AMS Short Course on Big Data, Distributed systems Analyses: UV-CDAT
Sunday, 6 January 2013
Climate scientists have made substantial progress in understanding the Earth's climate system, particularly at global and continental scales. Climate research is now focused on understanding climate changes over wider ranges of time and space scales. These efforts are generating ultra-scale datasets at very high spatial resolution. An insightful analysis in climate science depends on using software tools to discover, access, manipulate, and visualize the data sets of interest. These data exploration tasks can be complex and time-consuming, and they frequently involve many resources from both the modeling and observational climate communities. Because of the complexity of the explorations, the number of tools, and the amount of data involved, provenance is critical, allowing scientists to ensure reproducibility, revisit existing computational pipelines, and more easily share analyses and results. In addition, as the results of this work can impact policy, having provenance available is important for decision-making. In
this class, we will cover UV-CDAT, a workflow-based, provenance-enabled system that integrates climate data analysis libraries and visualization tools in an end-to-end application, making it easier for scientists to integrate and use a wide array of tools.
This course is targeted for an advanced audience of Python users and will assume prior knowledge of the basics of Python (looping patterns, function definition, NumPy arrays, object oriented programming). Knowledge of CDAT, ParaView and/or VisIt is also a plus. Students with no prior knowledge of Python are encouraged to sign-up for the Beginner's Python course; those with some knowledge should sign-up for the Intermediate Python course.
All attendees will need to bring a laptop (with power adapter) that has Python installed on it. Instructions will be emailed to registered attendees before the course begins on the basic requirements. The class will be taught using a Virtual Machine (VM) with all necessary materials and software pre-loaded.
The course will cover most features of the new Ultra-scale Visualization Climate Data Analysis Tools (UV-CDAT). Concept covered by the class are: Accessing distributed Dataset, Analyzing Big Distributed Data, Provenance and why it matters, novel 3D visualizations, building custom plugins for UV-CDAT.
Most of the course will be geared around discovering and using data for the Fifth Annual Report of the Intergovernmental Panel on Climate Change (IPCC AR5)
The course format consists of lectures mixed with hands-on exercises. Strong user participation will be encouraged that could reshape the flow of the course “on-the-fly”. The main instructor for the course is Charles S. Doutriaux, Program for Climate Model diagnosis and Intercomparison at the Lawrence Livermore National Laboratory, CA.
Users are expected to bring their own laptop. AC and internet access will be available. Note that this class is for users with a good knowledge of CDAT/Python. Sister classes, aimed at beginners, are also scheduled during the meeting.
For more information please contact Charles Doutriaux at PCMDI, 7000 East Avenue, Livermore, CA, 94450, USA. (tel: (925) 422-1487; email: firstname.lastname@example.org.