OS4.7

Marine pollution has been traditionally addressed in Earth Observation studies through the use of Synthetic Aperture Radar (SAR) imagery. In operational processes, the contrast between the dark oil surfaces, characterized by a low backscatter return, and the rough, bright sea surface with higher backscatter has been exploited for decades.

Many of the processing techniques involve the use of semi-automatic workflows. Dark spot segmentation and feature classification are, undisputedly, common computational tasks. However, effective discrimination between oil slicks and other ocean phenomena (e.g. biogenic slicks, wind streaks, greasy ice) remains a challenge. To complete this task, a trained human operator is often employed in the final validation step. Thus, the process is time and resource consuming over large expanses, while the results are highly subjective. Automating this process and reducing computation and analysis time is the ultimate goal.

New algorithms based on the use of artificial intelligence for oil slick detection have recently emerged. While there are studies proving their effectiveness in successfully segmenting and classifying oil slicks, questions regarding their operational feasibility remain. Do they improve the quality of the detection? Are they more capable of discriminating between the various dark formations? What are the computational and data resources required for training, validation, and deployment of such an algorithm?

This project focusses on the development of a new customized algorithm for natural oil slicks detection. As part of this development, we analyzed the state-of-the-art methods and performed a comparison of the latest deep learning methods and classic semi-automatic techniques. Here, we present an in-depth analysis of selected segmentation and convolutional neural networks algorithms and various frameworks. The primary objective is to evaluate their effectiveness and expose their deficiencies for the detection and classification of natural oil slicks against anthropogenic pollution and other dark formation caused by ‘look-alikes’.

This presentation centers on the results that have been obtained by utilizing high-resolution open SAR data acquired by the Copernicus Sentinel-1 satellites. The evaluation is based on study sites located in the Black Sea, where two known oil seepage areas are actively generating consistent productive slicks as well as underdeveloped oil traces. 

How to cite: Vrinceanu, C., Grebby, S., and Marsh, S.: Is Deep Learning more effective in detecting natural oil slicks? A comparison of semi-automated and AI techniques, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-5462, https://doi.org/10.5194/egusphere-egu21-5462, 2021.

Abstract: The thickness of oil spills on the sea is an important but poorly studied topic. Means to measure slick thickness are reviewed. More than 30 concepts are summarized. Many of these are judged not to be viable for a variety of scientific reasons. Two means are currently available to remotely measure oil thickness, namely, passive microwave radiometry and time of acoustic travel. Microwave radiometry is commercially developed at this time. Visual means to ascertain oil thickness are restricted by physics to thicknesses smaller than those of rainbow sheens (~3 µm), which rarely occur on large spills, and thin sheen. One can observe that some slicks are not sheen and are probably thicker. These three thickness regimes are not useful to oil spill countermeasures, as most of the oil is contained in the thick portion of a slick, the thickness of which is unknown and ranges over several orders of magnitude. There is a continuing need to measure the thickness of oil spills. This need continues to increase with time, and further research effort is needed. Several viable concepts have been developed but require further work and verification. One of the difficulties is that ground truthing and verification methods are generally not available for most thickness measurement methods.

How to cite: Fingas, M.: Remotely Measuring Oil Slick Thickness: An Epic Challenge, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-978, https://doi.org/10.5194/egusphere-egu21-978, 2021.

A layer of oil on the sea surface reduces the components of ocean wave spectra corresponding to the capillary and gravity-capillary waves leading to significantly reduced radar backscatter and making slicks appear dark in synthetic aperture radar (SAR) images.  The ratio of the backscatter between clean and slicked ocean surfaces is known as the ‘damping ratio,’ and has been shown to be sensitive to variations within slicks that correlate with the oil layer’s thickness and fractional water volume.  Although the relative thickness relationship is well established and can be used to identify areas of ‘thicker’ oil within a slick, quantifying the thickness from SAR alone remains to be shown.  There is considerable uncertainty regarding the potential capability of SAR to quantitatively determine the oil layer thickness for slicks in open water given the dependence on bulk and interfacial oil layer properties and the variation of the properties typical in this setting and for different types of oil.  Here, we report the results of a study modeling radar backscatter from slicked and unslicked ocean surfaces based on electromagnetic scattering theory and accounting for oil properties and meteorological conditions.  The electromagnetic scattering model is used to evaluate whether the oil thickness can be quantified with reasonable accuracy based upon SAR backscatter intensities alone, and how information about metocean conditions, oil properties and ocean temperature and salinity can be used to calibrate the model to obtain more accurate thickness estimates.

The research was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration.

How to cite: Jaruwatanadilok, S., Duan, X., Holt, B., and Jones, C.: Realistic Electromagnetic Modeling of SAR’s Capability for Oil Spill Thickness Measurement, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-9209, https://doi.org/10.5194/egusphere-egu21-9209, 2021.

We describe an effort to develop a quantifiable approach for determining the thicker components of oil spills using microwave synthetic aperture radar (SAR) imagery that can be utilized in an operational context to guide clean-up efforts. The presence of mineral oil on the surface can suppress the SAR returns in two ways. First, surface oil dampens the capillary waves making those areas darker in SAR imagery, an effect that been used to determine oil extent. The second is by modifying the dielectric properties of the surface from those of clean seawater to either pure oil or a mixture of oil and water as the oil weathers and thickens to form an emulsion. The emulsion provides an intermediate conductive surface layer between the highly conductive ocean itself and the very low, ‘radar transparent’ sheen layers, resulting in a further reduction in the radar returns for areas with thicker oil within an inhomogeneous oil slick. The challenges are to quantify the thickness and conditions for which this thicker layer becomes separable from the thinner oil, determine whether multiple thicker components can be identified, identify which airborne and spaceborne SAR systems can be used for this purpose, and determine under what range of environmental conditions, particularly wind speed, it is possible.

We are planning to hold field campaigns with in situ measurements and SAR and multispectral remote sensor data collections from drones, aircraft, and satellites. The field measurements include surface collections of oil, underwater spectrophotometry, and drone-based infrared, ultraviolet, and optical collections.  Coincident with the field measurements, the airborne L-band NASA-UAVSAR SAR system will image the seep fields to track temporal changes and overpassing satellite acquisitions will be acquired. UAVSAR provides fine resolution, low noise radar imagery under all weather and solar conditions and is fully polarimetric, which enables evaluation of multiple methods to characterize the oil slick. The system noise floor of this instrument, considerably less than all satellite SAR instruments, enables a detailed examination of the zones of reduced backscatter caused by varying oil thickness levels. The primary satellite SAR will be C-band Sentinel-1, accompanied potentially by C-band Radarsat-2 and L-band ALOS-2. Both the UAVSAR and satellite SAR analysis will utilize the contrast ratio, defined as the normalized radar cross section (NRCS) in open water divided by the NRCS in oil-covered water. The larger the ratio, the thicker the oil. The operational algorithm for oil thickness is under development using satellite SAR data and will be staged in NOAA’s SAR Ocean Product System (SAROPS) that currently produces SAR-derived wind speed and oil spill extent operationally, with the latter using the Texture-Classifying Neural Network (TCNNA) to automatically delineate oil versus non-oil covered areas. We are planning field campaigns at the natural oil seep area offshore of Santa Barbara, California, in March 2021 and during the 2022 Norwegian Clean Sea Association for Operating Companies’ (NOFO’s) coordinated releases of oil in the North Sea. Recent field collections and analysis will be shown, as available.

How to cite: Holt, B., Monaldo, F., Jones, C., and Garcia, O.: Transitioning SAR-derived Oil Spill Thickness Measurements into an Operational Context, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-1858, https://doi.org/10.5194/egusphere-egu21-1858, 2021.

This presentation discusses a downstream application from Copernicus Services, developed in the framework of the IMPRESSIVE project, for the monitoring of  the oil spill produced after the crash of the ferry “Volcan de Tamasite” in waters of the Canary Islands on the 21^st of April 2017. The presentation summarizes the findings of [1] that describe a complete monitoring of the diesel fuel spill, well-documented by port authorities. Complementary information supplied by different sources enhances the description of the event. We discuss the performance of very high resolution hydrodynamic models in the area of the Port of Gran Canaria and their ability for describing the evolution of this event. Dynamical systems ideas support the comparison of different models performance. Very high resolution remote sensing products and in situ observation validate the description.

Authors acknowledge support from IMPRESSIVE a project funded by the European Union’s Horizon 2020 research and innovation programme under grant agreement No 821922. SW acknowledges the support of ONR Grant No. N00014-01-1-0769

References

[1] G.García-Sánchez, A. M. Mancho, A. G. Ramos, J. Coca, B. Pérez-Gómez, E. Álvarez-Fanjul, M. G. Sotillo, M. García-León, V. J. García-Garrido, S. Wiggins. Very High Resolution Tools for the Monitoring and Assessment of Environmental Hazards in Coastal Areas.  Front. Mar. Sci. (2021) doi: 10.3389/fmars.2020.605804.

How to cite: Mancho, A. M., García-Sánchez, G., Ramos, A. G., Coca, J., Pérez-Gómez, B., Álvarez-Fanjul, E., Sotillo, M. G., García-León, M., García-Garrido, V. J., and Wiggins, S.: Monitoring and assessment of an oil spill event with very high resolution tools., EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-13528, https://doi.org/10.5194/egusphere-egu21-13528, 2021.

Understanding the pathways of floating material (e.g. larvae, plastics, oil) at the surface ocean is important to improve our knowledge on the surface circulation and for its ecological and environmental impacts.  For example, knowing where floating plastic and oil spills accumulate in the surface ocean can help ocean clean-up strategies.  One of the main methods of research is virtual particle simulations, which simulate the dispersion of floating material in the Ocean.  

Previous studies have tried to understand the surface dispersion and accumulation via these numerical simulations. To define the circulation, the velocity outputs of ocean general circulation models are needed. Oceanic models have improved in the past years, but many still do not fully represent the ocean dynamics at the fine-scales (below 100 km).  The spatial resolution of ocean models and whether they include a tidal-forcing are two important model parameterizations that can determine how well the ocean dynamics are represented at the fine-scales. In this study we try to answer: How do these model characteristics affect the dispersion and accumulation of virtual particles at the ocean surface?

To answer this, we use the ocean surface velocity outputs of different NEMO simulations to simulate the trajectories of virtual particles, and we evaluate the impact of different NEMO simulations’ spatial resolution and the presence or not of a tidal-forcing. As tidal-forcing has a big impact on the ocean model’s representation of internal tides and waves, we focus on a region where there is a high internal-tide signal: the Azores Islands.  We evaluate these impacts by looking at whether there is a difference in particles’ accumulation and dispersion in the different model scenarios.

How to cite: Gomez-Navarro, L., van Sebille, E., Albert, A., Molines, J.-M., Brodeau, L., Le Sommer, J., and Ubelmann, C.: The effect of ocean model tidal-forcing and spatial resolution on virtual particle dispersion and accumulation at the ocean surface, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-853, https://doi.org/10.5194/egusphere-egu21-853, 2021.

Being situated in a semi-enclosed Mediterranean lagoon, the Port of Taranto represents a transport, industrial and commercial hub, where the port infrastructure, a notorious steel plant, oil refinery and naval shipyards coexist with highly-dense urban zone, recreation facilities, mussel farms, and vulnerable environmental sites. A Single Buoy Mooring in the center of the Mar Grande used by tankers and subsea pipeline that takes oil directly from tanker to refinery are assumed to stay at risk of accidental oil spills, despite significant progress in technology and prevention.

The oil spill model MEDSLIK-II (http://medslik-ii.org) coupled to the high resolution Southern Adriatic Northern Ionian coastal Forecasting System (SANIFS http://sanifs.cmcc.it Federico et al., 2017) is used to model hypothetical oil spill scenarios in stochastic mode. 15,000+ hypothetical individual spills are generated from randomly selected start locations: 50% from a buoy and 50% along the subsea pipeline 2018–2020. Individual spill scenario is based on a real crude oil spill caused by a catastrophic pipeline failure happened in Genoa in April 2016 (Vairo et al., 2017). The model outputs are processed statistically to represent quantitively: (1) timing of the oil drift; (2) hazard maps in probability terms at the sea surface and on the coastline; (3) oil mass balance; (4) local-zone contamination assessment.

The simulations reveal that around 48% of the spilled oil will evaporate during the first 8 hours after the accident. Being transported by highly variable currents and waves, the rest is additionally exposed to multiply reflections from sea walls and concrete wharfs that dominate in the study area. As a result, the oil will be dispersed almost isotropically in the Mar Grande, indicating a rather moderate or small level of concentrations over the minimum threshold values (French McCay, 2016).

We have concluded that at a probability of 50%, the first oil beaching event will happen within 14 hours after the accident. The most contaminated areas are predicted on and around the nearest Port berths, on the coastlines of the urban area and on the tips of the breakwaters that frame the Mar Grande openings. The remote areas of the West Port and Mar Piccolo are expected to be the least contaminated ones.

Results are applicable to contingency planning, ecological risk assessment, cost-benefit analysis, and education.

This work is conducted in the framework of the IMPRESSIVE project (#821922) co-funded by the European Commission under the H2020 Programme.

References

Federico, I., Pinardi, N., Coppini, G., Oddo, P., Lecci, R., Mossa, M., 2017. Coastal ocean forecasting with an unstructured grid model in the southern Adriatic and northern Ionian seas. Nat. Hazards Earth Syst. Sci., 17, 45–59, doi: 10.5194/nhess-17-45-2017.

French McCay, D., 2016. Potential effects thresholds for oil spill risk assessments. Proc. of the 39 AMOP Tech. Sem., Environment and Climate Change Canada, Ottawa, ON, 285–303.

Vairo, T., Magrì, S., Qualgliati, M., Reverberi, A.P., Fabiano, B., 2017. An oil pipeline catastrophic failure: accident scenario modelling and emergency response development. Chem. Eng. Trans., 57, 373–378, doi: 10.3303/CET1757063.

How to cite: Liubartseva, S., Federico, I., Coppini, G., and Lecci, R.: Oil spill risk assessment for a Single Buoy Mooring terminal in the Port of Taranto (Southern Italy), EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-4662, https://doi.org/10.5194/egusphere-egu21-4662, 2021.

This interdisciplinary study with implications for fate and transport of pollutants from shipping, investigates the previously overlooked phenomenon of ship induced mixing. When a ship moves through water, the hull and propeller induce a long-lasting turbulent wake. Natural waters are usually stratified, and the stratification influences both the vertical and horizontal extent of the wake. The altered turbulent regime in shipping lanes governs the distribution of discharged pollutants, e.g. PAHs, metals, nutrients and non-indigenous species. The ship related pollutant load follows the trend in volumes of maritime trade, which has almost tripled since the 1980s. In heavily trafficked areas there may be one ship passage every ten minutes; today shipping constitutes a significant source of pollution.

To understand the environmental impact of shipping related pollutants, it is essential to know their fate following regional scale transport. However, previous modelling efforts assuming discharge at the surface will not adequately reflect the input values in the regional models. Therefore, it is urgent to bridge the gaps between the spatiotemporal scales from high-resolution numerical modeling of the flow hydrodynamics around the ship, mixing processes and interaction of the ship and wake with stratification, and parameterization in regional oceanographic modeling. Here this knowledge gap is addressed by combining an array of methods; in situ measurements, remote sensing and numerical flow modeling.

A bottom-mounted Acoustic Doppler Current Profiler was placed under a ship lane, for in-situ measurements of the vertical and temporal expansion of turbulent wakes. In addition, ex-situ measurements with Landsat 8 Thermal Infrared Sensor were used to estimate the longevity and spatial extent of the thermal signal from ship wakes. The computational modelling was conducted using well resolved 3D RANS modelling for the hull and the near wake (up to five ship lengths aft), a method typically used for the near wake behaviour in analysing the propulsion system. As this is not feasible to use for a far wake analysis, the predicted wake is then used as input for a 2D+time modelling for the sustained wake up to 30min after the ship passage. These results, both from measurements and numerical models, are then combined to analyse how ship-induced turbulence influence at what depth discharged pollutants will be found.

This first step to cover the mesoscales of the turbulent ship wake is necessary to assess the impact of ship related pollution. In-situ measurements show median wake depth 13.5m (max 31.5m) and median longevity 10min (max 29min). Satellite data show median thermal wake signal 13.7km (max 62.5km). A detailed simulation model will only be possible to use for the first few 100m of the ship wake, but the coupling to a simplified 2D+time modelling shows a promising potential to bridge our understanding of the impact of the ship wake on the larger scales. Our model results indicate that the natural stratification affects the distribution and retention of pollutants in the wake region. The depth of discharge and the wake turbulence characteristics will in turn affect the fate and transport of pollutants on larger spatiotemporal scales.

How to cite: Nylund, A. T., Bensow, R., Liefvendahl, M., Eslamdoost, A., Tengberg, A., Mallast, U., Hassellöv, I.-M., Broström, G., and Arneborg, L.: The importance of the turbulent ship wake regime for pollutant fate and transport, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-11011, https://doi.org/10.5194/egusphere-egu21-11011, 2021.

Near-surface Stokes drift in operational wave forecast

We apply an operational model for the mean drift of small particles near and at the surface. The drift is a sum of the Stokes drift and the current. The Stokes drift is derived from a wave model and the current is calculated by means of a 3D hydrodynamic model.

In an operational wave model setup (based on WAVEWATCH-III version 7) a parametric tail is supplied as an extension to the prognostic wave spectrum. The parametrisation is based on the modelled spectra level and the first circular moment (the spectral spread) near the highest prognostic frequency, which is typically around 0.5 Hz in operational wave modelling.

New source terms formulations has been introcuced in wave modelling (e.g. WAVEWATCH-III with effect from 2019) that reproduces the spreading and spectral level well compared to many independent observations.

The Stokes drift is calculated at a discrete number of depths down to Z = 2/Ks, where Ks is a wave number scale estimated from the prognostic spectrum. The calculation is integrated in the wave model output.

The application is evaluated in complex weather situations between January 1 and January 6 2020 in the European North West Shelf region. A set of wave model setups are compared with a variation of values of the prognostic spectral cut-off.

It is demonstrated that a the near-surface drift profile is resolved well with an order of ten discrete depths, and the parametric tail extension requires a low computational time.

How to cite: Hansen, C.: Near-surface Stokes drift in operational wave forecast, EGU General Assembly 2021, online, 19–30 Apr 2021, EGU21-9520, https://doi.org/10.5194/egusphere-egu21-9520, 2021.