Copernicus Sentinel-3A OLCI has opened an era of European operational ocean colour observations and represents a significant achievement in the maturity, accessibility, continuity and routine utilization of ocean colour measurements. A consistent time series of OLCI-A products is now available, spanning three years of operation, and the OLCI-B data stream has recently joined this record.

The presentation will describe the achievements in the ocean colour realm brought by the Sentinel-3 mission. OLCI’s special capabilities for advanced ocean colour products and data services will be shown, relying on examples from a number of algorithm development and application activities.

The presentation will also describe the quality status of OLCI ocean colour products based on validations with in situ measurements and time series inter-comparisons with contemporaneous missions and climatologies. Guidance on the use of the products will also be provided, including product limitations and the application of flags to mask cloudy or unreliable pixels.

Main validation outcomes from the Sentinel-3 Validation Team, ocean colour sub-group (S3VT-OC), will be presented based on the recent S3VT meeting contributions. S3VT-OC provides an independent evidence about the quality of products from the OLCI instruments for open oceans, regional seas, coastal zones and inland waters using separate validation measurements and techniques.

Ocean colour product improvements and evolutions are a part of on-going OLCI activities, led by EUMETSAT and the S3 MPC ESA contract. Several product evolutions are currently taking place for near-term implementation. Also, longer-term studies are in progress to fully utilize the capabilities of the OLCI instruments and to capture the advancements in ocean colour science. The goal is to ensure OLCI ocean colour observations are state-of-the-art and meet the mission requirements and the needs of Copernicus marine data users.

Sentinel-2/MSI (S2-MSI) are mainly designed to provide information on land surfaces for applications in agriculture, geology, forestry, mapping, global change research, etc. However, its performance in terms of signal-to-noise ratio (SNR) is sufficient to be used for marine applications, especially in turbid coastal waters. Compared to the mainstream ocean colour sensors (MODIS-AQUA, VIIRS and Sentinel-3/OLCI) this sensor offers great advantages in terms of spatial resolution enabling the development of the new generation of coastal water quality products: i.e. high resolution total suspended matter (TSM) and chlorophyll a (CHL-a) products.

The presence of clouds limits the usability of Sentinel-2 data, as happens with all optical satellite sensors. A more specific problem related to the high resolution of Sentinel-2 data is the presence of spatially resolved cloud shadows, obscuring partially the signal being measured. These cloud shadows appear as border features surrounding clouds, but also as detached features, not associated to pixels identified as clouds. This is the case of shadows resulting from small, scattered clouds like cumulus-type clouds. As cloud shadows do not show specific spectral features it is very difficult to accurately detect and flag them to exclude them from further processing.

A technique to detect and remove cloud shadows from high resolution satellite imagery will be presented. The approach is based on DINEOF (Data Interpolating Empirical Orthogonal Functions) and provides a pixel-by-pixel outlier index based on the departure of each data point to a truncated EOF basis provided by DINEOF. The more a pixel departs from the value expected from the truncated EOF basis, the higher it is penalized. Additional tests, penalizing proximity to clouds and low concentration values, are used in combination with the EOF-based index as supporting detection test to increase the accuracy of the technique.

This approach proves efficient for very small shadows, such as those resulting from offshore wind farm turbines, and medium-sized shadows, such as those from cumulus clouds. Larger shadows, from stratus-type clouds, cover large domains and cannot be considered as outliers by the EOF index. In these cases, a multi-scale detection approach will be investigated to effectively remove them.

Several examples on the detection of cloud shadows will be presented. The suggested approach does not depend on the satellite sensor used, and can also be used on land data.

The recent successful launch of Sentinel-3B allows for an unprecedented monitoring potential of the Global Ocean. The Polymer atmospheric correction algorithm for Ocean Colour uses the full available spectrum to discriminate the water reflectance from the atmospheric and surface signal, using an iterative optimization scheme and relying on analytical models. With its ability to process both oceanic and coastal waters, its robustness to the aerosol contamination, partial ice, cloud and haze contamination, adjacency effects and the ability to process the sun glint contaminated observations, this algorithm maximizes the usefulness of these observations. Polymer is being applied to OLCI data from both Sentinel-3A and Sentinel-3B to show the extended daily coverage of their combined observations. The consistency of the observations from OLCI-A and OLCI-B with Polymer will be evaluated. In particular, we use a methodology which evaluates the self-consistency of the water reflectance products by using the multiple daily revisits of the Arctic Ocean in summer. The standard OLCI product will also be evaluated with this method. Validation using in-situ matchups with Aeronet-OC will be presented to better assess the quality of the respective products.

The 2nd atmospheric correction intercomparison exercise (ACIX-II) is an ESA/NASA collaborative effort under CEOS Cal/Val framework evaluating various atmospheric correction (AC) algorithms applied to Sentinel-2 and Landsat images. The ACIX-II aquatic subgroup with two different teams involved, i.e., processing and data contribution teams, provides performance analyses of eight different atmospheric correction processors using radiometric data provided by the data contribution teams and those available via AERONET-OC. The in situ data are obtained in various field campaigns in lakes, reservoirs, and rivers across the Europe, Asia, and Americas. The objective is a) to evaluate the AC methods under various atmospheric, aquatic, environmental, and morphological/topographical conditions and b) to make recommendations on possible blending approaches to achieve highly accurate Sentinel-2-/Landsat-derived aquatic reflectance products in inland and nearshore waters. This presentation will report on the latest analyses associated with this intercomparison exercise.

The use of decametre scale imagery for inland and coastal water applications has exploded in the last couple of years, thanks to the free availability of high quality data from Landsat 8 (2013-present), Sentinel-2A (2015-present) and Sentinel-2B (2017-present). Together, these sensors provide unprecedented coverage of all land masses and coastal areas, and are the first public missions suitable for routine mapping of inland waters, including small lakes and narrow rivers.

Due to the near-nadir viewing of these sensors originally constructed for land applications, direct sun glint on water surfaces often contaminates their imagery. Due to the relatively high spatial resolution of the sensors, and the high degree of small-scale variability of the surface effects (caused by local wind patterns or currents) traditional glint correction schemes - e.g. using view geometry, wind and Cox & Munk (1954) wave slope statistics - are not applicable. Glint correction for these sensors has emerged, either by modelling realistic water reflectances as part of the atmospheric correction (POLYMER by Steinmetz et al., 2011, and C2RCC developed by Brockmann Consult), or by using the SWIR bands to directly estimate glint from the imagery (Harmel et al., 2018, Vanhellemont, submitted).

Additionally, adjacency effects, i.e. light scattered from adjacent non-water targets such as land, clouds, and ice can have varying impacts on the observed water signal. There have been some attempts to directly estimate the adjacency effects (Sterckx et al., 2011, 2015, De Keukelaere et al., 2018) using the similarity spectrum of Ruddick et al. (2006). This method may be inaccurate depending on the bands used and the water constituents, especially Red-NIR for Landsat, but also the NIR in general for extremely turbid waters. Approaches that have an explicit modelling of the water reflectance (see earlier) may deal well with adjacency effects if the target falls within the validity range of the model. Studies by Bulgarelli et al. (2014, 2018) show the large length scales at which adjacency effects manifest themselves, as well as a dependency on position, illumination and view geometry and the atmospheric optical depth. For inland water bodies of smaller dimensions than the typical adjacency length scale, the position effects may be of lesser importance, but in general the adjacency effects are more severe due to the proximity to land.

If glint and adjacency effects are both present and estimated separately their impacts may be exaggerated. Additionally, their effects may be amplified through the atmospheric correction. In the present study, the Dark Spectrum Fitting (DSF) atmospheric correction in ACOLITE is used to estimate atmospheric path reflectance, which is relatively insensitive of adjacency and glint (Vanhellemont & Ruddick, 2018; Vanhellemont, submitted). The reflectances retrieved after the DSF atmospheric correction however still contain glint and adjacency effects. These effects may in some cases cause the retrieved signal over water pixels to be larger in the NIR and SWIR bands than in the visible bands. By assuming zero water reflectance in the SWIR channels (at 1.6 and 2.2 µm), the relative contribution of glint and adjacency effects can be estimated simultaneously. The problem of glint and adjacency is illustrated using multiple Belgian inland waters, and different correction methods are evaluated qualitatively using spatial and temporal analysis of the imagery and quantitatively using in situ hyperspectral matchups.


References

Bulgarelli, B., Kiselev, V. and Zibordi, G., 2014. Simulation and analysis of adjacency effects in coastal waters: a case study. Applied optics, 53(8), pp.1523-1545.

Bulgarelli, B., Zibordi, G. and Mélin, F., 2018. On the minimization of adjacency effects in SeaWiFS primary data products from coastal areas. Optics express, 26(18), pp.A709-A728.

Cox, C. and Munk, W., 1954. Measurement of the roughness of the sea surface from photographs of the sun’s glitter. Josa, 44(11), pp.838-850.

De Keukelaere, L., Sterckx, S., Adriaensen, S., Knaeps, E., Reusen, I., Giardino, C., Bresciani, M., Hunter, P., Neil, C., Van der Zande, D. and Vaiciute, D., 2018. Atmospheric correction of Landsat-8/OLI and Sentinel-2/MSI data using iCOR algorithm: validation for coastal and inland waters. European Journal of Remote Sensing, 51(1), pp.525-542.

Harmel, T., Chami, M., Tormos, T., Reynaud, N. and Danis, P.A., 2018. Sunglint correction of the Multi-Spectral Instrument (MSI)-SENTINEL-2 imagery over inland and sea waters from SWIR bands. Remote Sensing of Environment, 204, pp.308-321.

Ruddick, K.G., De Cauwer, V., Park, Y.J. and Moore, G., 2006. Seaborne measurements of near infrared water‐leaving reflectance: The similarity spectrum for turbid waters. Limnology and Oceanography, 51(2), pp.1167-1179.

Steinmetz, F., Deschamps, P.Y. and Ramon, D., 2011. Atmospheric correction in presence of sun glint: application to MERIS. Optics express, 19(10), pp.9783-9800.

Sterckx, S., Knaeps, E. and Ruddick, K., 2011. Detection and correction of adjacency effects in hyperspectral airborne data of coastal and inland waters: the use of the near infrared similarity spectrum. International Journal of remote sensing, 32(21), pp.6479-6505.

Sterckx, S., Knaeps, S., Kratzer, S. and Ruddick, K., 2015. SIMilarity Environment Correction (SIMEC) applied to MERIS data over inland and coastal waters. Remote Sensing of Environment, 157, pp.96-110.

Vanhellemont, Q., submitted. Adaptation of the dark spectrum fitting atmospheric correction for aquatic applications of the Landsat and Sentinel-2 archives

Vanhellemont, Q. and Ruddick, K., 2018. Atmospheric correction of metre-scale optical satellite data for inland and coastal water applications. Remote Sensing of Environment, 216, pp.586-597.

The Earth's climate is changing; however, the time scale, consequences and rate of this change is subject to uncertainty. Significant efforts are already done to mitigate this uncertainty. Remote and continuous observation of the Earth by satellites provides cost-effective acquisition of global data, which can be fed into internationally agreed key datasets; such as the Essential Climate Variables identified by the Global Climate Observing System (GCOS) of the United Nations.

Commonly known that the radiometric signal is contaminated by atmospheric influence as the light travels from the earth surface to the satellite in outer space; and small errors in instrument calibration or atmospheric correction can introduce significant errors in the inferred ocean/land surface signal. Thus stable calibration and high radiometric accuracy are critical in providing meaningful and indisputable data for climate related research and applications. Moreover, creating a long time series of data requires that the data from different sensors be stitched together in a seamless manner to guarantee and/or quantify the consistency of their geophysical measurements.

In order to maintain the homogeneity and consistency of time-series observations, the GCOS-Climate Monitoring Principles recommended the introduction of an initial tandem overlap phase between new and already in-orbit satellite instruments. Following the aforementioned recommendations, the Copernicus Services, the European Space Agency (ESA) and EUMETSAT have introduced the Sentinel-3-Tandem phase to allow climate scientists to quantify and understand the small differences between each satellite, as these influence our ability to determine accurate climate trends. The Sentinel-3A and Sentinel-3B tandem flight at 30 seconds will minimise the differences between both S3A & S3B measurement to instrumental aspects only. Such as a homogenisation of S3A and S3B measurement would provide valuable time-series to the Earth observation community.

As part of the Sentinel-3 Mission Performance Centre (S3MPC) activities, ARGANS is tasked to assess the quality of the data product at both level-1 & level-2, to monitor both sensors evolution, and to ensure that the products meet the mission requirement accuracy. Cross-missions intercomparison is systematically performed (eg. with MODIS, VIIRS and MSI) over different test sites and using different methods and tools.

This presentation describes the radiometric cross-calibration methods and will address (i) the verification of OLCI Level-1B products from both S3A and S3B, (ii) the assessment of OLCI-L2 water products, and (iii) Sensor to sensor intercomparison with MODIS and VIIRS products.

The cross-comparison results of level-1 radiometry from OLCI-A vs OLCI-B show a discrepancy of about ~2%. Considering this difference is mainly related to instrumental aspects, we aim to analyse and understand its impact on Level-2 products. While the agreement of the level-1 products (TOA-reflectance) with the other sensors is better than 5%, it is not an easy task to understand and quantify the propagation of these differences to level-2 products. Moreover, how these aforementioned differences are scaled with respect to other optical sensors.