NCAR Strategic Capability projects 2015-2016

Evaluating a new orographic drag parameterization for CAM using data assimilation techniques

Project lead: Julio Bacmeister, CGD
Yellowstone allocation: million core-hours

The drag exerted on the atmosphere by topography and the vertically-propagating gravity waves generated by topography is an important component of the atmospheric momentum budget. At the current horizontal resolutions used in global climate and weather forecasting models much of this drag is unresolved and must be represented by parameterization schemes. Incorporation of orographic gravity wave drag (OGWD) parameterization schemes began in the 1980s and was shown to improve global atmospheric model performance in both weather and climate applications. Beginning in the late 1990s, increasingly sophisticated OGWD schemes incorporating topographic anisotropy and nonlinear low-level flow dynamics were developed and introduced in global models. These more sophisticated schemes were shown to have significant positive effects on both short-term forecast skill and climate simulation quality. The current OGWD scheme in the Community Atmosphere Model (CAM) is from the first generation of schemes (McFarlane 1987). Anisotropy and low-level nonlinearity are neglected. This project will evaluate a more complete OGWD scheme in CAM by using a data assimilation system. Innovations (corrections) produced during the assimilation cycle are objective measures of model error. We will compare the magnitude of innovations in assimilations using different OGWD schemes. Smaller RMS innovations indicate that the model performance is improving. After characterizing the model’s performance in assimilation mode we will perform unconstrained 10-day forecasts to examine longer-term model drift using different OGWD schemes.

Projection of future air quality over South Asia

Project lead: Mary Barth, ACD
Yellowstone allocation: 6.8 million core-hours

About one-seventh of the world’s population living in South Asia faces the risk of severe air pollution as anthropogenic emissions of air pollutants are increasing continuously in response to economic development and urbanization. Recent studies show that air pollution in South Asia is already reducing the lifespan of about 660 million people by about three years and destroying enough food to feed about 94 million people. This problem may worsen in the future as anthropogenic emissions are expected to increase further due to economic growth in South Asia, and climate change is expected to lead to atmospheric conditions conducive to the production and accumulation of air pollutants. Therefore, the objective of this project is to understand how air quality of South Asia will change in the future due to changes in both the anthropogenic emissions and climate, and how those changes will affect human health and ecosystems in this region. To address this objective, we will perform 10 years of air quality simulations for both the present day (2005-2014) and future (2046-2055) time periods using state-of-the-art chemical transport models. These simulations will cover South Asia at a 60 km resolution but will zoom into the densely populated and heavily polluted Indo-Gangetic Plain with a 12 km resolution. This project will help address the first grand challenge identified in the NCAR strategic document and provide a solid scientific base for designing strategies to mitigate air pollution in South Asia.

Continuation: Highly-scalable Data Assimilation Research Testbed (DART)

Project lead: Nancy Collins, CISL
Yellowstone allocation: 3.7 million core-hours

The Data Assimilation Research Section (DAReS) group at NCAR develops, distributes, and supports a state-of-the-art data assimilation system, the Data Assimilation Research Testbed (DART). It works with many different models and observation types, including the NCAR community climate and weather models. The DART code has been downloaded by people at over 150 UCAR member institutions; over 400 institutions in total. DART runs on hardware that spans laptops to supercomputers.

In collaboration with researchers in CGD and MMM, DART already runs with the Community Earth System Model (CESM), the Weather Research and Forecast (WRF) model, and the Model for Prediction Across Scales (MPAS). Moving forward, the DAReS group has been redesigning the internal algorithms of DART so that it will scale to larger processor counts while requiring less memory per node, as future hardware projections show this to be the model for upcoming machines.

The redesigned code will need to be run with several large models to verify the accuracy of the new code, and it will have to be run on large processor counts to verify the scaling properties. While the DAReS group itself does not directly do atmospheric science with these systems, the use of DART enables other researchers both at NCAR and across the world to do successful data assimilation and make effective use of their computational resources.

Investigating the importance of climate variability on water resources in a changing climate

Project lead: Clara Deser, CDG
Yellowstone allocation: 6 million core-hours

Previous research in two areas has highlighted important concepts in climate research. First, the CESM “Large Ensemble” (LENS) experiment (Kay et al., 2014) has illustrated the tremendous variability in climate change signals due to internal variability within the system. Second, high-resolution WRF simulations over the Colorado Rocky Mountains (The Colorado Headwaters) have shown that proper representation of the spatio-temporal distribution of precipitation and other climate variables is key to understanding the impact of climate change on water resources. In addition, we have recently performed simulations with WRF to downscale two members of LENS (WRF-LENS); these runs have shown that this combination can generate realistic estimates of temperature and precipitation in current climate, and that future climate of the two downscaled simulations may be different from the changes predicted by CESM alone. The original “Colorado Headwaters” WRF experiments and the two WRF-LENS runs only provide a few climate change cases, and thus provided little guidance on the uncertainty due to internal variability. Here we propose to add four additional WRF-LENS runs to quantify how a high-resolution regional climate model modifies the CESM climate signal, and to assess the physical processes behind those differences. The planned simulations will triple the number of downscaled WRF-LENS ensemble members, and add an additional future time period (2070-2079) to all ensemble members. This will address the issue of internal variability, and be useful to many others in the research community who would like to make use of these simulations.

Magneto-hydrodynamic (MHD) simulations of homologous and cannibalistic coronal mass ejections (CMEs)

Project lead: Yuhong Fan, HAO
Yellowstone allocation: 5.7 million core-hours

Observations have shown that emerging solar active regions that exhibit strong magnetic twist and rotating sunspots tend to be source regions of strong, repeated (homologous) flares and coronal mass ejections (CMEs). Homologous CMEs have been observed to give rise to “cannibalism” or CME–CME interactions and merging, which are among the most energetic and geo-effective space weather phenomena. In this proposed project, we will carry out MHD simulations of CME initiation driven by an imposed flux emergence at the base of the corona, with the presence of an ambient solar wind, to study the development of homologous and cannibalistic CMEs, exploring different pre-existing field configurations and properties of the imposed flux emergence. The main scientific questions we will address are:

(1) How do the properties of the flux emergence, in conjunction with the configuration of the pre-existing coronal magnetic field, affect the production of repeated eruptions (homologous flares/CMEs) and reformation of the coronal flux ropes?

(2) What are the properties of the resultant cannibal CMEs in regards to the speeds, shock propagation and interaction, and the 3D magnetic field structures produced by the repeated eruptions?

The proposed study directly contributes to the NCAR strategic Imperative I in relation to HAO’s science mission to understand and quantify the solar magnetic variability as driver of space weather in the sun-earth system, by specifically investigating the origin and the underlying magnetic field evolution of the most energetic solar eruptive activities.

Testing and applying WRF-Solar for irradiance and solar power prediction

Project lead: Pedro Jimenez Munoz, RAL
Yellowstone allocation: 3.7 million core-hours

WRF-Solar is an augmented numerical weather prediction model to provide improved representation of the cloud-aerosol-radiation system. The model has been tested extensively under clear skies. Our results indicate the necessity of incorporating the effects of the atmospheric aerosol to obtain accurate estimations of the surface irradiances. The present project will assess and improve the WRF-Solar model under cloudy skies. The project will analyze 1) the effects of horizontal resolution in the cloud formation and evolution and its impact on the surface irradiance. 2) The effects that aerosols produce on the cloud lifetime and evolution and its impact on the surface irradiances. 3) The adequacy of simulating the effects that unresolved clouds produce on the surface irradiances using a shallow cumulus parameterization. 4) And finally, the improvements in the estimations of the surface irradiances as a consequence of assimilating infrared irradiances recorded from satellites to infer the presence of clouds. The results will benefit the community of WRF well beyond the needs of the energy industry.

Continuation: Local and remote regional climate responses to regional forcings from short-lived climate forcers

Project lead: Jean-Francois Lamarque, ACD
Yellowstone allocation: 5.3 million core-hours

This project aims at answering the fundamental question: what are the local and remote impacts of the regional perturbations in short-lived climate forcers, as these will be key drivers of regional climate changes over the next several decades? More specifically, we will target perturbations in emissions of short-lived climate forcers or their precursors. We will use a control experiment to provide the baseline and statistical basis for significance analysis. Then we will isolate the impact of a specific region and short-lived climate forcer emissions through a series of regional perturbations (e.g., SO2 emissions from the United States) from this control state.

Identifying the climate impact of decreasing solar activity during the 21st century

Project lead: Scott McIntosh, HAO
Yellowstone allocation: 3 million core-hours

Observations of solar activity over the last several decades are indicating a significant decrease in the total solar irradiance (TSI). Based on those observations, the very recent understanding of the underlying physical processes driving solar variability, and the possible projection of Maunder-like minimum during the 21st century, we propose to study the climate impacts of this minimum in solar activity compared to the standard approximation of a repeated 11-yr solar cycle. This proposed study is particularly relevant to the upcoming Coupled Model Intercomparison Phase 6 (CMIP6) since this modeling exercise will use a repeated solar cycle. In order to identify potential statistically significant signals, we propose to use the WACCM model (to capture the full range of processes associated with solar variability and trends) and perform an ensemble of simulations initialized from the existing WACCM CMIP5 simulations.

Enabling continental-scale high resolution hydrologic predictions through NCAR-NSC supercomputing

Project lead: Andy Newman, RAL
Yellowstone allocation: 8.6 million core-hours

WRF-Hydro (Gochis et al., 2014) has been selected to be part of national scale hydrologic initiatives guided by the National Weather Service aimed at providing high resolution, distributed, physically based hydrologic forecasts across CONUS. This hydrologic forecast system will provide forecast output at any model grid and resolved river channel point, potentially transforming the National Weather Service River Forecast Center workflow and forecast data availability, and has never been implemented across CONUS before. To advance continental scale, physics-based hydrologic prediction to the point of providing skillful forecasts, many challenges must be overcome. These include estimation of model parameters for soil hydrology (e.g., infiltration), estimating input data uncertainties and how model configuration choices impact model behavior. Progress toward meeting these challenges can be made through the production of long-term retrospective hydrologic re-analyses and parameter perturbation experiments. This work proposes a combination of such simulations to develop near optimal a priori parameter sets for initial model deployment and re-analyses that provide model benchmark performance and statistical properties.

Demonstration of a quasi-real-time air quality forecast/data assimilation system using FRAPPÉ and DISCOVER-AQ field data

Project lead: Gabriele Pfister, ACD
Yellowstone allocation: 3.5 million core-hours

This project aims to set up and test a real-time chemical weather forecast (CWF)/data assimilation (DA) system. The system is expected to give clear improvement over currently available air quality (AQ) predictions by including a state-of-the-art forecast model, the regional chemical Weather Research and Forecast Model with Chemistry (WRF-Chem), and the assimilation of satellite and ground-based chemical observations with the NCAR/DAReS Data Assimilation Research Testbed (DART) ensemble adjustment Kalman filter (EAKF) system. This real-time CWF/DA system - WRF-Chem/DART – is being developed and tested at NCAR/ACD. Under this proposal we will be testing WRF Chem/DART in a quasi-real-time application with data collected during the 4-week long NCAR/NSF and State of Colorado Front Range Air Pollution and Photochemistry Éxperiment (FRAPPÉ) and the fourth deployment of the NASA DISCOVER-AQ campaign. Both campaigns took place last summer in Colorado and provide a comprehensive observational data set to evaluate the performance of WRF-Chem/DART. The state-of-the art AQ analyses and forecasts that will be produced by WRF-Chem/DART will also be valuable for interpretation of the field campaign data. A demonstration of how an advanced forecasting system can improve AQ predictions will provide guidance to the AQ community on future directions in air quality modeling and predictions.

A detailed evaluation of the water cycle over the contiguous United States including potential climate change scenarios

Project lead: Roy Rasmussen, RAL
Yellowstone allocation: 11.1 million core-hours

The NCAR Water System program and RAL’s Hydrometeorological Applications Program strive to improve the full representation of the water cycle in both regional and global models. Our previous high-resolution simulations using the WRF model over the Rocky Mountains revealed that proper spatial and temporal depiction of snowfall adequate for water resource and climate change purposes can be achieved with the appropriate choice of model grid spacing and parameterizations. The climate sensitivity experiment consistent with expected climate change showed an enhanced hydrological cycle with increased fraction of rain versus snow, increased snowfall at high altitudes, earlier melting of snowpack, and increased total runoff. In order to investigate regional differences between the Rockies and other major mountain barriers and to study climate change impacts over other regions of the contiguous U.S. (CONUS), we have proposed to expand our prior CO Headwaters modeling study to encompass most of North America. A domain expansion provides the opportunity to assess changes in orographic precipitation across different mountain ranges in the western U.S., as well as the very dominant role of convection in the eastern half of the U.S. The planned simulations leverage many other RAL and NCAR-wide imperatives and frontiers. The WRF-downscaled climate change data will also become a valuable community resource for many university groups who are interested in studying regional climate changes and impacts but unable to perform such long-duration and high-resolution WRF-based downscaling simulations of their own. Such large domain high-resolution regional climate simulations were made possible by the 2014 NCAR Strategic Capability (NSC2014) allocation. This request seeks an extension of our NSC2014 award to continue and complete the ongoing modeling project on the North America water cycle in current and projected future warmer climates.

Investigation of solar dynamo models and magnetic flux emergence

Project lead: Matthias Rempel, HAO
Yellowstone allocation: 3.75 million core-hours

The Sun as a magnetic star exhibits a wide variety of activity, ranging from the fairly regular 11-year cycles of major sunspot eruptions with orderly patterns on the global scales to the intense and more chaotic magnetic fields on the small scales. The Sun shows that magnetism in stars involves both beauty and complexity. The large-scale fields seen as sunspots and active regions follow orderly and systematic cyclic patterns that have long been studied (e.g., Charbonneau, 2010; Hathaway, 2010). Both the large-scale and small-scale components contribute to the total magnetic activity of the Sun, which is the primary driver for solar variability. A clear challenge is to understand how solar dynamo action within the interior can yield such diverse and evolving magnetic fields. In this project we will address two key components of this grand challenge: cyclic dynamo action in the solar convection zone and magnetic flux emergence leading to the formation of active regions (sunspot groups).

Real-time high resolution ensemble analyses and forecasts of high impact weather with NCAR’s DART facility and WRF model

Project lead: Glen Romine, MMM
Yellowstone allocation: 8.7 million core-hours

Forecasts of convective precipitation systems remain a considerable challenge for the weather community. Yet, recent progress in convection-permitting ensemble forecast system design has generated a path forward by providing guidance on convective forecast uncertainty. Ensemble data assimilation serves as a foundation for ensemble forecast systems, differentiating spatial regions with the greatest initial condition uncertainty, while ensemble forecasts evolve this uncertainty and map how the information deficiency in the initial state impacts forecast outcomes. The proposed project aims to continue recent progress in ensemble prediction of high impact convective weather using NCAR’s Data Assimilation Research Testbed (DART) facility and Advanced Research Weather Research and Forecast (WRF) model. During this allocation period we will:

1) Operate a real-time ensemble analysis and high-resolution ensemble prediction system to produce real-time, 48-hr, 10-member, 3-km WRF-based ensemble forecasts over the entire contiguous United States for one year.
2) Investigate convection-permitting ensemble analysis to initialize high-resolution ensemble forecasts.

We will also continue analysis activities on prior computational results.

Tropical cyclone predictability experiments

Project lead: Bill Skamarock, MMM
Yellowstone allocation: 7.3 million core-hours

Tropical cyclones (TCs) are poorly predicted in global models that do not resolve the mesoscale. Mesoscale resolutions, however, are not attainable in operational NWP global ensembles. In this project, we propose to use both coarse-resolution and mesoscale-resolution global ensembles to address critical questions concerning TC predictability and TC interactions with the larger scale on short and intermediate timescales (0 – 10 days).

Specifically, we wish to understand the role of large-scale waves in determining tropical cyclone tracks and their influence on track error, and the influence of tropical cyclones on the larger-scale flow predictability. We will also explore questions concerning the nature of error growth and predictability in the tropics versus the extra-tropics and implications for ensemble construction. We will use the atmospheric component of the Model for Prediction Across Scales (MPAS-A), and the predictability experiments will make use of ensemble forecasts produced on both a uniform 60 km mesh and 60-15 km variable resolution meshes with the high-resolution region centered over northern hemisphere TC basins (Atlantic, Eastern and Western Pacific).

NWP ensembles are under-dispersive (ensemble spread < error) when model error is not accounted for. The ensemble integrations will use the Stochastic Kinetic Energy Backscatter Scheme (SKEBS) to account for model error. The stochastic perturbations introduced by SKEBS have a theoretical rational, whereas other methods of accounting for model error, e.g. multi-physics and multi-model based ensembles, have no theoretical basis.

Uniform and variable-resolution data assimilation experiments with MPAS

Project lead: Chris Snyder, MMM
Yellowstone allocation: 6.3 million core-hours

We propose to further test, develop, and apply data-assimilation (DA) capabilities for the nonhydrostatic atmospheric component of the Model for Prediction Across Scales (MPAS-A). Our overall objective is to develop efficient, scalable DA for MPAS-A as a platform both for probabilistic weather prediction and for rigorously evaluating the model’s fidelity against observations, across a range of spatial scales. Computing in this proposal will support three science objectives: (i) reduction of systematic errors in MPAS-A, in part through objective identification of bias in MPAS-A and diagnosis of its sources, (ii) variational techniques that, as part of the minimization needed for a single deterministic analysis, also provide an ensemble of analysis perturbations that reflect the analysis uncertainty, and (iii) issues related to ensemble-based DA on variable resolution meshes. These science objectives, especially, will rest in part on assimilation of satellite radiances, which are the crucial observation type for global data assimilation, and computing in this proposal will also support testing related to the development of radiance assimilation for MPAS-A.