CISL Science

CISL research activities support scientific computation, numerical methods, geophysical modeling, and the analysis of geophysical data and model experiments. These activities are chosen to lead the geophysics community in adopting new computational methods and mathematical tools to improve research. Diverse scientific disciplines often share common tools and numerical methods. The kind of mathematical, computational, and physical sciences housed in CISL focus on general mathematical tools, models, and algorithms that have broad application across NCAR. Hallmarks of this research are innovative and standout contributions that not only have relevance for the overall NCAR scientific program, but also are significant in their specific area of mathematical, physical, or computational science.

Earth's electric system
The Earth’s electric system is an important geophysical between link between solar processes, the upper and lower atmosphere, and cloud system dynamics. In the past however, its role has not been represented in global climate models because they could not resolve the many spatial scales involved in describing the Earth’s currents and electrical fields. Challenges include representing dipole current sources from thunderclouds and the effect of topography at smaller scales than a typical atmospheric model can resolve. Another problem is accounting for the variation of the conductivity of the atmosphere and Earth over many orders of magnitude. This figure illustrates some initial research to model these electrical linkages. In the left panel is the current density derived from a 3D electrostatic model of the lower atmosphere (<70km), using sources from electrified clouds obtained by satellite data (right panel). This physical model is one of the first constructed for these processes and is based on a new numerical method known as radial basis function-generated finite differences that allows for complete geometric flexibility of the domain, yet is computationally efficient due to the sparse structure of the matrices.

The figures shown here illustrate some of the diversity of scientific research in CISL and also suggest some unifying themes. Numerical methods have always been an important component for making geophysical models run more efficiently on evolving computing environments and for handling both resolved and unresolved processes in models. The project highlighted in the first figure is an example of using a new numerical approach (radial basis functions) to model the electrical properties of the atmosphere. Here the emphasis is on a physical model capturing processes at multiple scales.

Urban heat analysis
The color shading depicts an estimate of how the frequency for 911 calls in Houston, Texas depends on the day of year and temperature (with reference contours at 0.1 and 0.05). This statistical surface is estimated from 911 call records, U.S. census demographic information, and daily weather. Note that the increased call intensity is associated with a higher heat index and with days that are closer to the middle of the summer. This suggests a seasonal effect that is partly independent of the temperature. This research is a part of the highly interdisciplinary SIMMER project, featuring scientists from across NCAR, the university community, and public health professionals from the city of Houston and from partners in Canada. By understanding the link between extreme heat and public health endpoints such as 911 calls, this research is aimed at providing public health professionals with important scientific information for planning and mitigation purposes. In this case, for example, this figure highlights the times of year and temperatures they can expect a high volume of heat-related 911 calls to occur.

At other the end of research activities is the analysis of observational data – 911 call records – and the connection between weather and human health. The second example illustrates the use of Bayesian statistical methods to estimate patterns in heat stress for the Houston metropolitan area. Finally, data assimilation is an additional area of research that is pursued in CISL. It combines observations and numerical models to improve estimates of a physical process or to quantify differences between model predictions and observations. All of these projects are challenged by large computational problems and data sets. CISL’s research in computational science supports these projects and additional efforts at NCAR to simulate the Earth’s climate system. Hallmarks of all these areas of research – numeric, data analytics, data assimilation, and computational sciences – are the development of tools and methods that have relevance beyond their specific applications. While CISL research is motiviated by specific scientific problems, it also produces useful outcomes applicable to other areas.

Additional notable accomplishments and examples CISL’s scientific diversity in FY2013 include:

  • A six-month reanalysis was completed that couples the atmosphere, ocean, and land components of the NCAR climate model. The value of assimilating observations into this system is the ability to compare these results to assimilation for each model component alone. The comparison will help to identify significant transfers of energy, momentum, and water within the system. The use of DART in this project is novel and represents a unique capability in climate modeling.

  • The impact of extreme heat events on Houston has been modeled statistically and related to the demographic and urban landscape of the city. This model is the first of its kind to quantify the uncertainty in the relationship of temperature and emergency calls over location and season.

  • A numerical method that supports nonhydrostatic dynamical motion of the atmosphere was successfully implemented in the primary model for the atmosphere used by CESM. Nonhydrostatic models are important for simulating motion at the scale of clouds but require very short spatial scales in the vertical dimension. This approach deals with the differences in the horizontal and vertical scales using a mix of explicit and implicit time stepping. Moreover, the implementation focusing on the flux form of the physical equations facilitates scalability of code and conservation in the simulated processes.

  • An exploration of computational accelerator technology focusing on the numerical core for atmospheric models demonstrated the potential of these micro-architectures. These results are important for evaluating next-generation supercomputing architectures that make heavy use of accelerator technology.

  • A prototype data preprocessing tool has been developed to convert simulation model output fields to time series. This “transpose” processing operation has been a bottleneck, and the prototype based on parallel libraries affords a factor of 30 speedup.

The activities outlined in this report advance the imperatives and frontiers of CISL’s research strategy. Moreover, the breadth of this research aligns with CISL’s strategic imperative to produce scientific excellence. For CISL’s interactions with the scientific community, a robust visitor program and a popular summer internship program provide numerous opportunites for collaboration and for dissemination of these results. Another scientific imperative addressed by CISL research is meeting the challenges of Earth System modeling as it moves to the petascale and exascale. Thus, research is pursued on adaptive numerical methods and multiscale models, both areas addressing the need for higher-resolution geophysical simulations. A companion to this effort is work in computational science for taking advantage of massively parallel supercomputers and new kinds of processors. Set against these computational goals is the creation of new analysis tools to interpret complex multifactor geophysical simulations and heterogeneous observational data. These are addressed in CISL research by focusing on the impacts of regional climate change and data assimilation.

This work is funded as specified in the following individual reports.