Sea surface temperature climate change initiative: alternative image classification algorithms for sea-ice affected oceans

We present a Bayesian image classification scheme for discriminating cloud, clear and sea-ice observations at high latitudes to improve identification of areas of clear-sky over ice-free ocean for SST retrieval. We validate the image classification against a manually classified dataset using Advanced Along Track Scanning Radiometer (AATSR) data. A three way classification scheme using a near-infrared textural feature improves classifier accuracy by 9.9 % over the nadir only version of the cloud clearing used in the ATSR Reprocessing for Climate (ARC) project in high latitude regions. The three way classification gives similar numbers of cloud and ice scenes misclassified as clear but significantly more clear-sky cases are correctly identified (89.9 % compared with 65 % for ARC). We also demonstrate the poetential of a Bayesian image classifier including information from the 0.6 micron channel to be used in sea-ice extent and ice surface temperature retrieval with 77.7 % of ice scenes correctly identified and an overall classifier accuracy of 96 %.

Ice-stream initiation, duration and thinning on James Ross Island, northern Antarctic Peninsula

Predicting the future response of the Antarctic Ice Sheet to climate change requires an understanding of the ice streams that dominate its dynamics. Here we use cosmogenic isotope exposure-age dating (26Al, 10Be and 36Cl) of erratic boulders on ice-free land on James Ross Island, north-eastern Antarctic Peninsula, to define the evolution of Last Glacial Maximum (LGM) ice in the adjacent Prince Gustav Channel. These data include ice-sheet extent, thickness and dynamical behaviour. Prior to ∼18 ka, the LGM Antarctic Peninsula Ice Sheet extended to the continental shelf-edge and transported erratic boulders onto high-elevation mesas on James Ross Island. After ∼18 ka there was a period of rapid ice-sheet surface-lowering, coincident with the initiation of the Prince Gustav Ice Stream. This timing coincided with rapid increases in atmospheric temperature and eustatic sea-level rise around the Antarctic Peninsula. Collectively, these data provide evidence for a transition from a thick, cold-based LGM Antarctic Peninsula Ice Sheet to a thinner, partially warm-based ice sheet during deglaciation.

Glaciodynamics of the central sector of the last British–Irish ice sheet in Northern England

The central sector of the last British–Irish Ice Sheet (BIIS) was characterised by considerable complexity, both in terms of its glacial stratigraphy and geomorphological signature. This complexity is reflected by the large number and long history of papers that have attempted to decipher the glaciodynamic history of the region. Despite significant advances in our understanding, reconstructions remain hotly debated and relatively local, thereby hindering attempts to piece together BIIS dynamics. This paper seeks to address these issues by reviewing geomorphological mapping evidence of palimpsest flow signatures and providing an up-to-date stratigraphy of the region. Reconciling geomorphological and sedimentological evidence with relative and absolute dating constraints has allowed us to develop a new six-stage glacial model of ice-flow history and behaviour in the central sector of the last BIIS, with three major phases of glacial advance. This includes: I. Eastwards ice flow through prominent topographic corridors of the north Pennines; II. Cessation of the Stainmore ice flow pathway and northwards migration of the North Irish Sea Basin ice divide; III. Stagnation and retreat of the Tyne Gap Ice Stream; IV. Blackhall Wood–Gosforth Oscillation; V. Deglaciation of the Solway Lowlands; and VI. Scottish Re-advance and subsequent final retreat of ice out of the central sector of the last BIIS. The ice sheet was characterised by considerable dynamism, with flow switches, initiation (and termination) of ice streams, draw-down of ice into marine ice streams, repeated ice-marginal fluctuations and the production of large volumes of meltwater, locally impounded to form ice-dammed glacial lakes. Significantly, we tie this reconstruction to work carried out and models developed for the entire ice sheet. This therefore situates research in the central sector within contemporary understanding of how the last BIIS evolved over time.

Landscape evolution and ice-sheet behaviour in a semi-arid polar environment: James Ross Island, NE Antarctic Peninsula

This study of landscape evolution presents both new modern and palaeo process-landform data, and analyses the behaviour of the Antarctic Peninsula Ice Sheet through the Last Glacial Maximum (LGM), the Holocene and to the present day. Six sediment-landform assemblages are described and interpreted for Ulu Peninsula, James Ross Island, NE Antarctic Peninsula: (1) the Glacier Ice and Snow Assemblage; (2) the Glacigenic Assemblage, which relates to LGM sediments and comprises both erratic-poor and erratic-rich drift, deposited by cold-based and wet-based ice and ice streams respectively; (3) the Boulder Train Assemblage, deposited during a Mid-Holocene glacier readvance; (4) the Ice-cored Moraine Assemblage, found in front of small cirque glaciers; (5) the Paraglacial Assemblage including scree, pebble-boulder lags, and littoral and fluvial processes; and (6) the Periglacial Assemblage including rock glaciers, protalus ramparts, blockfields, solifluction lobes and extensive patterned ground. The interplay between glacial, paraglacial and periglacial processes in this semi-arid polar environment is important in understanding polygenetic landforms. Crucially, cold-based ice was capable of sediment and landform genesis and modification. This landsystem model can aid the interpretation of past environments, but also provides new data to aid the reconstruction of the last ice sheet to overrun James Ross Island.

Dynamic Devensian ice flow in NE England: a sedimentological reconstruction

This study reconstructs the depositional environments that accompanied both ice advance and ice retreat of the last British–Irish Ice Sheet in NE England during the Last Glacial Maximum, and proposes three regional ice-flow phases. The Late Devensian (29–22 cal. ka BP) Tyne Gap Ice Stream initially deposited the Blackhall Till Formation during shelf-edge glaciation (Phase I). This subglacial traction till comprises several related facies, including stratified and laminated diamictons, tectonites, and sand and gravel beds deposited both in subglacial canals and in proglacial streams. Eventually, stagnation of the Tyne Gap Ice Stream led to ice-marginal sedimentation in County Durham (Phase II). During the Dimlington Stadial (21 cal. ka BP), the North Sea Lobe advanced towards the coastline of N Norfolk. This resulted initially in sandur deposition (widespread, tabular sand and gravel; the Peterlee Sand and Gravel Formation; Phase II) and ultimately in deposition of the Horden Till Formation (Phase III), a massive subglacial till. As the North Sea Lobe overrode previous formations, it thrusted and stacked sediments in County Durham, and dammed proglacial lakes between the east-coast ice, the Pennine uplands and the remaining Pennine ice. The North Sea Lobe retreated after Heinrich Event 1 (16 ka). This study highlights the complexity of ice flow during the Late Devensian glaciation of NE England, with changing environmental and oceanic conditions forcing a mobile and sensitive ice sheet.

Antarctic Peninsula ice sheet evolution during the Cenozoic Era

The Antarctic Peninsula region is currently undergoing rapid environmental change, resulting in the thinning, acceleration and recession of glaciers and the sequential collapse of ice shelves. It is important to view these changes in the context of long-term palaeoenvironmental complexity and to understand the key processes controlling ice sheet growth and recession. In addition, numerical ice sheet models require detailed geological data for tuning and testing. Therefore, this paper systematically and holistically reviews published geological evidence for Antarctic Peninsula Ice Sheet variability for each key locality throughout the Cenozoic, and brings together the prevailing consensus of the extent, character and behaviour of the glaciations of the Antarctic Peninsula region. Major contributions include a downloadable database of 186 terrestrial and marine calibrated dates; an original reconstruction of the LGM ice sheet; and a new series of isochrones detailing ice sheet retreat following the LGM. Glaciation of Antarctica was initiated around the Eocene/Oligocene transition in East Antarctica. Palaeogene records of Antarctic Peninsula glaciation are primarily restricted to King George Island, where glacigenic sediments provide a record of early East Antarctic glaciations, but with modification of far-travelled erratics by local South Shetland Island ice caps. Evidence for Neogene glaciation is derived primarily from King George Island and James Ross Island, where glaciovolcanic strata indicate that ice thicknesses reached 500–850 m during glacials. This suggests that the Antarctic Peninsula Ice Sheet draped, rather than drowned, the topography. Marine geophysical investigations indicate multiple ice sheet advances during this time. Seismic profiling of continental shelf-slope deposits indicates up to ten large advances of the Antarctic Peninsula Ice Sheet during the Early Pleistocene, when the ice sheet was dominated by 40 kyr cycles. Glacials became more pronounced, reaching the continental shelf edge, and of longer duration during the Middle Pleistocene. During the Late Pleistocene, repeated glacials reached the shelf edge, but ice shelves inhibited iceberg rafting. The Last Glacial Maximum (LGM) occurred at 18 ka BP, after which transitional glaciomarine sediments on the continental shelf indicate ice-sheet retreat. The continental shelf contains large bathymetric troughs, which were repeatedly occupied by large ice streams during Pleistocene glaciations. Retreat after the LGM was episodic in the Weddell Sea, with multiple readvances and changes in ice-flow direction, but rapid in the Bellingshausen Sea. The late Holocene Epoch was characterised by repeated fluctuations in palaeoenvironmental conditions, with associated glacial readvances. However, this has been subsumed by rapid warming and ice-shelf collapse during the twentieth century.

Accelerating shrinkage of Patagonian glaciers from the Little Ice Age (∼AD 1870) to 2011

We used Little Ice Age (LIA) trimlines and moraines to assess changes in South American glaciers over the last ∼140 years. We determined the extent and length of 640 glaciers during the LIA (∼ AD 1870) and 626 glaciers (the remainder having entirely disappeared) in 1986, 2001 and 2011. The calculated reduction in glacierized area between the LIA and 2011 is 4131 km2 (15.4%), with 660 km2 (14.2%) being lost from the Northern Patagonia Icefield (NPI), 1643 km2 (11.4%) from the Southern Patagonia Icefield (SPI) and 306 km2 (14.4%) from Cordillera Darwin. Latitude, size and terminal environment (calving or land-terminating) exert the greatest control on rates of shrinkage. Small, northerly, land-terminating glaciers shrank fastest. Annual rates of area loss increased dramatically after 2001 for mountain glaciers north of 52° S and the large icefields, with the NPI and SPI now shrinking at 9.4 km2 a–1 (0.23% a–1) and 20.5 km2 a–1 (0.15% a–1) respectively. The shrinkage of glaciers between 52° S and 54° S accelerated after 1986, and rates of shrinkage from 1986 to 2011 remained steady. Icefield outlet glaciers, isolated glaciers and ice caps south of 54° S shrank faster from 1986 to 2001 than they did from 2001 to 2011.

Dust sources and deposition during the last glacial maximum and current climate: a comparison of model results with paleodata from ice cores and marine sediments

Mineral dust aerosols in the atmosphere have the potential to affect the global climate by influencing the radiative balance of the atmosphere and the supply of micronutrients to the ocean. Ice and marine sediment cores indicate that dust deposition from the atmosphere was at some locations 2–20 times greater during glacial periods, raising the possibility that mineral aerosols might have contributed to climate change on glacial-interglacial time scales. To address this question, we have used linked terrestrial biosphere, dust source, and atmospheric transport models to simulate the dust cycle in the atmosphere for current and last glacial maximum (LGM) climates. We obtain a 2.5-fold higher dust loading in the entire atmosphere and a twenty-fold higher loading in high latitudes, in LGM relative to present. Comparisons to a compilation of atmospheric dust deposition flux estimates for LGM and present in marine sediment and ice cores show that the simulated flux ratios are broadly in agreement with observations; differences suggest where further improvements in the simple dust model could be made. The simulated increase in high-latitude dustiness depends on the expansion of unvegetated areas, especially in the high latitudes and in central Asia, caused by a combination of increased aridity and low atmospheric [CO2]. The existence of these dust source areas at the LGM is supported by pollen data and loess distribution in the northern continents. These results point to a role for vegetation feedbacks, including climate effects and physiological effects of low [CO2], in modulating the atmospheric distribution of dust.

September Arctic sea-ice minimum predicted by spring melt-pond fraction

The area of Arctic September sea ice has diminished from about 7 million km2 in the 1990s to less than 5 million km2 in five of the past seven years, with a record minimum of 3.6 million km2 in 2012 (ref. 1). The strength of this decrease is greater than expected by the scientific community, the reasons for this are not fully understood, and its simulation is an on-going challenge for existing climate models2, 3. With growing Arctic marine activity there is an urgent demand for forecasting Arctic summer sea ice4. Previous attempts at seasonal forecasts of ice extent were of limited skill5, 6, 7, 8, 9. However, here we show that the Arctic sea-ice minimum can be accurately forecasted from melt-pond area in spring. We find a strong correlation between the spring pond fraction and September sea-ice extent. This is explained by a positive feedback mechanism: more ponds reduce the albedo; a lower albedo causes more melting; more melting increases pond fraction. Our results help explain the acceleration of Arctic sea-ice decrease during the past decade. The inclusion of our new melt-pond model10 promises to improve the skill of future forecast and climate models in Arctic regions and beyond.

Impact of variable atmospheric and oceanic form drag on simulations of Arctic sea ice

Over Arctic sea ice, pressure ridges and floe andmelt pond edges all introduce discrete obstructions to the flow of air or water past the ice and are a source of form drag. In current climate models form drag is only accounted for by tuning the air–ice and ice–ocean drag coefficients, that is, by effectively altering the roughness length in a surface drag parameterization. The existing approach of the skin drag parameter tuning is poorly constrained by observations and fails to describe correctly the physics associated with the air–ice and ocean–ice drag. Here, the authors combine recent theoretical developments to deduce the total neutral form drag coefficients from properties of the ice cover such as ice concentration, vertical extent and area of the ridges, freeboard and floe draft, and the size of floes and melt ponds. The drag coefficients are incorporated into the Los Alamos Sea Ice Model (CICE) and show the influence of the new drag parameterization on the motion and state of the ice cover, with the most noticeable being a depletion of sea ice over the west boundary of the Arctic Ocean and over the Beaufort Sea. The new parameterization allows the drag coefficients to be coupled to the sea ice state and therefore to evolve spatially and temporally. It is found that the range of values predicted for the drag coefficients agree with the range of values measured in several regions of the Arctic. Finally, the implications of the new form drag formulation for the spinup or spindown of the Arctic Ocean are discussed.

A Lagrangian analysis of ice-supersaturated air over the North Atlantic

Understanding the nature of air parcels that exhibit ice-supersaturation is important because they are the regions of potential formation of both cirrus and aircraft contrails, which affect the radiation balance. Ice-supersaturated air parcels in the upper troposphere and lower stratosphere over the North Atlantic are investigated using Lagrangian trajectories. The trajectory calculations use ERA-Interim data for three winter and three summer seasons, resulting in approximately 200,000 trajectories with ice-supersaturation for each season. For both summer and winter, the median duration of ice-supersaturation along a trajectory is less than 6 hours. 5% of air which becomes ice-supersaturated in the troposphere, and 23% of air which becomes ice-supersaturated in the stratosphere will remain ice-supersaturated for at least 24 hours. Weighting the ice-supersaturation duration with the observed frequency indicates the likely overall importance of the longer duration ice-supersaturated trajectories. Ice-supersaturated air parcels typically experience a decrease in moisture content while ice-supersaturated, suggesting that cirrus clouds eventually form in the majority of such air. A comparison is made between short-lived (less than 24 h) and long-lived (greater than 24 h) ice-supersaturated air flows. For both air flows, ice-supersaturation occurs around the northernmost part of the trajectory. Short-lived ice-supersaturated air flows show no significant differences in speed or direction of movement to subsaturated air parcels. However, long-lived ice-supersaturated air occurs in slower moving air flows, which implies that they are not associated with the fastest moving air through a jet stream.