Foster research and technical collaborations

CISL’s research efforts are enhanced by a robust set of ongoing partnerships, workshops, and training activities. These partnerships take the form of joint appointments with other NCAR laboratories and long-term research and development projects that include resource investments by industry, vendor, university, and research laboratory partners. Recurring workshops, symposia, hackathons, and other training events foster community understanding and use of emerging tools, technologies, and techniques developed through CISL research activities.

Joint appointments

NCAR joint appointments are positions that are funded by two or more NCAR laboratories on a semi-permanent basis under a formal agreement that is reviewed and approved by the NCAR Directorate. One example is a long-standing project scientist joint appointment between CISL and the Mesoscale and Microscale Meteorology Laboratory using ensemble data assimilation techniques to understand the behavior and predictability of severe thunderstorms and other high-impact weather phenomena.

Two other project scientist positions shared with the Research Applications Laboratory (RAL) are devoted to machine learning and form the backbone of the Analytics and Integrative Machine Learning (AIML) group in CISL. These two joint appointments have served to strengthen ties between RAL and CISL and have helped create an NCAR-wide collaborative research network focused on applied research and stakeholder-driven applications. Similarly, two new appointments between CISL and the Climate and Global Dynamics (CGD) Laboratory are being implemented. The first position, filled by a new hire to start early in FY2020, enhances collaboration between the labs in the area of Earth system prediction by sharing scientific expertise. The second is an existing project scientist who will provide shared data workflow expertise between CISL and CGD, thus formalizing a long-standing, co-funded project scientist position in that area.

University workforce development partnerships

CISL works to develop the future workforce at partner universities by collaboratively training computer science and electrical and computer engineering students in the porting and optimization of GPU-enabled model components. This experience provides students with employable HPC skills and helps fill a growing need for experts in optimization and porting of scientific applications to GPUs. Students can be graduate or undergraduate level. Their work is performed as part of a paid research position and typically provides partial credit toward their degree.

In 2019, students at the University of Wyoming and the University of Delaware worked on porting and optimizing the MPAS-A global atmospheric model and the MURaM solar physics model for GPUs. Over the past three years, a total of eight university students have been trained in this way, and five have obtained HPC-related employment in industry or at U.S. national laboratories as a result of their participation. The work performed by these students has been of excellent quality and has helped NCAR provide GPU-CPU portable models, which benefits the entire scientific user community. Both universities we have engaged are in Experimental Program to Stimulate Competitive Research (EPSCoR) states – thus helping to advance the research competitiveness of these states.

The work of the students on NCAR projects at partner universities was performed under subawards with NSF Core funds.

Vendor R&D partnerships

CISL enters into strategic research and development partnerships in high-performance computing where appropriate. These activities enable CISL to ensure that NCAR applications, tools, and data sets are able to exploit the latest technologies. This work is done on an open-source basis that furthers progress on the scientific goals of the NCAR and university scientific communities. Following are examples of these partnership activities.

NVIDIA Corporation

The Weather and Climate Alliance (WACA) is a CISL partnership with NVIDIA Corporation. WACA, which focused initially on GPU application acceleration, continued its focus on machine-learning emulation. To this end, NVIDIA supplied $75,000 in mid-2018 to fund CISL activities in the AIML group. These funds fostered rapid progress on three machine-learning emulation prototypes in 2019: predicting the magnetic field strength of coronal mass ejections in Earth’s geospace, emulating the atmospheric surface layer from tower observations, and emulating autoconversion processes in cloud microphysics.

Joint development agreement with IBM/The Weather Company

An ongoing joint-development agreement with IBM Corporation and its subsidiary, The Weather Company (TWC), seeks to create an operational, GPU-enabled weather prediction system based on the MPAS atmospheric model. In FY2019 NCAR neared completion of a GPU-enabled version of the model for use in global weather predictions by TWC on the IBM Power Server architecture as part of the Global High-Resolution Atmospheric Forecast System. The joint development activity provided $100,000 of funding to CISL in FY2019 and was also supported with co-sponsored staff time from NSF Core funds.

Amazon collaboration

The Amazon Web Services (AWS) Public Dataset Program and the Amazon Sustainability Data Initiative provided ~100 TB of free data hosting and egress in AWS Simple Storage Service (S3) in FY2019. This allocation has been used to host a cloud-optimized version of the CESM LENS data set.

Training activities

Training activities help the community better utilize the resources we provide in order to further the accomplishment of their research goals. These training activities were supported by NSF Core funds.

Python training

In June of 2019, CISL presented a three-day Python tutorial and hackathon in collaboration with Climate and Global Dynamics Lab staff and Unidata personnel. The event focused primarily on teaching NCAR scientists and visitors how to work with geoscientific data in Python. The tutorial/hackathon had formal lecture segments and time for participants to work on personal projects. Sixty-six people registered for the event.

Based on feedback from the June event, a second Python tutorial in September focused more on Python basics for beginners. It drew 74 registrants, all but 15 affiliated with NCAR and UCAR. 

Fortran training 

Fortran remains the dominant scientific programming language, especially for large-scale model development. To support Fortran programming competency in our user base, CISL conducted a series of Fortran workshops in September 2019. It consisted of lectures that covered:

  • Scalar Fortran

  • Vector Fortran

  • Object-Oriented Fortran

  • Parallel Fortran

  • Interoperability with C

  • IEEE 754 Standards for Floating Point Arithmetic

There were 20 attendees aggregated over five lectures, half of them from within NCAR and half from outside institutions.

Cloud-based CESM user training 

CISL and CGD teamed up to provide two tutorials for new users of the Community Earth System Model (CESM). The training used commercial cloud services, which enabled presenters to have full control over the environment and resulted in easier access and a more uniform and productive experience for the trainees. More than 30 users attended the tutorial during the American Geophysical Union Conference in December 2018, and more than 10 new CESM users attended during the American Meteorological Society annual meeting in January 2019. Amazon Web Services provided the cloud resources in each session at moderate cost. This approach will also be used to provide scientists with the ability to run CESM in the cloud in the future. 

Meetings

Recurring workshops and conferences give NCAR a way to track emerging research fields and technologies, monitor the state of the art, and promote their use in the community. Three such activities are highlighted below, focusing on climate informatics, the use of many-core processors in modeling applications, and hands-on training in machine learning. These conferences and workshops were supported using NSF Core funds. In some cases, vendor donations, registration fees, or discretionary funds were used for ancillary items such as food and refreshments in accordance with NSF policies.

iCAS 2019

The International Computing in the Atmospheric Sciences (iCAS) symposium brings together an international group of center directors and specialists from the atmospheric and related geosciences communities to discuss advances in the ever-changing computing landscape. More than 50 attendees from around the world heard about machine-learning efforts and CMIP6 progress at sites worldwide at the September 2019 symposium in Stresa, Italy.

In addition to machine learning and CMIP6, the symposium’s themes included new technologies and storage and computing in the cloud, but machine learning permeated many presentations. From Richard Gerber’s discussion of the machine-learning use cases to be served by the Perlmutter supercomputer system being acquired by the National Energy Research Supercomputing Center to Niall Robinson’s talk on machine learning at the U.K. Met Office, machine learning is clearly on the rise at centers around the world.

Attendees also learned about the technical, philosophical, and data management challenges related to the construction and operations of the Daniel K. Inouye Solar Telescope (DKIST) in a keynote by Alisdair Davey, data center scientist for the DKIST project at the U.S. National Solar Observatory.

Improving Scientific Software Conference 2019

The annual Improving Scientific Software Conference (ISSC) took place at the NCAR Center Green Campus in April 2019. More than 80 non-speakers attended. The conference featured 24 talks and 11 tutorials across four tracks:

  • Utilizing modern HPC Architectures for Scientific Software

  • Utilizing Machine Learning Techniques in Scientific Software

  • Utilizing Software Engineering Practices to Improve Scientific Software

  • Utilizing Containers and Cloud for HPC

As in previous years, the project hosted students who received travel support to attend the conference to develop their scientific software skills.

Travel scholarship recipients and mentors
Student Travel Scholarship awardees and NCAR/UCAR mentors pose after the April 9 scholarship luncheon at ISSC-19. 

Multicore 9

The ninth annual Multicore Workshop (MC-9) was held in September 2019 and was attended by 57 participants. The workshop is a forum for open discussion and learning to better understand the application of new HPC technologies for the next generation of weather, climate, and Earth system models. The new generation of HPC platforms has diverse, heterogeneous architectures that present significant challenges to the community working on these models. In addition to providing a forum for participants to share their experience and lessons learned from developing models on these platforms, the workshop goals include creating a community of developers to collect and enunciate requirements for next-generation HPC programming models, tools, systems, and hardware.

Topics discussed at MC-9 included advancements in machine learning and deep learning for modeling, data analysis, and data assimilation; programming, performance analysis and optimization; model development and correctness; alternative language and tool development; and involving students and early-career computational scientists in HPC for Earth system applications.

Multicore 9 participants
The two-day Multicore 9 workshop drew 57 participants. NCAR uses this strategic workshop to share knowledge and expertise about emerging computing technologies broadly with students, faculty, and staff from peer centers.