Skip Navigation
Return to Layout View | Home | A-Z Directory | my UMaine | MaineStreet | Campus Map | Calendar
Follow UMaine on Twitter | Join UMaine on Facebook | Watch UMaine on YouTube | Admissions | Parents & Family | Emergency

School of Computing and Information Science


Site Navigation:


Increasing the Scalability of PISM for High Resolution Ice Sheet Models

P. Dickens and T. Morey (2013), Increasing the Scalability of PISM for High Resolution Ice Sheet Models14th IEEE International Workshop on Parallel and Distributed Scientific and Engineering Computing (PDSEC-14), Boston, MA.

Abstract

The issue of global climate change is of great interest to scientist and a critical concern of society at large. One important piece of the climate puzzle is how the dynamics of large-scale ice sheets, such as those in Greenland and Antarctic, will react in response to such climate change. Domain scientists have developed several simulation models to predict and understand the behavior of large-scale ice sheets, but the depth of knowledge gained from such models is largely dependent upon the resolution at which they can be efficiently executed. The problem, however, is that relatively small increases in the resolution of the model result in very large increases in the size of the input and output data sets, and an explosion in the number of grid points that must be considered by the simulation. Thus increasing the resolution of ice-sheet models, in general, requires the use of supercomputing technologies and the application of tools and techniques developed within the high-performance computing research community. In this paper, we discuss our work in evaluating and increasing the performance of the Parallel Ice Sheet Model  (PISM) [6, 25, 38], using a high-resolution model of the Greenland ice sheet, on a state-of-the-art supercomputer. In particular, we found that the computation performed by PISM was highly scalable, but that he I/O demands of the higher-resolution model were a significant drag on overall performance. We then performed a series of experiments to determine the cause of the relatively poor I/O performance and how such performance could be improved. By making simple changes to the PISM source code and one of the I/O libraries used by PISM we were able to provide an 8-fold increase in I/O performance.


LocalAlert: Simulating decentralized ad-hoc collaboration in emergency situations

S. Nittel, C. Dorr, and J.C. Whittier, 2012, LocalAlert: Simulating decentralized ad-hoc collaboration in emergency situations, in: N.Xioa, M.P. Kwan, M. Goodchild, and S. Shekhar (eds.), Geographic Information Science — Seventh International Conference, GIScience 2012, Columbus, OH, Lecture Notes in Computer Science, Vol. 7478, Springer, pp. 146-159.

 

Abstract

Today, advances in short-range ad-hoc communication and mobile phone technologies allow people to engage in ad-hoc collaborations based solely on their spatial proximity. These technologies can also be useful to enable a form of timely, self-organizing emergency response. Information about emergency events such as a fire, an accident or a toxic spill is most relevant to the people located nearby, and these people are likely also the first ones to encounter such emergencies. In this paper we explore the concept of decentralized ad-hoc collaboration across a range of emergency scenarios, its feasibility, and potentially effective communication protocols. We introduce the LocalAlert framework, an open source agent simulation framework that we have developed to build and test various form s of decentralized ad-hoc collaboration in different emergency situations. Initial experiments identify a number of parameters that affect the likelihood of a successful response under such scenarios.

 

Spatial reasoning with augmented points: Extending cardinal directions with local distances

R. Moratz and J.O. Wallgrün, 2012, Spatial reasoning with augmented points: Extending cardinal directions with local distances, Journal of Spatial Information Science, Vol. 5.

Abstract

We present an approach for supplying existing qualitative direction calculi with a distance component to support fully fledged positional reasoning. The general underlying idea of augmenting points with local reference properties has already been applied in the OPRAm calculus. In this existing calculus, point objects are attached with a local reference direction to obtain oriented points and able to express relative direction using binary relations. We show how this approach can be extended to attach a granular distance concept to direction calculi such as the cardinal direction calculus or adjustable granularity calculi such as OPRAm or the Star calculus. We focus on the cardinal direction calculus and extend it to a multi-granular positional calculus called EPRAm. We provide a formal specification of EPRAm including a composition table for EPRA2 automatically determined using real algebraic geometry. We also report on an experimental performance analysis of EPRA2 in the context of a topological map-learning task proposed for benchmarking qualitative calculi. Our results confirm that our approach of adding a relative distance component to existing calculi improves the performance in realistic tasks when using algebraic closure for consistency checking.

Qualitative Reasoning about Relative Direction of Oriented Points

T. Mossakowski and R.  Moratz, 2012, Qualitative Reasoning about Relative Direction of Oriented Points. Artificial Intelligence Journal, Volumes 180-181, pages 34-45.

 

Abstract

 

An important issue in qualitative spatial reasoning is the representation of relative directions. In this paper we present simple geometric rules that enable reasoning about the relative direction between oriented points. This framework, the oriented point algebra OPRAm, has a scalable granularity m. We develop a simple algorithm for computing the OPRAm composition tables and prove its correctness. Using a composition table, algebraic closure for a set of OPRAm statements is very useful to solve spatial navigation tasks. It turns out that scalable granularity is useful in these navigation tasks.

 

Spatial working memory for locations specified by vision and audition: testing the amodality hypothesis

J.M. Loomis , R.L. Klatzky, B. McHugh, and N.A. Giudice, 2012, Spatial working memory for locations specified by vision and audition: testing the amodality hypothesis, Attention, Perception, & Psychophysics 74(6):1260-7.

Abstract
Spatial working memory can maintain representations from vision, hearing, and touch, representations referred to here as spatial images. The present experiment addressed whether spatial images from vision and hearing that are simultaneously present within working memory retain modality-specific tags or are amodal. Observers were presented with short sequences of targets varying in angular direction, with the targets in a given sequence being all auditory, all visual, or a sequential mixture of the two. On two thirds of the trials, one of the locations was repeated, and observers had to respond as quickly as possible when detecting this repetition. Ancillary detection and localization tasks confirmed that the visual and auditory targets were perceptually comparable. Response latencies in the working memory task showed small but reliable costs in performance on trials involving a sequential mixture of auditory and visual targets, as compared with trials of pure vision or pure audition. These deficits were statistically reliable only for trials on which the modalities of the matching location switched from the penultimate to the final target in the sequence, indicating a switching cost. The switching cost for the pair in immediate succession means that the spatial images representing the target locations retain features of the visual or auditory representations from which they were derived. However, there was no reliable evidence of a performance cost for mixed modalities in the matching pair when the second of the two did not immediately follow the first, suggesting that more enduring spatial images in working memory may be amodal.

Mid-latitude Amazonian glaciation on Mars

J L Fastook and J W Head, 2012, “>Mid-latitude Amazonian glaciation on Mars: Controls on accumulation and glacial flow patterns, Lunar and Planetary Science Conference XXXXIII, Houston TX.

Abstract

A model of flow into a crater from episodic layering events driven by obliquity demonstrates that even at Amazonian temperatures, concentric crater fill can be formed in as little as 50 m.y., whereas flow from a single persistent layer takes 450 m.y.

Early Mars climate near the Noachian–Hesperian boundary

J L Fastook, J W Head, D R Marchant, F Forget, and J-B Madeleine, 2012, Early Mars climate near the Noachian–Hesperian boundary: Independent evidence for cold conditions from basal melting of the south polar ice sheet (Dorsa Argentea Formation) and implications for valley network formation, Icarus 219(1):  25-40.

 

Currently, and throughout much of the Amazonian, the mean annual surface temperatures of Mars are so cold that basal melting does not occur in ice sheets and glaciers and they are cold-based. The documented evidence for extensive and well-developed eskers (sediment-filled former sub-glacial meltwater channels) in the south circumpolar Dorsa Argentea Formation is an indication that basal melting and wetbased glaciation occurred at the South Pole near the Noachian–Hesperian boundary. We employ glacial accumulation and ice-flow models to distinguish between basal melting from bottom-up heat sources (elevated geothermal fluxes) and top-down induced basal melting (elevated atmospheric temperatures warming the ice). We show that under mean annual south polar atmospheric temperatures (100 C) simulated in typical Amazonian climate experiments and typical Noachian–Hesperian geothermal heat fluxes (45–65 mW/m2), south polar ice accumulations remain cold-based. In order to produce significant basal melting with these typical geothermal heat fluxes, the mean annual south polar atmospheric temperatures must be raised from today’s temperature at the surface (100 C) to the range of 50 to 75 C. This mean annual polar surface atmospheric temperature range implies lower latitude mean annual temperatures that are likely to be below the melting point of water, and thus does not favor a “warm and wet” early Mars. Seasonal temperatures at lower latitudes, however, could range above the melting point of water, perhaps explaining the concurrent development of valley networks and open basin lakes in these areas. This treatment provides an independent estimate of the polar (and non-polar) surface temperatures near the Noachian–Hesperian boundary of Mars history and implies a cold and relatively dry Mars climate, similar to the Antarctic Dry Valleys, where seasonal melting forms transient streams and permanent ice-covered lakes in an otherwise hyperarid, hypothermal climate.

Full Paper

An Ordering of Convex Topological Relations

M. Dube and M. Egenhofer, An Ordering of Convex Topological Relations, in: N.Xioa, M.P. Kwan, M. Goodchild, and S. Shekhar (eds.), Geographic Information Science — Seventh International Conference, GIScience 2012, Columbus, OH, Lecture Notes in Computer Science, Vol. 7478, Springer, pp. 72-86, September 2012.

Abstract

Topological relativity is a concept of interest in geographic information theory. One way of assessing the importance of topology in spatial reasoning is to analyze commonplace terms from natural language relative to conceptual neighborhood graphs, the alignment structures of choice for topological relations. Sixteen English-language spatial prepositions for regionregion relations were analyzed for their corresponding topological relations, each of which was found to represent a convex subset within the conceptual neighborhood graph of the region-region relations, giving rise to the construction of the convex ordering of region-region relations. The resulting lattice of the convex subgraphs enables an algorithmic approach to explaining unknown prepositions.

 

Analyzing Spatial and Temporal Radon-222 Trends in Maine

C. Farah, K, Beard, C.T. Hess, and J. Hock, 2012, Analyzing Spatial and Temporal Radon-222 Trends in Maine. Health Physics. 102(2) 115-123.

Abstract

Prolonged radon exposure has been linked to lung cancer. Cancer registry data indicates excess risk for age-adjusted lung cancer in Maine. Maine’s mean residential radon activity exceeds the EPA maximum contaminant level (MCL). This paper describes the application of spatial autocorrelation methods to retrospective data as a means of analyzing radon activity in Maine. Retrospective air and well water radon activity data, sampled throughout Maine between 1993 and 2008, are standardized and geocoded for analysis. Three spatial autocorrelation algorithms-local Getis-Ord, local Moran, and spatial scan statistic-are used to identify spatial, temporal, and spatiotemporal radon activity clusters and/or outliers. Spatial clusters of high air- and well water-Rn activity are associated with Maine’s Lucerne and Sebago granitic formations. Spatial clusters of low air- and well water-Rn activity are associated with Biddeford Granite and the metamorphic bedrock formation Silurian Ordovician Vassalboro. Space-time analysis indicates that most spatial clusters persist over the period of sampling. No significant temporal clusters are identified. Persistent spatial variations in radon may help to better understand and predict radon-related health risks associated with Maine residences.

Efficient Data Collection and Event Boundary Detection in Wireless Sensor Networks Using Tiny Models

K. King and S. Nittel, 2010, Efficient Data Collection and Event Boundary Detection in Wireless Sensor Networks Using Tiny Models, International Conference on Geographic Information Science (GIScience), Zurich, Switzerland, Springer LNCS.

Using wireless geosensor networks (WGSN), sensor nodes often monitor a phenomenon that is both continuous in time and space. However, sensor nodes take discrete samples, and an analytical framework inside or outside the WSN is used to analyze the phenomenon.. In both cases, expensive communication is used to stream a large number of data samples to other nodes and to the base station. In this work, we explore a novel alternative that utilizes predictive process knowledge of the observed phenomena to minimize upstream communication. Often, observed phenomena adhere to a process with predictable behavior over time.We present a strategy for developing and running so-called ‘tiny models’ on individual sensor nodes that capture the predictable behavior of the phenomenon; nodes now only communicate when unexpected events are observed. Using multiple simulations, we demonstrate that a significant percentage of messages can be reduced during data collection.


Sidebar


Contact Information

School of Computing and Information Science
348 Boardman Hall
Orono, ME 04469
Phone: (207) 581-2188 | Fax: (207) 581-2206E-mail: kkidder@spatial.maine.edu
The University of Maine
Orono, Maine 04469
207.581.1865