PolarGrid

1. PolarGrid Open Grid Forum OGF21, Seattle Washington October 16 2007 Geoffrey Fox Computer Science, Informatics, Physics Pervasive Technology Laboratories Indiana University Bloomington IN 47401 http://grids.ucs.indiana.edu/ptliupages/presentations/ http://www.polargrid.org/ http://cerser.ecsu.edu/ https://www.cresis.ku.edu/index.php gcf@indiana.eduhttp://www.infomall.org

3. Cyberinfrastructure Center for Polar Science To address the Cyberinfrastructure challenges presented immediately by the Center for Remote Sensing of Ice Sheets (CReSIS) and the polar science community in general, the Cyberinfrastructure Center for Polar Science (CICPS), with experts in Polar Science, Remote Sensing and Cyberinfrastructure, has been established. This center includes the University of Kansas (KU), the lead CReSIS institution; Indiana University (IU), which is internationally known for its broad expertise in research and infrastructure for eScience; and Elizabeth City State University (ECSU), a founding member of CReSIS with a center of excellence in remote sensing.

4. ECSU When ECSU first began operation on January 4, 1892, with 23 students, it did so in rented quarters with a budget of $900.  In the fall 2005, the university made history with its highest enrollment ever of 2,664 students.  As of May 2006, 15,663 students have been conferred with undergraduate and graduate degrees.

5. Climate Change and Ice Cores Climate Change Source: IPCC Third Assessment Report Alley,2004 Hansen, 2004

6. Greenland • Changes in the Velocity Structure of the Greenland Ice Sheet • Eric Rignot and Pannir Kanagaratnam • Science 17 February 2006: Vol. 311. no. 5763, pp. 986 – 990 • Extensive press coverage

7. Jacobshavn • Greenland’s mass loss doubled in the last decade: • 0.23 ± 0.08 mm slr / yr in 1996 • 0.57 ± 0.1 mm slr / yr in 2005 • 2/3 of the loss is caused by ice dynamics • 1/3 is due to enhanced runoff Rignot and Kanagaratnam, Science (2006) Jakobshavns Discharge: 24 km3 / yr (5.6 mile3 / yr) in 1996 46 km3 / yr (10.8 mile3 / yr)in 2005

8. Global Warming …… In its summary report to policy makers, IPCC identified ice-sheet response to the projected global warming as a major unknown in assessing future sea level rise. The models used by IPCC to make sea level rise projections do not include dynamical processes related to ice flow and cannot explain rapid changes being observed. In fact, IPCC explicitly stated in its report that understanding of these processes is limited, there is no consensus on their magnitude (IPCC 2007), and there is an urgent need to improve the models. Improving numerical ice-sheet models and constraining future evolution of the Greenland and Antarctic ice sheets requires better understanding of the processes causing the rapid changes, which is a major goal of the CReSIS science program.

9. Glacial Change Data While there is considerable evidence now that fast glacial changes occur and have occurred in the past, there is no consensus yet regarding how this takes place. Better understanding will only come from an ambitious campaign to collect more data, and develop new and better models using such data.

10. CReSIS CReSIS science research presents a substantial challenge for PolarGrid. The traditional view regarding the relationship of polar ice sheets and climate has been that polar ice sheets respond slowly to changes in climate. Most existing models, including those used by the Intergovernmental Panel on Climate Change (IPCC), predict only a small contribution from Greenland and Antarctica to global sea level over the next century.

11. Synthetic Aperture Radar (SAR) Synthetic Aperture Radar (SAR) imaging of the ice-bed interface has the potential to make a significant contribution to glaciology by providing insights regarding the conditions at this critical boundary, including terrain relief and roughness as well as presence and distribution of liquid water. However, SAR image formation is computationally intensive. While the raw radar data, called phase histories, can be collected and stored for post-processing analysis, the image formation process requires a significant amount of signal processing, regardless of whether it is done in real-time or in post-processing. SAR imaging of the ice-bed interface through 3-km thick ice may require iterative focusing to compensate for the unknown variability in the electrical properties of the intervening ice.

12. Data Collection Flights Initial CReSIS SAR data collection flights may produce in excess of 6 million complex samples per second per receiver channel. A single 5-hour flight mission could collect as much as one terabyte of raw SAR data. CReSIS field programs are designed to perform geophysical measurements in the transition region (from deformation to streaming flow) of areas in Greenland and Antarctica undergoing rapid changes. The real-time field processing will allow the field team to adjust location and flight patterns to collect optimum data sets. Thus, access to a state-of-the-art computing facility that can be used to process data to generate quick-look images will allow the team to collect optimum SAR data. .

13. Current UAV Design Concept Aircraft Summary: – WTO = 1,083 lbs – WE = 618 lbs – WF = 295 lbs – WPL = 165 lbs – Wingspan = 26.4 ft – Length = 17 ft – Range = 1,750 km (950 nm) – Endurance = 13 hours The most critical design requirements are: – Payload Integration (Antennae Size) – Takeoff/Landing Distance – Size limitations • Shipping • Hangar Size – Fuel Type – Cold Weather Requirements (Anti-Icing)

14. CICPS Projects CICPS is committed to the effort needed to build the portal, workflow and Grid (Web) services that are required to make PolarGrid real. Two CICPS projects are being implemented and proposed in support of this effort. • Cyberinfrastructure for Remote Sensing of Ice Sheets • PolarGrid

16. PolarGrid • Important PolarGrid-specific Cyberinfrastructure components include • Managed data from sensors and satellites • Data analysis such as SAR processing – possibly with parallel algorithms • Electromagnetic simulations (currently commercial codes) to design instrument antennas • 3D simulations of ice-sheets (glaciers) with non-uniform meshes • GIS Geographical Information Systems • Also need capabilities present in many Grids • Portal i.e. Science Gateway • Submitting multiple sequential or parallel jobs • The need for three distinct types of components: Continental USA with multiple base and field camps • Base and field camps must be power efficient • Terrible connectivity from base and field camps to Continental subGrid

17. Activities • Need to develop • 1) Real time SAR • 2)  Modeling • 3) Data fusion into products (second and third year) • 4) Science Gateway using Web 2.0 • Current SAR uses Matlab. Need efficient multicore parallel algorithm. • Training is essential (ECSU) • Expeditions • May--June 2008 Greenland and • November 2008  - February 2009 Antarctica

18. ECSU Training • Wed Aug 29 3:00pm: Overview: What the Heck is Supercomputing? • Wed Sep 5 3:00pm: The Tyranny of the Storage Hierarchy • Wed Sep 12 3:00pm: Instruction Level Parallelism • Wed Sep 19 3:00pm: Stupid Compiler Tricks • Wed Sep 26 3:00pm: Shared Memory Multithreading • Wed Oct 10 3:00pm: Distributed Multiprocessing • Wed Oct 17 3:00pm: Multicore Madness • Wed Oct 24 3:00pm: High Throughput Computing • Wed Oct 31 3:00pm: Grab Bag: Scientific Libraries, I/O, Visualization

19. Basic Hardware • Initially Build 1 Base and Field camps • Send out vendor-neutral RFP for Indiana University and ECSU systems Data Centers • Also 60 workstations and Polycom for ECSU Training facilities

20. Data Rates • Initial Field Station about 500GB per day • Distance of 100-300 Km between base and field camps and the remoteness of the base camp from the data center (3000-6000 Km) • Primary data will be stored on-site and returned to the data center by air on removable media on a regular schedule. • Continuous low-bandwidth connection to the continental US allows users can see what data has been acquired from the field and can potentially request that ad hoc analyses be run on data sets in the field to answer questions long before the primary data are available at the CICPS data center.

21. Cyberinfrastructure for Remote Sensing of Ice Sheets The first of these projects is an NSF CI-TEAM project (PI: Hayden, Co-PIs: Fox and Gogineni), “Cyberinfrastructure for Remote Sensing of Ice Sheets,” which establishes a virtual classroom environment and a CReSIS Science Gateway for TeraGrid working with IU, Minority-Serving Institution Cyberinfrastructure Empowerment Coalition (MSI-CIEC), and TeraGrid.

22. ECSU Grid ECSU will establish a minimal Grid configuration and a virtual classroom environment with broadcast and receiving capabilities. With regards to the Grid, access will be implemented through web and research portals for use of computational and data resources. The portals will be customized with educational and science interfaces while still allowing access to large amounts of data.

23. Vision and Goals The vision and goals of the NSF CI-TEAM project, “Cyberinfrastructure for Remote Sensing of Ice Sheets,” are based on the fact that “educational settings, audiences, and goals are too important to be adequately addressed as afterthoughts or add-ons to Cyberinfrastructure projects and, instead, must be treated as high priorities integrated in a project’s overall design”

24. Components The virtual classroom configuration consists of a Polycom 8000S system with 14-kHz wideband audio, 2 channels of 14-kHz audio, stand-alone audio conferencing phone, IP (H.232 and SIP) interfaces and Global Management System collaborative communications. ECSU will establish a Grid through PolarGrid grant. This will have an classroom and an analysis component ECSU Team has a brand new building with a PolarGrid lab 2 New Faculty Master’s Degree in applied math with a concentration in remote sensing.

25. ADMI The Association of Computer and Information Science/Engineering Departments at Minority Institutions (ADMI) was founded in August 1989. (http://cerser.ecsu.edu/admi2006/purp_gls.html) Represented on the Board of ADMI are Spelman College, Mississippi Valley State University, North Carolina Central University, ECSU, Howard University, Jackson State University, University of the District of Columbia, Hampton University, Fisk University and Florida A&M University.

26. ADMI Involvement ADMI’s role in the proposed ECSU Grid project will be to establish capability within their institutions to allow students and faculty to participate in the virtual workshops, training and courses. In addition, students from the ADMI institutions will participate in an 8-week Grid technology workshop to be conducted during the summers at ECSU.

27. NSF CI-Team With particular attention to the current and next generation of traditionally underrepresented minority scientists engineers and educators, the NSF CI-TEAM project prepares students with the knowledge and skills necessary to conduct interdisciplinary research in areas including Cyberinfrastructure, remote sensing, engineering and modeling related to glaciers and ice sheets. • Summer Research Training in GRID and CReSIS related science • Academic year project, workshops and seminars