Cloud computing and virtualization within the regional climate model and evaluation system

The Regional Climate Model Evaluation System (RCMES) facilitates the rapid, flexible inclusion of NASA observations into climate model evaluations. RCMES provides two fundamental components. A database (RCMED) is a scalable point-oriented cloud database used to elastically store remote sensing observations and to make them available using a space time query interface. The analysis toolkit (RCMET) is a Python-based toolkit that can be delivered as a cloud virtual machine, or as an installer package deployed using Python Buildout to users in order to allow for temporal and spatial regridding, metrics calculation (RMSE, bias, PDFs, etc.) and end-user visualization. RCMET is available to users in an “offline”, lone scientist mode based on a virtual machine dynamically constructed with model outputs and observations to evaluate; or on an institution’s computational cluster seated close to the observations and model outputs. We have leveraged RCMES within the content of the Coordinated Regional Downscaling Experiment (CORDEX) project, working with the University of Cape Town and other institutions to compare the model output to NASA remote sensing data; in addition we are also working with the North American Regional Climate Change Assessment Program (NARCCAP). In this paper we explain the contribution of cloud computing to RCMES’s specifically describing studies of various cloud databases we evaluated for RCMED, and virtualization toolkits for RCMET, and their potential strengths in delivering user-created dynamic regional climate model evaluation virtual machines for our users.     

Bayesian modelling the retreat of the Irish Sea Ice Stream

We present an 8000‐year history spanning 650 km of ice margin retreat for the largest marine‐terminating ice stream draining the former British–Irish Ice Sheet. Bayesian modelling of the geochronological data shows the ISIS expanded 34.0–25.3 ka, accelerating into the Celtic Sea to reach maximum limits 25.3–24.5 ka before a collapse with rapid marginal retreat to the northern Irish Sea Basin (ISB). This retreat was rapid and driven by climatic warming, sea‐level rise, mega‐tidal amplitudes and reactivation of meridional circulation in the North Atlantic. The retreat, though rapid, is uneven, with the stepped retreat pattern possibly a function of the passage of the ice stream between normal and adverse ice bed gradients and changing ice stream geometry. Initially, wide calving margins and adverse slopes encouraged rapid retreat (～550 m a^?1) that slowed (～100 m a^?1) at the topographic constriction and bathymetric high between southern Ireland and Wales before rates increased (～200 m a^?1) across adverse bed slopes and wider and deeper basin configuration in the northern ISB. These data point to the importance of the ice bed slope and lateral extent in predicting the longer‐term (>1000 a) patterns and rates of ice‐marginal retreat during phases of rapid collapse, which has implications for the modelling of projected rapid retreat of present‐day ice streams. Copyright © 2013 John Wiley & Sons, Ltd.     

Taphonomic differentiation of Acropora palmata facies in cores from Campeche Bank Reefs, Gulf of México

A common assumption in the geological analysis of modern reefs is that coral community zonation seen on the surface should also be found in cores from the reef interior. Such assumptions not only underestimate the impact of tropical storms on reef facies development, but have been difficult to test because of restrictions imposed by narrow‐diameter cores and poor recovery. That assumption is tested here using large‐diameter cores recovered from a range of common zones across three Campeche Bank reefs. It is found that cores from the reef‐front, crest, flat and rubble‐cay zones are similar in texture and coral composition, making it impossible to recognize coral assemblages that reflect the surface zonation. Taphonomic signatures imparted by variations in encrustation, bioerosion and cementation, however, produce distinct facies and delineate a clear depth zonation. Cores from the reef‐front zone (2–10 m depth) are characterized by sections of Acropora palmata cobble gravel interspersed with sections of in‐place (but truncated) A. palmata stumps. Upper surfaces of truncated colonies are intensely bioeroded by traces of Entobia isp. and Gastrochaenolites isp. and encrusted by mm‐thick crustose corallines before colony regeneration and, therefore, indicate punctuated growth resulting from a hurricane‐induced cycle of destruction and regeneration. Cores from the reef crest/flat (0–2 m depth) are also characterized by sections of hurricane‐derived A. palmata cobble‐gravels as well as in‐place A. palmata colonies. In contrast to the reef front, however, these cobble gravels are encrusted by cm‐thick crusts of intergrown coralline algae, low‐relief Homotrema and vermetids, bored by traces of Entobia isp. and Trypanites isp. and coated by a dense, peloidal, micrite cement. Cores from the inter‐ to supratidal rubble‐cay zone (+0–5 m) are only composed of A. palmata cobble gravels and, although clasts show evidence of subtidal encrustation and bioerosion, these always represent processes that occurred before deposition on the cay. Instead, these gravels are distinguished on the basis of their limited bioerosion and marine cements, which exhibit fabrics formed in the intertidal zone. These results confirm that hurricanes have a major influence on facies development in Campeche Bank reefs. Instead of reflecting the surface coral zonation, each facies records a distinctive, depth‐related set of taphonomic processes, which reflect colonization, alteration and stabilization following the production of new substrates by hurricanes.     

Pattern,style and timing of British–Irish Ice Sheet advance and retreat over the last 45 000 years: evidence from NW Scotland and the adjacent continental shelf

Predicting the future response of ice sheets to climate warming and rising global sea level is important but difficult. This is especially so when fast-flowing glaciers or ice streams, buffered by ice shelves, are grounded on beds below sea level. What happens when these ice shelves are removed? And how do the ice stream and the surrounding ice sheet respond to the abruptly altered boundary conditions? To address these questions and others we present new geological, geomorphological, geophysical and geochronological data from the ice-stream-dominated NW sector of the last British–Irish Ice Sheet (BIIS). The study area covers around 45 000 km² of NW Scotland and the surrounding continental shelf. Alongside seabed geomorphological mapping and Quaternary sediment analysis, we use a suite of over 100 new absolute ages (including cosmogenic-nuclide exposure ages, optically stimulated luminescence ages and radiocarbon dates) collected from onshore and offshore, to build a sector-wide ice-sheet reconstruction combining all available evidence with Bayesian chronosequence modelling. Using this information we present a detailed assessment of ice-sheet advance/retreat history, and the glaciological connections between different areas of the NW BIIS sector, at different times during the last glacial cycle. The results show a highly dynamic, partly marine, partly terrestrial, ice-sheet sector undergoing large size variations in response to sub-millennial-scale climatic (Dansgaard–Oeschger) cycles over the last 45 000 years. Superimposed on these trends we identify internally driven instabilities, operating at higher frequency, conditioned by local topographic factors, tidewater dynamics and glaciological feedbacks during deglaciation. Specifically, our new evidence indicates extensive marine-terminating ice-sheet glaciation of the NW BIIS sector during Greenland Stadials 12 to 9 – prior to the main ‘Late Weichselian’ ice-sheet glaciation. After a period of restricted glaciation, in Greenland Interstadials 8 to 6, we find good evidence for rapid renewed ice-sheet build-up in NW Scotland, with the Minch ice-stream terminus reaching the continental shelf edge in Greenland Stadial 5, perhaps only briefly. Deglaciation of the NW sector took place in numerous stages. Several grounding-zone wedges and moraines on the mid- and inner continental shelf attest to significant stabilizations of the ice-sheet grounding line, or ice margin, during overall retreat in Greenland Stadials 3 and 2, and to the development of ice shelves. NW Lewis was the first substantial present-day land area to deglaciate, in the first half of Greenland Stadial 3 at a time of globally reduced sea-level c. 26 kabp , followed by Cape Wrath at c. 24 kabp. The topographic confinement of the Minch straits probably promoted ice-shelf development in early Greenland Stadial 2, providing the ice stream with additional support and buffering it somewhat from external drivers. However, c. 20–19 kabp , as the grounding-line migrated into shoreward deepening water, coinciding with a marked change in marine geology and bed strength, the ice stream became unstable. We find that, once underway, grounding-line retreat proceeded in an uninterrupted fashion with the rapid loss of fronting ice shelves – first in the west, then the east troughs – before eventual glacier stabilization at fjord mouths in NW Scotland by ~17 kabp. Around the same time, ~19–17 kabp , ice-sheet lobes readvanced into the East Minch – possibly a glaciological response to the marine-instability-triggered loss of adjacent ice stream (and/or ice shelf) support in the Minch trough. An independent ice cap on Lewis also experienced margin oscillations during mid-Greenland Stadial 2, with an ice-accumulation centre in West Lewis existing into the latter part of Heinrich Stadial 1. Final ice-sheet deglaciation of NW mainland Scotland was punctuated by at least one other coherent readvance at c. 15.5 kabp , before significant ice-mass losses thereafter. At the glacial termination, c. 14.5 kabp , glaciers fed outwash sediment to now-abandoned coastal deltas in NW mainland Scotland around the time of global Meltwater Pulse 1A. Overall, this work on the BIIS NW sector reconstructs a highly dynamic ice-sheet oscillating in extent and volume for much of the last 45 000 years. Periods of expansive ice-sheet glaciation dominated by ice-streaming were interspersed with periods of much more restricted ice-cap or tidewater/fjordic glaciation. Finally, this work indicates that the role of ice streams in ice-sheet evolution is complex but mechanistically important throughout the lifetime of an ice sheet – with ice streams contributing to the regulation of ice-sheet health but also to the acceleration of ice-sheet demise via marine ice-sheet instabilities.     

Multi‐scale stratigraphic forward modelling of the Surat Basin for geological storage of CO2

Underground geological storage of CO₂ (GSC) requires a high level of subsurface understanding that is often hindered by a lack of data. This study demonstrates the use of stratigraphic forward modelling (SFM) in generating and characterising a static reservoir model using limited well data, with multiple potential applications within the GSC workflow. Sedsim SFM software was used to create a static model of the Surat Basin, including a high‐resolution nested model of the EPQ‐7 GSC tenement within the basin. Deposition and burial of the Jurassic Precipice Sandstone, Evergreen Formation and Hutton Sandstone were simulated. Modelling results show a close match with gamma‐ray well logs in the tenement area, and the model can be considered a credible model of the subsurface. The Sedsim‐predicted formation thicknesses and porosity and permeability distributions meet criteria set for GSC, suggesting that the EPQ‐7 tenement may be a prospective GSC location.     

Automated AF-demagnetization on the 2G-Enterprises through-bore,cryogenic magnetometer

Since the early sixties, alternating field demagnetization (AFD) has been a standard laboratory technique for demagnetizing rocks to expose the multicomponent structure of their natural remanent magnetization (NRM). In the majority of AFD implementations, however, the procedure remains as labour-intensive as ever. The implementation that we have developed at the Australian Geological Survey Organisation, automates the procedure for AFD based on the static method, and results in significant productivity and efficiency gains without compromising data quality. A properly formulated procedure for static AFD may be the only method of retrieving higher-coercivity components of natural remanence in samples prone to developing gyroremanence at higher alternating fields (AFs). Our AFD environment comprises: a 2G-Enterprises through-bore, cryogenic magnetometer; 2G AF-coils and control equipment; and personal computer software, developed by us, to control all procedural aspects for a complete AFD of a sample including, importantly, a counteracting procedure to neutralize the effects of gyroremanence build-up at higher AFs. With our system, AFD of 8 samples/day, each of 20+ steps, requires only 20 min of user attention compared with a full day for conventional systems.     

Complex statistics in Hamiltonian barred galaxy models

We use probability density functions (pdfs) of sums of orbit coordinates, over time intervals of the order of one Hubble time, to distinguish weakly from strongly chaotic orbits in a barred galaxy model. We find that, in the weakly chaotic case, quasi-stationary states arise, whose pdfs are well approximated by q-Gaussian functions (with 1 <?q < 3), while strong chaos is identified by pdfs which quickly tend to Gaussians (q =?1). Typical examples of weakly chaotic orbits are those that ??stick?? to islands of ordered motion. Their presence in rotating galaxy models has been investigated thoroughly in recent years due to their ability to support galaxy structures for relatively long time scales. In this paper, we demonstrate, on specific orbits of 2 and 3 degree of freedom barred galaxy models, that the proposed statistical approach can distinguish weakly from strongly chaotic motion accurately and efficiently, especially in cases where Lyapunov exponents and other local dynamic indicators appear to be inconclusive.