Source: UTAH STATE UNIVERSITY submitted to NRP
HIGH PERFORMANCE COMPUTING, UTAH
Sponsoring Institution
National Institute of Food and Agriculture
Project Status
COMPLETE
Funding Source
Reporting Frequency
Annual
Accession No.
0221597
Grant No.
2010-34610-20744
Cumulative Award Amt.
(N/A)
Proposal No.
2010-01517
Multistate No.
(N/A)
Project Start Date
Jul 1, 2010
Project End Date
Jun 30, 2011
Grant Year
2010
Program Code
[CC-F]- High Performance Computing, UT
Recipient Organization
UTAH STATE UNIVERSITY
(N/A)
LOGAN,UT 84322
Performing Department
Agricultural Experiment Station
Non Technical Summary
The project goal is to extend the use and application of high performance computing to the agricultural research community. The Center for High Performance Computing at Utah State University (HPC@USU) addresses this goal by developing an integrated, easy to use computational infrastructure. This initiative is in concert with the Agricultural, Rural Development, Food and Drug Administration, and Related Agencies Appropriations Act of 2010, as well with the President's Information Technology Advisory Committee (PITAC) report (Committee, 2005) and the recently passed High Performance Computing (HPC) Revitalization Act and addresses the critical needs for high performance computing education and resources on the local, state and national level. Through this program, the high performance computing resources at HPC@USU will be expanded to meet new challenges related to producing, storing, processing and archiving substantially larger datasets. Agricultural researchers often require a variety of externally and internally generated datasets to both build their research process as well as validate their results. The ability to store and process a greater variety of larger datasets represents an essential infrastructural improvement for HPC@USU that will allow research simulations to achieve a new level of detail and confidence. In addition, a great research emphasis will be on directly educating, through hands-on computational research, a PhD student working with Dr. Jin to learn specifics of computational research as applied to snowpack and drought research.
Animal Health Component
(N/A)
Research Effort Categories
Basic
100%
Applied
(N/A)
Developmental
(N/A)
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
1020210207020%
1020430207020%
1327299208050%
9036099303010%
Goals / Objectives
1. Quantify the Relationship Between Snowpack and Drought in the Western United States using Regional Climate Modeling. In this work we will investigate how climate change affects snowpack and drought in the Western United States. To achieve this goal, the relationships among climate change, snowpack, and drought will be qualitatively assessed using historical data. Both field observations and numerical modeling results will be applied in this investigation. This study will provide insight into how climate change affects past and future snowpack, drought occurrence, and water supply. It will also provide decision-makers with better data upon which to make informed decisions. 2. Enhanced Capacity And Capabilities for Cluster Attached Storage. To accommodate the large and varied datasets that projects such as the Regional Climate Modeling project require, we will build a scale-out network attached storage (NAS) solution that will expand our current storage capacity, as well as provide our high performance computing environment with significant new storage capabilities. Features previously considered high-end storage capabilities such as filesystem parallelization, near-linear scalability, non-disruptive maintenance, enhanced availability through advanced RAID and data replication for offsite data storage archival, are now necessary tools for effectively managing large research datasets. Despite the premium nature of these features, the price per terabyte must be less than or equal to what we have historically paid for low-end storage. Recent vendor pricing shows that this is possible. Therefore, HPC@USU will build and grow a scalable-NAS storage solution with the appropriate balance of price, performance, scalability and availability that will allow us to maintain larger datasets for the life of a project as well as archive datasets that have value beyond the life of a project.
Project Methods
1. The advanced regional climate model Weather Research and Forecasting (WRF) developed by the National Center for Atmospheric Research (NCAR) will be used for this investigation. We recently coupled the Community Land Model version 3 (CLM3) with WRF (WRFC) to improve the simulations of snow and related land surface variables as well as heat and water flux exchanges between the land surface and atmosphere. The advanced CLM3 has been shown to be accurate in describing snow, soil, and vegetation processes for global and regional applications. CLM3 includes a 5-layer snow scheme, a 10-layer soil scheme, and a single layer vegetation scheme. Solid ice and liquid water are described in the snowpack as prognostic variables. A sophisticated snow compaction scheme is used to calculate the height and density of snow, where snow density is a critical variable for describing water and heat transfer within the snowpack. The model physically describes frozen soil processes and their impact on soil properties. A 20 km resolution domain will be set within WRFC; this domain covers the entire WUS that is the focus of this study. WRFC will be configured with 28 vertical sigma layers from the surface to the 50 hPa level. In order to properly represent the planetary boundary layer processes, the vertical layers will be closely spaced near the surface and coarsely spaced in the upper atmosphere. In addition, the 6 hr 2.5o x 2.5o National Centers for Environmental Predictions/ Department of Energy Reanalysis (NCEP 2) data will drive WRFC to perform historical simulations for the period September 1, 1979 to December 31, 2007. 2. HPC@USU will build and grow a scale-out NAS storage solution to replace and increase our existing general storage pool. We expect to initially build a 120TB general storage pool and then grow the solution from there. This is the best way for us to meet the challenge of processing, managing and archiving the mountain of research data that is beyond our current capabilities. One core feature of this solution will be filesystem parallelization via a protocol known as Parallel NFS (pNFS). The pNFS protocol is part of the NFS v4.1 standard, and it is an important consideration for compatibility reasons. While moving away from stand-alone NFS towards a scale-out NAS solution, compatibility and ease of migration is a principle concern and is largely addressed by the pNFS protocol. However, there are other important considerations that must be addressed as well. The appropriate balance of performance, scalability, availability, and price also need to be considered. The scale-out NAS solution that best balances these considerations is based on the Panasas ActiveStore Series 7 product that offers many high-end features in an entry-level scale-out NAS solution.

Progress 07/01/10 to 06/30/11

Outputs
OUTPUTS: Climate Modeling: To improve simulations of regional-scale snow processes and related cold-season hydro climate, the Community Land Model version 3 (CLM3) was coupled with the Pennsylvania State University/NCAR fifth generation Mesoscale Model (MM5) and the next generation regional weather and climate model, the Weather and Research Forecasting (WRF) model. CLM3 physically describes the mass and heat transfer within the snowpack using five snow layers that include liquid water and solid ice. Interactions among the snow, soil, and vegetation are characterized by the CLM3 mass and energy equations. We have carried out extensive simulations with the both improved MM5 and WRF to better understand the physical mechanisms that control regional climate and land surface processes such snow dynamics. In addition, we have developed the innovative techniques to better simulate and predict climate for the western United States. These techniques were to combine statistical and dynamical approaches to reduce the uncertainties in regional and global climate model outputs. River Modeling:This effort involves the implementation and evaluation of a new turbulence model. Current turbulence models are not able to accurately predict where turbulence occurs in a river. These current turbulence limitations exemplify the need for improvements. The mathematics behind this new turbulence model employs fewer approximations and in essence should yield improvements in predictive capability. Traditionally turbulence models were first derived for flows over a smooth surface and then extended to rough surfaces. The turbulence model being tested was developed primarily for fully rough flow in a pipe. Most commonly used turbulence models are not able to accurately predict the bulk properties of fully rough flow. The closure coefficients of this improved turbulence model were selected to match the friction factor and velocity distributions available from either experimental data or well-established empirical correlation such as the law of the wall. The resulting turbulent eddy viscosity was compared satisfactorily with experimental data. This new turbulence model is based on two transportable properties: the turbulent kinetic energy and the mean vortex turbulent wavelength.The equation for the turbulent kinetic energy is obtained from the Reynolds-Averaged Navier Stokes equations dotted with the velocity vector rather than the Reynolds stress transport tensor. Even though the turbulent kinetic energy has the same definition as in traditional models,the equation used to model the turbulent kinetic energy is somewhat different. The mean vortex wavelength is modeled using an empirical relation developed based on experimental data obtained for fully rough flow in a pipe. Using fewer approximations than traditional models, this new turbulence model could significantly improve the way turbulent flows are modeled using the RANS equations. Although developed primarily for fully rough pipe flow, this model will be extended to transitionally rough surfaces in later stages. PARTICIPANTS: Nothing significant to report during this reporting period. TARGET AUDIENCES: The target audiences for this work involve the scientific communities. In particular, the climate modeling communities and turbulence modeling communities. PROJECT MODIFICATIONS: Nothing significant to report during this reporting period.

Impacts
Climate Modeling: The coupled MM5-CLM3 model performance was evaluated for the snowmelt season in the Columbia River Basin in the Pacific Northwest using gridded temperature and precipitation observations,along with station observations from an automated snowpack telemetry (Snotel) system. The results show that the averaged snow water equivalent(SWE)observation is 494 mm during the snowmelt season. However, the averaged SWE produced by the original version of MM5 with the Noah land surface model is 174 mm for the same period, while it is 511 mm in our coupled MM5-CLM3. Thus, a significant improvement in the SWE simulations is seen. In addition, we quantified the influence of land use change and irrigation in the California Central Valley using MM5-CLM3. Our modeling results show that modern-day daily maximum near-surface air temperature(Tmax)decreases by 0.55 oC due to agricultural expansion since pre-settlement. Additionally, irrigation over the cropland has also affected the hydroclimate processes within the California Central Valley. Results show that irrigation lowers the temperature of the cropland surface. The daytime cooling is produced by irrigation-related evaporation enhancement. In the nighttime, the land-use-change-induced near-surface warming is alleviated by low intensity irrigation (17%
Publications

  • No publications reported this period