NSC projects May 2017 - April 2018

Developing guidance for short-fused, high-impact weather prediction

Project lead: Glenn Romine, MMM
Cheyenne allocation: 13.2 million core-hours

Despite the formidable challenge of predicting convective weather, forecast capability is steadily progressing through advances in knowledge of the physics of convective systems, data assimilation techniques, model improvements, and probabilistic guidance enabled by high-resolution ensemble forecast systems. Forecasts of high-impact weather benefit considerably from both high-resolution forecasts that explicitly represent deep convection and from ensemble forecast systems that provide probabilistic guidance by sampling forecast variability owing to uncertainties in atmospheric state and model characteristics that impact explicit predictions of weather hazards. Most operational and experimental prediction systems conduct the analysis on a much coarser grid than the forecast, degrading the value of forecasts for short-term, high-impact weather hazards. The proposed activities will advance the highly successful NCAR ensemble project, which provides high-resolution ensemble initialized from a mesoscale ensemble analysis appropriate for next-day predictions, to a convection-permitting ensemble analysis and forecast system appropriate for 1-36 h predictions of ongoing and near-future, high-impact weather. Specific design aspects that will be investigated in the proposed work include continuous versus partial cycling, covariance inflation methods, and hybrid variational-ensemble analysis re-centering.

The impact of climate change on the physics and biology of the ocean on scales down to the submesoscale

Project lead: Matthew Long, CGD
Cheyenne allocation: 12 million core-hours

Climate warming will drive substantial changes in the upper ocean environment, with implications for ocean ecosystems and biogeochemistry. Climate projections capture aspects of these changes, but are nearly always conducted with coarse-resolution ocean models that do not resolve mesoscale dynamics. This is problematic, given that the mesoscale is the dominant energetic scale in the ocean and plays a fundamental role in controlling the structure of the upper ocean habitat. High-resolution climate projections are not often feasible, however—particularly with ocean biogeochemistry—given the substantial computational cost associated with eddy-resolving ocean models. In this project, we employ a novel experimental framework to confront this challenge and enable investigations of the role of scale-interactions in climate change. The approach involves conducting ``present-day'' and "future-climate'' timeslice experiments with the CESM 1/10$^\circ$ ocean and sea-ice components. The future climate state is defined through application of forcing and initial state anomalies derived from a large ensemble of coarse-resolution coupled-model integrations. The timeslice experiments will enable quantification of changes in mesoscale dynamics and their impact on ocean biogeochemistry with climate warming.

Dynamics of the atmospheric kinetic energy spectrum

Project lead: Bill Skamarock, MMM
Cheyenne allocation: 15.2 million core-hours

One of the unsolved questions in atmospheric dynamics concerns the energetics responsible for the k^-5/3 scaling for the mesoscale portion of the atmospheric kinetic energy (KE) spectrum (where k is the horizontal wavenumber). High vertical and horizontal resolution atmospheric simulations will be performed with the Model for Prediction Across Scales (MPAS) to examine the dynamics of the mesoscale motions producing the KE spectra. Previous simulation results indicate that atmospheric models under-resolve the vertical structure in the waves that dominate the mesoscale structure in the atmosphere, particularly in the stratosphere. Very-high vertical grid spacing, down to 50 meters for the 15 km global mesh, and 100 meters for a 3.75 km mesh, will be used to resolve the waves and converge the KE spectra. The simulations should provide evidence for or against existing theories for the energy cascade, and they also should allow us to understand the role of topographically generated waves, convection, and large-scale baroclinic waves in driving the mesoscale dynamics. The simulations will also shed light on what is missing in atmospheric simulation using the much-coarser vertical resolutions typically employed in our applications.

Simulating mid-Pliocene tropical cyclones and mean climate state

Project lead: Ran Feng, CGD
Cheyenne allocation: 12.7 million core-hours

The mid-Pliocene warm period (3.0-3.3 Millions of years Ago, Ma, mPWP) is widely considered an analogue for future climate change due to its similar paleo-geography, topography, and atmospheric CO2 concentration to present-day. However, simulating the climate of the mPWP has proven challenging for complex Earth system models (ESM) participating in Pliocene Model Intercomparison Project Phase 1 (PlioMIP1). When given mPWP boundary conditions and external forcings, ESMs (including both CCSM4 and CESM1.2) fail to reproduce the basic features of mPWP tropical climate. These features are 1) reduced SST gradient between the equator and subtropics and between the eastern and western equatorial Pacific, and 2) moist Sahel (10-20 °N, 20°W-40 °E) and northward shift of the Sahara Desert boundary. These mismatches with proxy data limit our confidence in ESMs to simulate climate conditions different from present-day that might occur in the future. Here, we propose to revisit this problem using the high-resolution Community Earth System Model version 1.3 (CESM1.3) at 0.23×0.31° atmosphere and 1° ocean resolution. The high resolution permits explicit simulations of tropical cyclones and intense storm activities, which will help us understand tropical cyclone and storm-related changes in the heat budget and climate energetics under an enhanced greenhouse condition.

Simulating quasi-periodic bursts of solar activity from Tachocline Nonlinear Oscillation (TNO)

Project lead: Mausumi Dikpati, HAO
Cheyenne allocation: 6.2 million core-hours

This proposal seeks to improve predictive capability for bursts of solar activity on timescales of weeks to months using data assimilation techniques. Episodes of extreme solar activity occur as a result of global bursts in solar magnetism that take place on time scales of 6-18 months (McIntosh et al. 2015). These episodic bursts of solar magnetism, called the “seasons of solar activity,” are associated with the energetic output and have the same order-of-magnitude as the decadal-scale variability known as the solar cycle. These bursty phases are exhibited in coronal mass ejections (CMEs) and solar flares that pose a significant threat to our technology-dependent society. These global-scale outbursts of magnetic activity have been associated with instabilities present in the global-scale magnetic field that is rooted deep in the Sun's convective interior at a shear interface dubbed the “tachocline.” Dikpati et al. (2017) have recently shown, using a nonlinear shallow water model of the solar tachocline, that a back-and-forth exchange of energy between tachocline differential rotation and Rossby waves occurs with a periodicity of 3-19 months, within the observed range of burst periods. These Tachocline Nonlinear Oscillations (TNO) can cause the activity bursts when the Rossby waves’ energy grows to its maximum, because the tachocline top surface is maximally deformed then. Hence, nearly “frozen-in” spot-producing toroidal magnetic fields can enter the convection zone from the tachocline and start their buoyant rise to the surface to erupt as active regions. Building on the successful completion of a series of high-resolution simulations of tachocline nonlinear oscillations (TNO) in an MHD shallow water tachocline model (MHD-SWT model), we plan to simulate the periods of magnetic activity bursts and their locations of occurrence at the tachocline level. Then their corresponding locations at the solar surface will be estimated by employing a magnetic flux emergence recipe (forward operator here) followed by data assimilation using DART.

High-resolution simulation of the effects of climate-urbanization-crop interactions

Project lead: Michael Barlage, RAL
Cheyenne allocation: 5.9 million core-hours

In this joint ASU/NCAR EaSM-3 project, the NCAR team is tasked to employ WRF-Urban-Crop to additional 1-km simulations for selected U.S. metropolitan areas. The overarching goal is to advance our understanding and modeling capabilities of linked land-use and land-cover agricultural and urban processes on decadal and regional scales, and the consequences owing to future agricultural productivity within and adjacent to urbanized regions for the U.S. We address the following questions:

  • What are the decadal and regional environmental impacts associated with urban/agricultural expansion?
  • What additional feedback loops and interactions are realized with two-way coupling and how do they vary with place and season?
  • What are the impacts of changing landscapes on the modulation of the diurnal cycle and its seasonal variability at decadal time scales, and on the length of growing seasons? What are implications for yield for a variety of crops?

A number of WRF-Urban-Crop simulations are conducted under current and future climatic conditions, which leverage an on-going RAL decadal 4-km WRF CONUS modeling effort (supported by previous NSC requests and led by Roy Rasmussen). Those proposed ensemble simulations with explicit representation of urbanization and crop-growth processes are be used to contrast current 4-km CONUS simulations to discern effects of rapid urbanization and agriculture expansion on regional climate. Further, numerous 1-km WRF-Urban-Crop simulations are performed over selected metropolitan regions in the U.S. to investigate the regional climate effects and social-economical benefits of urban farms and gardens.

Transient evolution of the Greenland Ice Sheet over the last interglacial warm period

Project lead: Marcus Loefverstroem, CGD
Cheyenne allocation: 8.4 million core-hours

The Last Interglacial (LIG, c. 130-115 kyrs BP; thousands of years before present) was the most recent time in Earth's history that the Greenland Ice Sheet (GrIS) was significantly smaller than at present as a result of climate warming. Estimates of the total mass loss are uncertain but typically fall between 1.5 and 4.5 meters of sea-level equivalent, or approximately 20 to 65% of its modern volume. These estimates are based on surface mass balance calculations from simplified models with interactive ice sheets, and from comprehensive models only partially coupled to an ice-sheet model. The advent of the second version of the NCAR Community Earth System Model (CESM2) will support a fully interactive ice-sheet component over Greenland (Community Ice Sheet Model; CISM2). This is the first comprehensive Earth system model that has this capability and is therefore expected to significantly improve our understanding of how the GrIS interacts with the rest of the climate system. Here we propose running a 10,000-year transient fully coupled simulation over the LIG warm period and the subsequent transition into cooler conditions (128-118 kyrs BP). This simulation would be the first of its kind and will therefore be informative of how the coupled system responds to climate change (both warming and cooling), and how a massive volume change of the GrIS influences the atmosphere and ocean general circulation. This is important for assessing the ice sheet's sensitivity to climate change and for testing the model's capability of realistically simulating past climate states.

Enhanced Eocene polar warming due to biogeophysical feedbacks

Project lead: Jeffrey T Kiehl, CGD
Cheyenne allocation: 3.3 million core-hours

Earth’s deep past provides a unique opportunity to test Earth system models (ESMs) for a wide range of climate states. Given that projected future levels of atmospheric CO2 will be much higher than pre-Industrial times, it is of great interest to explore how ESMs simulate past climates in which CO2 was much higher. One deep time period of particular interest is the early Eocene, where proxies indicate significant warming occurred at high latitudes. Simulating the high-latitude warmth during this period has proven to be challenging for many climate models. Recently, Kiehl and Shields found that amplified polar warming is well simulated by including the effects of changes in high-latitude gas emissions by vegetation and aerosol particle-cloud interaction processes. However, the version of CCSM used by Kiehl and Shields did not include detailed mechanisms linking aerosols to cloud microphysics. It is proposed that a suite of new early Eocene simulations be carried out using CESM1.2 to further explore the interaction between terrestrial vegetation, aerosols and cloud properties as high-latitude climate feedbacks.