How valid are the Products and Services delivered by EODataBee?

As part of the DCS4COP project, a whole Work Package is dedicated to quantify and report on the validity of the Products and Services that will be delivered to Users through EODataBee. The purpose is to document the uncertainty associated to each optical or biogeochemical parameter (e.g., water turbidity or chlorophyll-a concentration) that can be accessed and used for water quality monitoring at various spatial and temporal scales. The second objective is to assess the quality of services delivered and the satisfaction of users.

The Validation activities aim at:

  • Issuing detailed and rigorous plans for the validation of the products and services delivered to Intermediate Business Users (IBUs)
  • Validating products (implemented as DataLayers in the EODataBee tool) with scientifically sound methods and adequate independent data
  • Validating the service delivered to IBUs on the basis of adequate and comprehensive (KPIs).

The Datalayers implemented in EODataBee are key optical and biogeochemical parameters derived from quality-checked in situ measurements, satellite observations at high temporal (up to 1 hour) and high (20 m), medium (300 m) and reduced (1000 m) spatial resolutions but also model outputs over several use cases representative of various coastal waters. These key parameters notably include water reflectance, sea surface temperature, water turbidity, concentrations of suspended particulate matter and chlorophyll-a, diffuse attenuation coefficients and euphotic depth.

Following methods developed and tested during the HIGHROC FP7 project (HIGH spatial and temporal Resolution Ocean Colour products and services), the validity of satellite and model derived parameters is systematically assessed for each use case based on numerous quality-checked match-ups with field measurements. These field measurements mainly come from autonomous sensors operated on fixed (e.g., buoys, pontoons) or moving (oceanographic research vessels, ferries) platforms. Coincident satellite observations (or model outputs) and field measurements provide regular direct comparisons between parameters implemented in EODataBee and ‘ground-truth’ at various spatio-temporal scales. These comparisons can be easily and rapidly extracted by IBUs to generate scatterplots and temporal trends but also detailed statistics representing the mathematical differences from which are computed the uncertainties associated to DataLayers.

The validation datasets consist of ‘historical’ match-ups notably generated as part of the HIGHROC project regularly completed by new ones performed as part of DCS4COP and are directly available to IBUs in EODataBee as an operational service.

In addition to uncertainties associated to each DataLayer, IBUs regularly evaluate the quality and performance of the services provided by EOdataBee by marking eleven KPIs, information reported automatically to the DCS4COP service operations.

Ongoing validation activities

Validation activities are ongoing in several European coastal areas such as the southern North Sea, UK and Norwegian waters, French estuaries and lagoons.

As examples the Gironde Estuary (south-West France) is characterized by highly turbid waters and a well-developed maximum turbidity zone. Several autonomous stations operating along the estuary provide numerous match-ups with satellite data to quantify the uncertainties associated to two key satellite-derived products: water turbidity and concentration of suspended particulate matter. In the Berre coastal lagoon (south-east of France), highly impacted by human activities, a water-quality monitoring program provides monthly match-ups with satellite observations used to map the water reflectance (Figure 1) and derived products (concentration of phytoplankton and water turbidity).

Figure 1.

A) Sentinel2-MSI image of the Berre lagoon on 04/12/2018 with location of the match-ups between in situ and S2-MSI and S3-OLCI satellite data on 04/12/2018 and 05/12/2018.

B) Water reflectance (Rrs) match-up between in-situ and S3-OLCI on 04/12/2018. The satellite Rrs was derived using 3 different atmospheric corrections for validation. 

C) Rrs matchup between in-situ and S3-OLCI on 05/12/2018. The satellite Rrs was derived using 3 different atmospheric corrections for validation.

 

The Norwegian ship of opportunity FerryBox network runs routes through a variety of water bodies that extend from Germany in the south, to Svalbard in north, and Iceland in west. The new national FerryBox infrastructure program NorSOOP will develop and extend the present network by using new instruments and including new lines that travel to the Arctic and Antarctic. 

The FerryBox system takes punctual water samples and measures continuously core parameters such as salinity, temperature, oxygen, turbidity, cDOM and chlorophyll a fluorescence, as well as wind speed and direction, and hyperspectral remote sensing reflectance (Rrs) using TrioS Ramses sensors.

These instrumented ferries provide valuable in situ environmental measurements that can be used to validate water property retrieval algorithms and atmospheric correction algorithms.

Figures 2 and 3 show an example of atmospheric corrections comparison using matchups between Sentinel-2 MultiSpectral Instrument (MSI) and FerryBox in-situ measurement from the MS Trollfjord traveling over Trondheim region in the west coast of Norway.  

Figure 2 shows the positions of ferry measurements after quality controls selection on the Sentinel-2 MSI RGB image from the 10th of July 2018. In situ Rrs spectra calculated from RAMSES TriOS radiometers (Fig. 3-a) are used to compare performance over the area of 2 different atmospheric corrections algorithms; Acolite (Fig. 3-b) and C2RCC (Fig. 3-c), applied on the same satellite image. The colors of fig.3-a-b-c correspond to the color of the spatial positions (Fig. 2).

The NorSOOP ferries network covers a large diversity of water types and has provided crucial insight into the dynamics of environmental conditions in these regions. This network is also highly valuable for the validation of atmospheric correction and water property retrieval algorithms satellite data.

 

Figure 2

 

Figure 3

 

Authors: Laboratoire d’Océanographie de Villefranche (LOV) and the French National Science Centre (CNRS), and Norwegian Institute for Water Research (NIVA)