The next generation GCM ICON of the Max Planck Institute for Meteorology is now available. Additionally, we are providing a benchmark for one of our key file formats decoding/encoding libraries. If you would like to get access to this benchmarks contact Luis Kornblueh.
The Max Planck Institute for Meteorology has been providing its Earth System Model for RAPS benchmarking in rather informal way. Because of fundamental changes in the models retrieve and build system, we would like to announce this time its availability and recommend to have a look into this new version.
Another change done is related to the availability of oasis3/mct. The new model version is updated to this coupler version. Setup and scaling is supposed to be improved by this change.
In case you are interested in accessing the model, please send an email to Luis Kornblueh.
Papers and extended abstract submitted to our last workshop at SC’12 can now be found here: Submissions of CKD@SC'12.
The submissions along with the workshop slides will also be published as part of the SC’12 post-conference digital proceedings - coming soon!
The emphasis of this workshop will be on running meteorological applications at sustained teraflops performance in a production environment. Particular emphasis will be placed on the future scalability of NWP codes and the tools and development environments to facilitate this as we move towards petaflop computing.
You can find details at http://www.ecmwf.int/newsevents/meetings/workshops/2012/high_performance_computing_15th/index.html.
We will hold the 4th Climate Knowledge Discovery workshop at the Supercomputing 2012, November 12, Salt Lake City. We are calling for participation and papers, the CfP can be found here:
Please join us in Salt Lake City!
Per Nyberg will be presenting an overview of the CKD effort, co-authored with Reinhard Budich and Karsten Steinhaeuser, at the 14th international specialist meeting on the next generation models of climate change and sustainability for advanced high performance computing facilities, March 12-15, Hawaii.
More information available here: http://www.prime-pco.com/climate2012/index.html
Workshop website: http://www.nd.edu/~dial/climkd11
The analysis of climate data, both observed and model-generated, poses a number
of unique challenges: (i) massive quantities of data are available for mining,
(ii) the data is spatially and temporally correlated so the IID assumption does
not apply, (iii) the data-generating processes are known to be non-linear,
(iv) the data is potentially noisy, and extreme events exist within the data.
In the computational data sciences, temporal, spatial and space-time data mining
differ fundamentally from traditional data mining in that the learning samples
are dependent, making auto- and cross-correlations important. Climate data mining
is based on geographic data and inherits the attributes of space-time data mining.
In addition, climate relationships are nonlinear, spatial correlations can be over
long range (teleconnections) and have long memory in time. Thus, in addition to
new or state of the art tools from temporal, spatial and spatio-temporal data
mining, new methods from nonlinear modeling and analysis are motivated along with
analysis of massive data for teleconnections and long-memory dependence.
Climate extremes may be inclusively defined as severe weather events as well as
significant regional changes in hydro-meteorology, which are caused or exacerbated
by climate change, and climate modelers and statisticians struggle to develop
precise projections of such phenomena. The ability to develop predictive insights
about extremes motivates the need to develop indices based on nonlinear
dimensionality reduction and anomaly analysis in space-time processes from massive
data. Knowledge discovery is broadly construed here to include high-performance
data mining of geographically-distributed climate model outputs and observations,
analysis of space-time correlations and teleconnections, geographical analyses of
extremes and their consequences obtained through fusion of heterogeneous climate
and GIS data along with their derivatives, geospatial-temporal uncertainty
quantification, as well as scalable geo-visualization for decision support.
TOPICS OF INTEREST include but are not limited to:
- Methods for mining climate datasets for patterns, trends, or extremes
- Complex networks and climate
- Spatio-temporal data mining
- Mining for rare events or phenomena in climate data
- Algorithms and implementations for the analysis of climate data, including
) Patterns / Clusters
) Extremes / Outliers
) Change Detection
- Methods addressing the role of uncertainty in space-time prediction
- High-performance data mining for the analysis of climate data
- Studies assessing the impacts of climate change and/or extremes
- Applications that demonstrate success stories of knowledge discovery
from climate data
Paper Submission: August 5, 2011
Notification to Authors: September 23, 2011
Camera-Ready Papers: October 11, 2011
Workshop: December 10, 2011
Paper Submission: This is an open call for papers. Only original and high-quality
papers (regular length 8 pages, short papers 6 pages) conforming to the ICDM 2011
guidelines will be considered for this workshop. Papers must be submitted using
the ICDM workshop paper submission system.
Proceedings: Accepted papers will be included in an ICDM Workshop Proceedings volume,
to be published by IEEE Computer Society press, which will also be archived in the
IEEE Digital Library.
Nitesh V Chawla, University of Notre Dame, USA
Auroop R Ganguly, Oak Ridge National Lab, USA
Vipin Kumar, University of Minnesota, USA
Michael Steinbach, University of Minnesota, USA
Karsten Steinhaeuser, University of Minnesota, USA
Read the full interview here:
As was discussed during the conclusion of the April 2011 Climate Knowledge Discovery workshop, a follow-on workshop is being organized in conjunction with the SC11 (Supercomputing 2011) conference in Seattle, U.S. The date for the full-day workshop is Sunday November 13, 2011 -more details here. For those of you not familiar with the SC conferences, these are the largest annual events bringing together the international community in high performance computing, networking, storage and analysis.
This second CKD workshop fits in very well with the SC11 cross-cutting thrust of data-intensive computing. Data intensive science is impacting all fields of study, and SC11 will focus on the challenges and opportunities for addressing the exponential growth and demands in the generation and analysis of data. We fully expect the SC11 CKD workshop to be a milestone in an ongoing series of community events in this area. Proceedings of the SC11 CKD workshop will be published online in association with the April 2011 CKD workshop.
Similar to the April meeting in Hamburg, we propose to hold a workshop that brings together research scientists from diverse disciplines to discuss the design and development of methods and tools for knowledge discovery in climate science. The breakthroughs needed to address CKD challenges will come from collaborative efforts involving several disciplines, including end-user scientists, computer and computational scientists, computing engineers, and mathematicians. The multi-disciplinary nature of SC11 provides a unique opportunity to attract and leverage each of these communities. As a follow-on workshop, the objective will be to build on the outcomes of the first workshop.
The workshop is the second in a series of planned workshops to discuss the design and development of methods and tools for knowledge discovery in climate science.Proposed agenda topics include:
- Science Vision for Advanced Climate Data Analytics
- Application of massive scale data analytics to large-scale distributed interdisciplinary environmental data repositories
- Application of networks and graphs to spatio-temporal climate data, including computational implications
- Application of semantic technologies in climate data information models, including RDF and OWL
- Enabling technologies for massive-scale data analytics, including graph construction, graph algorithms, graph oriented computing and user interfaces.
We hope that you will be able to participate and look forward to seeing you in Seattle. For further information or questions, please do not hesitate to contact any of the co-organizers listed below.
Reinhard Budich - MPI für Meteorologie (MPI-M) - reinhard.budich /a/ zmaw.de
John Feo - Pacific Northwest National Laboratory (PNNL) – john.feo /a/ pnnl.gov
Per Nyberg - Cray Inc. – nyberg /a/ cray.com
Tobias Weigel - Deutsches Klimarechenzentrum GmbH (DKRZ) - weigel /a/ dkrz.de
At the Climate Knowledge Discovery workshop (Mar./Apr. 2011) in Hamburg, considerable interest was expressed on applying graph algorithms to large scientific data sets, such as climate data, on architectures which are tailored for these techniques. The Cray XMT system was mentioned specifically as at target architecture. As part of the new EUREKA data analysis facility at the Swiss National Supercomputing Centre (CSCS), a Cray XMT2 has now been installed, and is now in a pre-production stage. An overview of the EUREKA facility can be found here:
CSCS is open to collaborations which can productively exploit the EUREKA data analysis facility. For technical questions, please contact Will Sawyer (wsawyer at cscs.ch<http://cscs.ch>). For inquiries about access to the XMT2 and other EUREKA systems, please contact Maria-Grazia Giuffreda (mgg at cscs.ch<http://cscs.ch>).
Also available in: Atom