A novel deep learning method for marine oil spill detection from satellite synthetic aperture radar imagery.

Oil spill discharges from operational maritime activities like ships, oil rigs and other structures, leaking pipelines, as well as natural hydrocarbon seepage pose serious threats to marine ecosystems and fisheries. Satellite synthetic aperture radar (SAR) is a unique microwave instrument for marine oil spill monitoring, as it is not dependent on weather or sunlight conditions. Existing SAR oil spill detection approaches are limited by algorithm complexity, imbalanced data sets, uncertainties in selecting optimal features, and relatively slow detection speed. To overcome these restrictions, a fast and effective SAR oil spill detection method is presented, based a novel deep learning model, named the Faster Region-based Convolutional Neural Network (Faster R-CNN). This approach is capable of achieving fast end-to-end oil spill detection with reasonable accuracy. A large data set consisting of 15,774 labeled oil spill samples derived from 1786C-band Sentinel-1 and RADARSAT-2 vertical polarization SAR images is used to train, validate and test the Faster R-CNN model. Our experimental results show that the proposed method exhibits good performance for detection of oil spills with wide swath SAR imagery. The Precision and Recall metrics are 89.23% and 89.14%, respectively. The average Precision is 92.56%. The effects of environmental conditions and sensor parameters on oil spill detection are analyzed. The expected detection results are obtained when wind speeds and incidence angles are between 3 m/s and 10 m/s, and 21° and 45°, respectively. Furthermore, the computer runtime for oil spill detection is less than 0.05 s for each full SAR image, using a workstation with NVIDIA GeForce RTX 3090 GPU. This suggests that the present approach has potential for applications that require fast oil spill detection from spaceborne SAR images.

Harmful algae and climate change on the Canadian East Coast: Exploring occurrence predictions of Dinophysis acuminata, D. norvegica, and Pseudo-nitzschia seriata.

Harmful algal blooms (HABs) are a threat to human health, local economies, and coastal ecosystems. Generalized additive mixed models (GAMMs) were fitted using a 24-y database in order to predict future occurrences of three distinct species of HABs on the Canadian East Coast, the dinoflagellates Dinophysis acuminata and D. norvegica, and the diatom Pseudo-nitzschia seriata. GAMMs produced for each species were combined with two downscaled climate simulations (MPI-ESM-LR and CanESM2) under the representative concentration pathway (RCP) 8.5 over the 21st century. D. acuminata, D. norvegica, and P. seriata GAMMs were fitted using sea surface salinity and sea surface temperature, with wind speed averaged over seven days added to the P. seriata model. GAMMs succeeded at various degrees at reproducing past HAB events, with D. acuminata and D. norvegica being accurately modelled, and P. seriata producing less precise model results. Both climate simulations lead to similar conclusions in regards to the spatio-temporal shift in occurrences of the three studied species. D. acuminata and D. norvegica blooms (≥ 1000 cells L - 1) are predicted to increase in the future, whereas P. seriata bloom events (≥ 5000 cells L - 1) will tend to stabilise/decrease overall on the Canadian East Coast. Dinophysis blooms are most likely to increase in the St. Lawrence Estuary. Pseudo-nitzschia blooms will move to the northeastern part of the Gulf of St. Lawrence and will increase in the Bay of Fundy/Gulf of Maine regions. On average, earlier blooms and larger seasonal windows of opportunity are predicted across all species investigated. We conclude that changes in D. acuminata, D. norvegica, and P. seriata bloom dynamics and their spatial distributions could threaten aquaculture industries and ecosystem health on Canada's East Coast in localities and during seasons which were not previously impacted by these species.

Surfactant-associated bacteria in the near-surface layer of the ocean.

Certain marine bacteria found in the near-surface layer of the ocean are expected to play important roles in the production and decay of surface active materials; however, the details of these processes are still unclear. Here we provide evidence supporting connection between the presence of surfactant-associated bacteria in the near-surface layer of the ocean, slicks on the sea surface, and a distinctive feature in the synthetic aperture radar (SAR) imagery of the sea surface. From DNA analyses of the in situ samples using pyrosequencing technology, we found the highest abundance of surfactant-associated bacterial taxa in the near-surface layer below the slick. Our study suggests that production of surfactants by marine bacteria takes place in the organic-rich areas of the water column. Produced surfactants can then be transported to the sea surface and form slicks when certain physical conditions are met. This finding has potential applications in monitoring organic materials in the water column using remote sensing techniques. Identifying a connection between marine bacteria and production of natural surfactants may provide a better understanding of the global picture of biophysical processes at the boundary between the ocean and atmosphere, air-sea exchange of greenhouse gases, and production of climate-active marine aerosols.

Variability in the backscattering efficiency of particles in the Bohai and Yellow Seas and related effects on optical properties.

The backscattering efficiency of particles is a crucial factor that relates light backscattering with biogeochemical properties. In this study, based on in situ measurements of the backscattering coefficient (bbp(λ)), particle biogeochemical variables and remote sensing reflectance (Rrs(λ)) in two typical shallow and semi-enclosed seas, namely the Bohai Sea (BS) and Yellow Sea (YS) during the late spring, late summer and late autumn, we examined particulate pseudo-backscattering efficiency variability at 640 nm (P_Qbbe(640)) and related optical effects. The results show that the P_Qbbe(640) levels varied by nearly two orders for all of the samples examined. This high degree of P_Qbbe(640) variability significantly affected bbp(640) and the mass-specific backscattering coefficient (bbp*(640)), showing that approximately 63.7% and 20.8% of the variability in the bbp*(640) and bbp(640) was attributed to the P_Qbbe(640), respectively. More importantly, consistent with the observations of Wang et al. [J. Geophys. Res.: Oceans 121, 3955 (2016)], the P_Qbbe(640) results clearly showed two clusters and this clustering changed the relationships between bbp*(640), bbp(640) and Rrs(640) with the biogeochemical variables. However, we confirm that P_Qbbe(640) clustering generally remained intact across seasons. Therefore, a simple scheme based on a threshold of the P_Qbbe(640) data is proposed for the classification of particle types. With this classification, impacts of P_Qbbe(640) on bbp*(640) and bbp(640) were clearly reduced, and co-variation trends of bbp*(640), bbp(640) and Rrs(640) with biogeochemical variables can be in turn more accurately described. Overall, this study provides general information on P_Qbbe(640) variability in the BS and the YS and consequent effects on optical properties. The scheme for particle type classification may also provide a useful basis for better modeling marine biogeochemical processes related to particulate backscattering and for the development of ocean color algorithms.

Detection of macroalgae blooms by complex SAR imagery.

Increased frequency and enhanced damage to the marine environment and to human society caused by green macroalgae blooms demand improved high-resolution early detection methods. Conventional satellite remote sensing methods via spectra radiometers do not work in cloud-covered areas, and therefore cannot meet these demands for operational applications. We present a methodology for green macroalgae bloom detection based on RADARSAT-2 synthetic aperture radar (SAR) images. Green macroalgae patches exhibit different polarimetric characteristics compared to the open ocean surface, in both the amplitude and phase domains of SAR-measured complex radar backscatter returns. In this study, new index factors are defined which have opposite signs in green macroalgae-covered areas, compared to the open water surface. These index factors enable unsupervised detection from SAR images, providing a high-resolution new tool for detection of green macroalgae blooms, which can potentially contribute to a better understanding of the mechanisms related to outbreaks of green macroalgae blooms in coastal areas throughout the world ocean.