR) or by an equivalent parameter, such as euphotic depth.SEVIRI can only support parameterization of the attenuation coefficient in turbid waters in which Non-Algal Particles (NAP) are the dominant factor determining algal bloom timing and duration.
Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper | Discussion Paper |

Figure 1 .Figure 2 .
Figure 1.Sequence of daily observations of Bay of Biscay by NASA's polar-orbiting MODIS sensor on the Aqua platform and by SEVIRI on MSG3.Development of coccolithophore blooms can be clearly followed on the SEVIRI daily-mean images.The figure is reproduced from (Vanhellemont et al., 2013).

Table 4 .
Summary of conditions contributing to SEVIRI ocean colour product uncertainties, where the largest errors arise at high airmasses, in the sun-glint geometry and at high aerosol optical depths.