Predicting ocean-induced ice-shelf melt rates using a machine learning image segmentation approach

Through their role in buttressing upstream ice flow, Antarctic ice shelves play an important part in regulating future sea level change. Reduction in ice-shelf buttressing caused by increased ocean-induced melt along their undersides is now understood to be one of the key drivers of ice loss from the Antarctic Ice Sheet. However, despite the importance of this forcing mechanism most ice-sheet simulations currently rely on simple melt-parametrisations of this ocean-driven process, since a fully coupled ice-ocean modelling framework is prohibitively computationally expensive. Here, we provide an alternative approach that is able to capture the greatly improved physical description of this process provided by large-scale ocean-circulation models over currently employed melt-parameterisations but with trivial computational expense. We introduce a new approach that brings together deep learning and physical modelling to develop a deep neural network framework, MELTNET, that can emulate ocean model predictions of sub-ice shelf melt rates. We train MELTNET on synthetic geometries, using the NEMO ocean model as a ground-truth in lieu of observations to provide melt rates both for training and to evaluate the performance of the trained network. We show that MELTNET can accurately predict melt rates for a wide range of complex synthetic geometries and outperforms more traditional parameterisations for > 95 % of geometries tested. Furthermore, we find MELTNET's melt rate estimates show sensitivity to established physical relationships such as a changes in thermal forcing and ice shelf slope. This study demonstrates the potential for a deep learning framework to calculate melt rates with almost no computational expense, that could in the future be used in conjunction with an ice sheet model to provide predictions for large-scale ice sheet models.

Modeling studies of the bioluminescence potential dynamics in a high Arctic fjord during polar night

During polar nights in January 2012 and 2017, significantly higher bioluminescence (BL) potential emissions in the upper 50 m were observed in the fjord Rijpfjorden (Svalbard, Norway) in comparison to offshore stations (located on the shelf-break, shelf-slope areas and in the deeper water). The objective of this paper is to better understand why, during two polar nights (separated by 5 years), the values of BL potential in the northern Svalbard fjord are higher than at offshore stations, and what the role of advection is in observed elevated BL potential values in the top 50 m of the fjord. To address the above objective, we applied the same BL potential modeling approach and strategies during polar nights for both 2012 and 2017. For both years, advection of BL potential from offshore (including upwelling along the shelf, shelf-slope) produced an increase of BL potential in the fjord area, in spite of the introduction of mortality in bioluminescent organisms. Observations of BL potential indicated high emissions at depths below 100 m at offshore stations for both polar nights. Our modeling studies demonstrated that these high values of BL potential below 100 m are upwelled and advected to the top 50 m of the fjord. We demonstrated that upwelling and advection of these deep high BL potential values (and therefore, upwelling and advection of corresponding bioluminescent taxa) from offshore areas are dominant factors in observed BL potential dynamics in the top 50 m in the fjord.

Depositional History of Paleocene Sediments in the Offshore Canterbury Basin, New Zealand

<p>The Paleocene interval within the Canterbury Basin has been relatively understudied with respect to the Neogene and Cretaceous intervals. Within the Paleocene interval is the Tartan Formation and the Charteris Bay Sandstone, which are potential source and reservoir rocks respectively. These two formations have not been previously mapped in the offshore Canterbury Basin and their limits have not been defined. This study utilises a database of nearly 12,000km of 2D seismic data together with data from four open–file wells and sidewall core samples from three wells and newly availiable biostratigraphic information to better constrain the chronostratigraphical interpretation of seismic data. Seismic mapping together with corroboration from well correlation and core lithofacies analysis revealed new insights into the development of the offshore Canterbury Basin through the Paleocene. These include the delineation of the lateral extents and thicknesses of the Tartan Formation and Charteris Bay Sandstone and location of the palaeo shelf–slope break and also the development of a new well correlation panel that incorporates the Tartan Formation for the first time. This study presents four new paleogeographic maps for the offshore Canterbury Basin that significantly improves our understanding of the development of the basin during the Paleocene. These maps show that during the Earliest Paleocene, the mudstones of the Katiki Formation were being deposited in the south of the study area, with the siltier sediments of the Conway Formation being deposited in the north. The coarser grained Charteris Bay Sandstone was deposited from Early to possibly Middle Paleocene in the northeast. The mudstones of the Moeraki Formation were being deposited in the south at this time. From Middle to Late Paleocene, the mudstones of the Moeraki Formation were deposited in the south and these mudstones onlapped against the Charteris Bay Sandstone which remained as a high in the north. The Tartan Formation was deposited during the Late Paleocene in the central and southern areas of the offshore Canterbury Basin, during a relative fall in sea–level. Deposition had ceased in the north of the study area or erosion possibly removed Late Paleocene sediments from there. During the Latest Paleocene, the mudstones of the Moeraki Formation were deposited over the Tartan Formation in the central and southern parts of the offshore Canterbury Basin with the northern area undergoing erosion, sediment bypass or both.</p>

Depositional History of Paleocene Sediments in the Offshore Canterbury Basin, New Zealand

<p>The Paleocene interval within the Canterbury Basin has been relatively understudied with respect to the Neogene and Cretaceous intervals. Within the Paleocene interval is the Tartan Formation and the Charteris Bay Sandstone, which are potential source and reservoir rocks respectively. These two formations have not been previously mapped in the offshore Canterbury Basin and their limits have not been defined. This study utilises a database of nearly 12,000km of 2D seismic data together with data from four open–file wells and sidewall core samples from three wells and newly availiable biostratigraphic information to better constrain the chronostratigraphical interpretation of seismic data. Seismic mapping together with corroboration from well correlation and core lithofacies analysis revealed new insights into the development of the offshore Canterbury Basin through the Paleocene. These include the delineation of the lateral extents and thicknesses of the Tartan Formation and Charteris Bay Sandstone and location of the palaeo shelf–slope break and also the development of a new well correlation panel that incorporates the Tartan Formation for the first time. This study presents four new paleogeographic maps for the offshore Canterbury Basin that significantly improves our understanding of the development of the basin during the Paleocene. These maps show that during the Earliest Paleocene, the mudstones of the Katiki Formation were being deposited in the south of the study area, with the siltier sediments of the Conway Formation being deposited in the north. The coarser grained Charteris Bay Sandstone was deposited from Early to possibly Middle Paleocene in the northeast. The mudstones of the Moeraki Formation were being deposited in the south at this time. From Middle to Late Paleocene, the mudstones of the Moeraki Formation were deposited in the south and these mudstones onlapped against the Charteris Bay Sandstone which remained as a high in the north. The Tartan Formation was deposited during the Late Paleocene in the central and southern areas of the offshore Canterbury Basin, during a relative fall in sea–level. Deposition had ceased in the north of the study area or erosion possibly removed Late Paleocene sediments from there. During the Latest Paleocene, the mudstones of the Moeraki Formation were deposited over the Tartan Formation in the central and southern parts of the offshore Canterbury Basin with the northern area undergoing erosion, sediment bypass or both.</p>

Utilization of Soot and 210 Po-210 Pb Disequilibria to Constrain Particulate Organic Carbon Fluxes in the Northeastern South China Sea

Black carbon (BC) is believed to be refractory and thus affects the timescale of organic carbon conversion into CO2 and the magnitude of the sink of CO2. However, the fate of BC in the oceans remains poorly understood. Here, 210Po and 210Pb were measured to examine the export of soot in the northeastern South China Sea (SCS). Concentrations of soot decreased from 0.141 ± 0.021 μmol-C L–1 (mean ± SD) in the mixed layer (0–30 m) to 0.087 μmol-C L–1 at the euphotic base (150 m) due to potential photodegradation within the euphotic zone. In the twilight zone, however, the soot showed an increasing pattern along with the total particulate matter and total particulate organic carbon (POC) contents, corresponding to additions from the shelf/slope sediment resuspension through lateral transport. Using the deficits of 210Po, the export flux of soot from the euphotic zone was calculated to be 0.172 ± 0.016 mmol-C m–2 d–1 and increased with depth. Assuming that the soot is entirely refractory below the euphotic zone, the sediment-derived soot fluxes were estimated based on the increase in soot fluxes relative to the base of the euphotic zone, with values varying from 0.149 ± 0.030 to 0.96 ± 0.10 μmol-C L–1. This indicates that sediment resuspension is an important source of soot to the ocean interior in the SCS. Coupling the sediment-derived soot and 210Po-derived POC fluxes gave rise to a Martin Curve-like flux attenuation of local euphotic zone-derived POC in the twilight zone with b value of 0.70 ± 0.01. These results suggest that soot could be useful for constraining in situ POC fluxes and their transport.

Late Neogene Radiolaria from the East Coast Deformed Belt, New Zealand

<p>Within the East Coast Deformed Belt there are a number of Late Neogene sedimentary basins with relatively deep-water sediments which, at places, contain abundant radiolarian skeletons. The region was subject to relatively open ocean circulation patterns during the Neogene which, combined with the input of rhyolitic glass shards, has enhanced the siliceous microfossil preservation. A short review of the silica budget is presented and discussed in relation to the preservation of siliceous microfossils in the New Zealand sequences. Techniques were developed to extract and quantitatively study fossil Radiolaria from some of the relatively barren shelf/slope sediments. One hundred and thirty-eight radiolarian taxa are described, most of which can be assigned at the generic level, but thirty-one of which can not be assigned specific names and may eventually prove to be new species. The radiolarian zonation presented is based on detailed analysis of 155 samples from 26 sections and sites ranging in age from basal Tongaporutuan (early Late Miocene) to middle Nukumaruan (early Pleistocene). Sediments of the Kapitean (uppermost Miocene) were generally deposited in shallow water environments or are missing in unconformities in the East Coast Deformed Belt, consequently the radiolarian zonation is based on very poor data in this time segment. Also upper Opoitian and Waipipian (middle Pliocene) sediments, although at places deposited in relatively deep water, generally lack siliceous tuffs, and radiolarian preservation is poor. Five major radiolarian zones can be recognised: Diartus hughesi Zone, Didymocyrtis sp. A Zone, Didymocyrtis sp. A Zone, Didymocyrtis tetrathalmus tetrathalmus Zone, Lamprocyrtis heteroporos Zone, and Lamprocyclas gamphonycha Zone. In samples with good radiolarian preservation six subzones can be identified. The Diartus hughesi Zone can be divided into the Heliodiscus umbonatum Subzone, Didymocyrtis laticonus Subzone, Heliodiscus asteriscus forma large pores Subzone, and Anthocyrtidium ehrenbergi pliocenica Subzone. Additionally the Didymocyrtis tetrathalmus tetrathalmus Zone can be divided into the Lychnocanium sp. aff. grande Subzone and Lamprocyrtis hannai Subzone. The bioevents that define the zonal boundaries are discussed along with other biostratigraphically useful radiolarian datums. These zones and zubzones are correlated to the foraminiferal zonation which in turn has been related, in part, to the paleomagnetic time scale. Correlation are then made with other radiolarian zonations in the north Pacific, tropics, and southern ocean. Points to emerge from these correlations include the apparent provincialism in the transition from Stichocorys delmontense to Stichocorys peregrine in the tropical Pacific. This transition has been reported to occur during approximately 1.5Ma but in New Zealand occurs over a time segment of at least 5.5Ma. The first appearance of Lamprocyclas gamphonycha appears to be an isochronous datum level in temperate radiolarian faunas of the northern and southern Pacific. The last appearance datum of Diartus hughesi at about 7.5Ma is in good agreement with its level in the tropics. The presence of this taxon in lower Gilbert Antarctic cores suggests either a grossly diachronous event between tropical/temperate areas and the southern ocean or, more probably, a misinterpretation of the paleomagnetic signature from key southern ocean piston cores. If the latter situation is the case then the real age estimates on the "Pre middle Gilbert" southern ocean diatom and silicoflagellate stratigraphies are questionable because they are based on the same key cores. Statistical faunal analysis shows that during the Miocene there was not much change in the radiolarian faunas with time and a major change, probably climatically controlled, took place across the Miocene/Pliocene boundary. Variability in preservation has probably affected the faunas to obscure more precise time variation although post-Miocene faunas indicate that some is present. In conclusion, the Radiolaria, although not as common in the fossil record as the foraminifera, definitely contribute to New Zealand Late Neogene integrated stratigraphy and suggest that our knowledge could be greatly enhanced by the study of other siliceous microfossil groups.</p>

Late Neogene Radiolaria from the East Coast Deformed Belt, New Zealand

<p>Within the East Coast Deformed Belt there are a number of Late Neogene sedimentary basins with relatively deep-water sediments which, at places, contain abundant radiolarian skeletons. The region was subject to relatively open ocean circulation patterns during the Neogene which, combined with the input of rhyolitic glass shards, has enhanced the siliceous microfossil preservation. A short review of the silica budget is presented and discussed in relation to the preservation of siliceous microfossils in the New Zealand sequences. Techniques were developed to extract and quantitatively study fossil Radiolaria from some of the relatively barren shelf/slope sediments. One hundred and thirty-eight radiolarian taxa are described, most of which can be assigned at the generic level, but thirty-one of which can not be assigned specific names and may eventually prove to be new species. The radiolarian zonation presented is based on detailed analysis of 155 samples from 26 sections and sites ranging in age from basal Tongaporutuan (early Late Miocene) to middle Nukumaruan (early Pleistocene). Sediments of the Kapitean (uppermost Miocene) were generally deposited in shallow water environments or are missing in unconformities in the East Coast Deformed Belt, consequently the radiolarian zonation is based on very poor data in this time segment. Also upper Opoitian and Waipipian (middle Pliocene) sediments, although at places deposited in relatively deep water, generally lack siliceous tuffs, and radiolarian preservation is poor. Five major radiolarian zones can be recognised: Diartus hughesi Zone, Didymocyrtis sp. A Zone, Didymocyrtis sp. A Zone, Didymocyrtis tetrathalmus tetrathalmus Zone, Lamprocyrtis heteroporos Zone, and Lamprocyclas gamphonycha Zone. In samples with good radiolarian preservation six subzones can be identified. The Diartus hughesi Zone can be divided into the Heliodiscus umbonatum Subzone, Didymocyrtis laticonus Subzone, Heliodiscus asteriscus forma large pores Subzone, and Anthocyrtidium ehrenbergi pliocenica Subzone. Additionally the Didymocyrtis tetrathalmus tetrathalmus Zone can be divided into the Lychnocanium sp. aff. grande Subzone and Lamprocyrtis hannai Subzone. The bioevents that define the zonal boundaries are discussed along with other biostratigraphically useful radiolarian datums. These zones and zubzones are correlated to the foraminiferal zonation which in turn has been related, in part, to the paleomagnetic time scale. Correlation are then made with other radiolarian zonations in the north Pacific, tropics, and southern ocean. Points to emerge from these correlations include the apparent provincialism in the transition from Stichocorys delmontense to Stichocorys peregrine in the tropical Pacific. This transition has been reported to occur during approximately 1.5Ma but in New Zealand occurs over a time segment of at least 5.5Ma. The first appearance of Lamprocyclas gamphonycha appears to be an isochronous datum level in temperate radiolarian faunas of the northern and southern Pacific. The last appearance datum of Diartus hughesi at about 7.5Ma is in good agreement with its level in the tropics. The presence of this taxon in lower Gilbert Antarctic cores suggests either a grossly diachronous event between tropical/temperate areas and the southern ocean or, more probably, a misinterpretation of the paleomagnetic signature from key southern ocean piston cores. If the latter situation is the case then the real age estimates on the "Pre middle Gilbert" southern ocean diatom and silicoflagellate stratigraphies are questionable because they are based on the same key cores. Statistical faunal analysis shows that during the Miocene there was not much change in the radiolarian faunas with time and a major change, probably climatically controlled, took place across the Miocene/Pliocene boundary. Variability in preservation has probably affected the faunas to obscure more precise time variation although post-Miocene faunas indicate that some is present. In conclusion, the Radiolaria, although not as common in the fossil record as the foraminifera, definitely contribute to New Zealand Late Neogene integrated stratigraphy and suggest that our knowledge could be greatly enhanced by the study of other siliceous microfossil groups.</p>

The propagation and decay of a coastal vortex on a shelf

A coastal eddy is modelled as a barotropic vortex propagating along a coastal shelf. If the vortex speed matches the phase speed of any coastal trapped shelf wave modes, a shelf wave wake is generated leading to a flux of energy from the vortex into the wave field. Using a simple shelf geometry, we determine analytic expressions for the wave wake and the leading-order flux of wave energy. By considering the balance of energy between the vortex and wave field, this energy flux is then used to make analytic predictions for the evolution of the vortex speed and radius under the assumption that the vortex structure remains self-similar. These predictions are examined in the asymptotic limit of small rotation rate and shelf slope and tested against numerical simulations. If the vortex speed does not match the phase speed of any shelf wave, steady vortex solutions are expected to exist. We present a numerical approach for finding these nonlinear solutions and examine the parameter dependence of their structure.