NCAR Accelerated Scientific Discovery projects 2017

The Accelerated Scientific Discovery (ASD) initiative provides dedicated, large-scale computational resources to a small number of projects for a very short time period, usually two months or so following acceptance of a new HPC system. These projects are selected to help put the new system through its paces and pursue scientific objectives that would not be possible through normal allocation opportunities. In most cases, three to five NSF-supported, university-led projects from across the geosciences or the supporting computational sciences are chosen.

For the Cheyenne HPC system, the second system installed at the NCAR-Wyoming Supercomputing Center (NWSC), CISL issued calls for university and NCAR proposals in mid-2016 and received 11 proposals for ASD projects, ranging from 10 million core-hours to 58 million core-hours. Altogether the requests asked for 300 million core-hours, but only 250 million core-hours were available. The NCAR proposals were reviewed by a panel of NCAR scientists, and the following six awards were approved by the NCAR Executive Committee.

Predicting near-term changes in the likelihood of climate extremes: Initialized decadal climate prediction using large ensembles

Project lead: Stephen Yeager, CGD
Cheyenne allocation: 35.4 million core-hours

The aim of decadal climate prediction research is to enhance our foreknowledge of high-impact, near-term climate shifts associated with both the intrinsic and extrinsic variations of the Earth system. The five-year NSF-funded EaSM2 project (OCE-1234015) has played a leading role in advancing the decadal prediction objectives spelled out in the most recent NCAR and CGD Strategic Plans. Ensemble climate simulations initialized from historical conditions and subject to prescribed external radiative forcings have been shown to possess significant multi-year predictive skill, particularly for upper-ocean heat content and surface air temperature in the North Atlantic region as well as for Arctic winter sea ice extent. In this proposal, we seek to tackle the compelling scientific challenge of predicting changes in the likelihood of regional climate extremes while at the same time accelerating the development and testing of the Community Earth System Model (CESM) decadal prediction system. The key questions we hope to address are:

  • How does ensemble size impact the assessment of decadal prediction skill?
  • Are there predictable shifts in the probability of extreme weather events (such as heat waves, cold spells, or floods) associated with interannual-to-decadal sea surface temperature and sea ice extent variations?

To answer these questions, we will expand the ensemble size of an existing suite of prediction experiments from 10 to 40 in order to better resolve the probability density distributions that allow extreme events to be identified and to improve the coverage and power of the statistical tests used in prediction skill score assessment.

Sub-seasonal to seasonal prediction using high-resolution CESM

Project lead: Joe Tribbia, CGD
Cheyenne allocation: 30.8 million core-hours

We propose to research the capability of CESM in sub-seasonal to seasonal (S2S) climate forecast mode. The challenges of accurate S2S forecasts span both weather prediction, in which accuracy is highly dependent on accurate initial conditions and high resolution in the atmosphere, and seasonal coupled prediction, in which accuracy depends strongly on coupled interactions with the ocean. In this research study we will perform S2S historical forecasts (hindcasts) for the past 30 years using initial conditions for the atmosphere and ocean supplied by the European Center for Medium Range Forecasts (ECMWF) using a 10-member ensemble. The start dates for the hindcasts will be May 1 and November 1 of each year from 1985 to 2015. This research and computing resource proposal is linked to a companion university proposal submitted from George Mason University requesting resources for a similar study using the ECMWF coupled forecast system. Both systems will perform the identical hindcast suite of initial dates using the same analyses for initials but the details of the forecast system used by each group will differ. These companion proposals will afford both groups the unprecedented opportunity to compare and contrast results at the highest spatial resolution affordable within the guidelines of the ASD call.

Climate impacts based on a multiple-member ensemble geoengineering study with WACCM

Project lead: Simone Tilmes, ACOM/CGD
Cheyenne allocation: 23.1 million core-hours

We propose to perform a suite of multiple-member ensemble simulations with a relatively new, but already well tested version of CESM-WACCM to identify impacts of geoengineering on climate. For this, a simple feedback mechanism has been coupled to WACCM, which determines the amount and location of continuous SO2 injections into the stratosphere, with the goal of achieving specific climate objectives. Here, we are aiming to keep surface temperatures at 2020 conditions, while following the RCP8.5 climate forcing scenario. While global surface temperature will be adjusted to maintain certain conditions, other climate variables are not and significant climate impacts and side effects of geoengineering may occur, including regional temperature extremes and changes in the hydrological cycle. A multiple-member ensemble simulation of such a geoengineering experiment has not been pursued before and is essential to analyze the significance of regional changes, extremes, and other impacts of geoengineering. The multiple-member ensemble will be shared with the community to allow experts from different areas, including land and ocean, to analyze the simulations and further help identify important impacts of such a geoengineering approach.

Working towards data-driven models of solar eruptions

Project lead: Matthias Rempel, HAO
Cheyenne allocation: 18.5 million core-hours

The magnetic field of the Sun plays a key role in solar activity that greatly impacts the solar-terrestrial environment. It is crucial to understand how the magnetic field is generated and transported inside the Sun, and how the magnetic energy in the solar atmosphere is stored and released. The whole process spans over several orders of magnitude in time and spatial scales, which makes this problem a great challenge to numerical modeling. We propose to build a comprehensive and realistic numerical model of the emergence of the magnetic field from the upper layers of the convection zone into the corona of the Sun that can be directly compared with available observations and that provides 3D data with sufficient temporal and spatial resolution to analyze the underlying physical processes. While we exploit novel approaches to allow for an efficient computation of coronal dynamics, we still require a significant amount of computational resources in order to capture the typical length and time scales of active region flux emergence on the sun. This research is aligned with the NWSC Community Science Objective: S.1.2 Modeling the emergence of the magnetic flux from the solar convection zone and the conditions that lead to solar flares and CMEs, S.2.1 Coupled or integrated models of radiative MHD processes and magnetic flux emergence, and S.3.4 MHD simulations of flux emergence in a vertical domain from 20 Mm below to 100 Mm above photosphere and 100 Mm x 100 Mm horizontal domain.

CESM2 regional climate community simulations

Project lead: Andrew Gettelman, CGD/ACOM
Cheyenne allocation: 13.7 million core-hours

Climate impacts are typically felt through extreme weather. We propose to use static mesh refinement in CESM2 to perform advanced regional climate simulations over the United States at 1/8˚ (14km) horizontal resolution with several different configurations of a consistent global model. These unique high-resolution simulated climate statistics for the U.S. will be evaluated against observed climate extremes to determine what configurations best reproduce observed extreme weather. Some examples of extreme events are extreme precipitation from mesoscale convective systems in the Central U.S., or the effects of ENSO on Western U.S. orographic rainfall. We will explore how climate interacts with the land surface and hydrology over the United States, and how climate extremes may change in the future.

Wind forecast improvement Project 2

Project lead: Pedro A. Jimenez Munoz, RAL WSAP
Cheyenne allocation: 10 million core-hours

Numerical weather prediction (NWP) models are being run at sub-kilometer horizontal grid spacings. However, NWP models use planetary boundary layer (PBL) parameterizations that assume statistically homogeneous turbulent motions in the horizontal. Under this simplification, PBL schemes only account for one-dimensional (1D) mixing resulting from vertical gradients in turbulent fluxes. Although a convenient assumption for grid spacings of several kilometers, assuming horizontally homogeneous turbulence motions no longer holds at sub-kilometer grid spacings. This limits the benefit of fine grid spacings over complex terrain regions. To overcome this limitation, we have implemented in WRF a three-dimensional (3D) PBL parameterization to account for both vertical and horizontal gradients in turbulent fluxes. The 3D-PBL parameterization is being validated against data from a field campaign. The present project will run WRF in a turbulent-resolving large-eddy simulation mode (WRF-LES) at fine horizontal grid spacing (i.e., 30 m and 10 m) to complement the observations and provide a comprehensive data set to validate the 3D-PBL parameterization. The evaluation focuses on the performance of the 3D-PBL scheme to reproduce the wind over complex terrain at resolutions where the assumption of horizontal homogeneity is no longer valid.