Climate for the Masses

Understanding climate change is important, and therefore has been discussed in a large variety of media over the last several years. Changes in climate resonate though a broad range of fields, including public health, infrastructure, water resources, and many others. Over the last few years, climate simulations have produced a wealth of climate simulation data, including output from high-resolution, long-term, climate change projections performed as part of the United States Global Change Research Program [1] and the Coupled Model Intercomparison Project [2]. Widespread use of climate data is impeded by the lack of easy-to-use, effective analysis and visualization tools for policy makers, researchers (from non-climate communities), and other possible users of climate data. Thus, there is a need for communicating climate data to all age, social, cultural, and professional groups.

Phase I Achievements
Visualization is visual communication for the purpose of the presentation and exploration of data, concepts, relationships and processes; it presents powerful capabilities that can yield useful insights into climate data. Kitware, in collaboration with NYU-Poly, has leveraged ParaViewWeb [3], VisTrails [4], Ultrascale Visualization – Climate Data Analysis Tools [5] (see Figure 1), Earth System Grid Federation [6], and the latest web technologies to provide a prototype of our state-of-the-art web tool, ClimatePipes, to climate data users.

 

Figure 1: Various 2D and 3D plots in UV-CDAT

 

 Figure 2:  Dataflow to list relevant data from ESGF

 

ClimatePipes enables users to access, query and visualize climate datasets from multiple sources. For the Phase I of the project, ClimatePipes provided a simple, easy-to-use form-based interface, and a more advanced dataflow interface to construct more complex workflows. Figure 2 shows an example dataflow created using the pipeline interface to fetch a list of datasets available from an ESGF source.

The simpler interface (see Figure 3) enabled users to select a region of interest, select a data source; set spatial and temporal filters, and desired output; and view a list of matched data items or visualization of closest-matched climate data.

 

Figure 3: A simple, form-based interface for  Phase I ClimatePipes


When executed, this interface automatically generated the corresponding dataflow, which could then be modified to construct more complex workflows. Figure 4 shows an advanced use case of ClimatePipes.

 

Figure 4: An advanced use case of ClimatePipes that involves regridding one dataset to match the other, computing the overlapping time series, computing the difference between the two, and displaying it visually
with a legend.

 

For the  Phase I system, ClimatePipes used an ad-hoc algorithm, which uses Natural Language Toolkit (NLTK) to generate improved results in response to a search query.

Phase II Vision
Recently, ClimatePipes received a notice of Phase II award from the DOE to transform the ClimatePipes prototype into an open-source production quality tool, which could be used by researchers or non-researchers to understand long-term climate change projections. In Phase II of the project, ClimatePipes will enable users to run complex computations, and generate effective visualizations in the cloud or on a user-provided cluster.  Users will be able to upload data or visualizations to a server for analysis and integration. Figure 4 provides an overview of the Phase II design.

 


Figure 5: ClimatePipes system overview. Climate data along with data from other domains is used by the ClimatePipes Server for computations that leverage HPC resources and generate analyses, visualizations, and output data. Provenance and other metadata are stored in the ClimatePipes database.


The key objectives of the Phase II of ClimatePipes are as follows:

Enable users to explore real-world questions related to climate change.
ClimatePipes will enable users to find answers to real-world questions regarding climate change. Using semantic search on climate datasets will enable ClimatePipes to provide relevant results to user queries.

Provide tools for data access, analysis, and visualizations.
ClimatePipes will enable users to create state-of-the-art info-views, which will provide the ability to combine data analysis and visualization into a single view. Drawing from a variety of visual results including 2D charts, geographic plots, and 3D visualizations, ClimatePipes will enables users to not only access but also explore climate data efficiently and effectively.

Facilitate collaboration by enabling users to share datasets, workflows, and visualizations.
ClimatePipes will provide the means to share datasets, workflows, and visualizations between users of communities. Similar to other social networks, users will be empowered to create groups of communities. ClimatePipes will also enable users to annotate data with specific tags and notes.

Conclusion
ClimatePipes is a community-supported tool, and as such, will foster collaboration between various commercial, non-profit, and governmental groups. Different agencies and companies publish datasets and tools that are often difficult to reconcile with other data or software, but ClimatePipes provides an extensible infrastructure to integrate them. The tool will also allow users to not only browse and visualize datasets in a Web browser, but also to download their data and analyses to work offline with their own tools of choice. Additionally, it will track the provenance of such analyses so the entire process can be reviewed and shared by other researchers, agencies, and the public.

This work has been supported by the Department of Energy (DOE) under award number DE-SC0006493.

References:
[1] United States Global Change Research Program, http://www.globalchange.gov
[2]  Taylor, Karl E, Ronald J Stouffer, and Gerald A Meehl. “An Overview of CMIP5 and the Experiment Design.” Bulletin of the American Meteorological Society 93.4 (2012) : 485-498.
[3]  Jourdain S., Ayachit U., Geveci B. “ParaViewWeb, A web framework for 3D Visualization and Data Processing”. 2010.
[4]   Bavoil, L et al. “VisTrails: enabling interactive multiple-view visualizations.” Ed. C T Silva, E Croller, & H Rushmeier. VIS 05 IEEE Visualization 2005 2005 135-142.
[5]    UV-CDAT, http://uv-cdat.llnl.gov
[6]    ESGF, http://esgf.org

Aashish Chaudhary is an R&D Engineer on the Scientific Computing team at Kitware. Prior to joining Kitware, he developed  a graphics engine and open-source tools for information and geo-visualization. Some of his interests are software engineering, rendering, and visualization

Questions or comments are always welcome!