Climate and Global Dynamics Division

Global Dynamics Section

CGD ASR00 Home | ASR00 Home | CGD Home

The objective of the Global Dynamics Section (GDS) is to further develop the scientific understanding of the dynamics and predictability of large-scale atmospheric variability on timescales of days to decades. This process will allow construction of the scientific basis for predicting transient, global circulations in the atmosphere beyond the present practical limits. GDS scientists take three approaches to their research: (1) numerical and theoretical experimentation with a hierarchy of physical models ranging from the non-divergent, barotropic model to coupled atmosphere-ocean models; (2) diagnostic analyses of the cause of atmospheric climatic variability and its theoretical and practical predictability in simulation and forecast experiments using the NCAR Community Climate Model (CCM), Regional Community Climate Model (RegCCM), and Community Climate System Model (CCSM); and (3) sensitivity analyses of numerical prediction models to atmospheric initial and boundary conditions using variational techniques, which will aid in the design of improved methods of data assimilation, for conventional and non-conventional meteorological data, e.g., precipitation, soil moisture, and sea surface temperature.

Predictability and Prediction Studies of Weather and Climate Variations

The studies described below are highlights of the research in GDS devoted to the prediction and predictability of climate variations and extreme events. These studies are integral to our section goals of extending and defining the spatio-temporal domain over which scientifically and societally useful forecasts can be made. GDS scientists have continued their interest in the inherent predictability of atmospheric phenomena and have utilized their expertise gained in ensemble prediction techniques to address the prediction of extreme events.

A focused research effort centered around the goals of the U.S. Weather Research Program (USWRP) has reinvigorated GDS efforts in the predictability of synoptic time and space scales with particular emphasis on the predictability of precipitation.  The effort in GDS focuses on two main topics: the inherent predictability of synoptic scale precipitation and the influence of the loss of predictability on synoptic and larger scales on mesoscale processes. With Ronald Errico (GDS) and Joseph Tribbia (GDS), David Baumhefner (GDS) has conducted rigorous experiments with several versions of CCM to evaluate how differences in synoptic scale predictability error growth (PEG) affect the ensemble mean properties of prediction. The forecast skill of individual forecasts and ensemble mean forecasts of CCM3 were compared at 3 resolutions, T42, T63, and T106. Thirty-two 10-member cases were used to evaluate differences in the skill of individual forecasts and the skill of the ensemble mean forecasts. No significant differences were seen in the skill of the individual forecasts; however, significant differences were exhibited in the skill of the ensemble means. The skill of the ensemble means increased with increasing model resolution, which coincides with the progressive increase in PEG, lowest for T42 to highest for T106. This is the first conclusive evidence that a representative, accurate depiction of forecast uncertainty is necessary not only for accurate prediction of forecast reliability but also to achieve the maximum nonlinear filtering benefits of ensemble prediction.  An extension of this research analyzed PEG in the perfect data imperfect model framework. In this experimental design, models with differing horizontal resolution (T42, T63, T106, and a T170 control) are integrated with identical initial states in the scales that are resolved in common.  Error growth then enters the system through the inverse cascade of variance from unresolved scales into resolved scales. The results of this experimental design demonstrate that the initial cascade from unresolved into resolved scales is not an inertial cascade typical of turbulent flows since the transfer rate is far too rapid. Higher order nonlinearities associated with threshold physical parameterizations are much more important in the early growth phase, confirming some earlier results by David Williamson (Climate Modeling Section, CMS). Also, because such parameterizations are so common in numerical weather prediction (NWP), there is little utility, on average, in the use of prediction models with resolution higher than T63 if the goal is accurate synoptic scale prediction.

The early motivation for ensemble prediction studies in GDS was the problem of extended range prediction, both as a means to dynamically ascertain probabilistic information in a system with limited deterministic predictability and to sharpen signals forced by anomalous boundary conditions. Tribbia (GDS) with Jeff Lee (GDS) has continued research in seasonal forecast skill concentrating on a forecast comparison project set up by the Center for Ocean-Land-Atmosphere Studies (COLA) in which several models were tested for skill in a 3-month time frame. Sixteen  winter and summer cases have been run with 10-member ensembles using CCM3.These forecasts were all forced by observed sea surface temperatures (SSTs). In the continuation of this work, a subset of 5 winter and summer cases were integrated using the CCM3 generated boundary conditions to specify lateral boundary conditions over the Pacific to embed a high-resolution, limited area model, RegCM2, over North America. The forecast skill of the monthly and seasonal precipitation over the Mississippi River Basin is being compared with that forecast by CCM3 in the Dynamical Seasonal Prediction (DSP) intercomparison. There is some improvement in the location and amplitude of the precipitation in the nested high-resolution forecasts in the winter season. In the summer season, however, the limited area forecasts are degraded.

Predictability on seasonal-to-interannual timescales is often associated with the influence of the El Niño-Southern Oscillation (ENSO) phenomenon in the tropical Pacific.  However, there are a few other regions of the globe that also contribute to predictability on seasonal-to-interannual timescales. Perhaps the most notable of these regions is the tropical Atlantic. For example, rainfall in the northern Nordeste region of Brazil is known to be highly correlated with SST anomalies in the tropical Atlantic.

For the past several years, Ramalingam Saravanan (GDS) and Ping Chang (visitor, Texas A&M University) have been engaged in a collaborative study of tropical Atlantic variability. They have developed a Hybrid Coupled Model based on a statistical atmosphere and a coarse resolution ocean general circulation model (GCM). They have used this model to investigate the atmospheric response to both local SST forcing in the tropical Atlantic and remote influence from Pacific ENSO using CCM3.

As the next step in the hierarchy of models, they have now developed a version of CCM3 that can be forced by observed SST, climatological SST, or a slab ocean model on a grid-point by grid-point basis. Using this model, experiments have been carried out where CCM3 is forced by observed monthly SST in the tropical Atlantic region but is coupled to a slab ocean model elsewhere. This configuration allows for potential thermodynamic air-sea interaction in the midlatitude ocean, which can contribute to enhanced persistence of atmospheric anomalies.

A 10-member ensemble of 45-year integrations has been carried out using this model to study the remote influence of tropical Atlantic SST anomalies.  Preliminary results from this study suggest that the tropical Atlantic has a weak but statistically robust influence on the North Atlantic and European regions. Surprisingly, the spatial pattern associated with this influence is not the well-known North Atlantic Oscillation, but a pattern almost orthogonal to it! Further research is in progress to isolate the exact mechanisms behind this remote influence.

Because it is the largest external factor influencing interannual variability over North America, much effort continues to be devoted to understanding how equatorial Pacific warm and cold events affect midlatitudes.  In ongoing work in collaboration with T.-C. Chen (Iowa State University), Siegfried Schubert (NASA/Goddard Space Flight Center, GSFC), and Max Suarez (NASA/GSFC), Grant Branstator (GDS) has investigated several facets of this problem with the intent of determining whether there are steps that can be taken to improve the simulation of the response to tropical SST anomalies in atmospheric global circulation models.  This work makes it clear that one necessary ingredient for producing a skillful response is an accurate reproduction of the observed climatological mean circulation.  For example, it appears that the latest version of NASA's Aries GCM has a mean state that refracts disturbance wavetrains stimulated from the central Pacific in a more northerly direction than did the mean state in previous versions of that model, thus leading to an enhanced and more realistic midlatitude response in the updated model.  Along similar lines, because the new model has a mean subtropical Asian jet that extends farther across the Pacific, medium-scale disturbances originating over southern Asia can penetrate all the way to North America.  This means that rainfall anomalies in the western tropical Pacific that are associated with central Pacific SST events can influence the U.S. via a medium-scale wavetrain that the study has found is detectable in nature.  Another example of mean state influences on perturbation components of the circulation has to do with the effect of seasonal mean conditions on the amplitude of internal variability in the atmosphere.  Using ensembles of GCM simulations that explore realization-to-realization variability under cold and warm equatorial Pacific conditions, the study has found that there is much higher midlatitude variability during cold conditions.  This means that there is much less certainty in the reaction of the atmosphere to La Niñas than to El Niños.  Experiments with stochastically excited linear dynamical models have traced this effect to the difference in the mean circulation induced by changes in the equatorial SSTs during these events.

Diagnostic and Theoretical Studies of Variability and Validation

Within GDS the purpose of diagnostic analyses is twofold: diagnosis is used to test theoretical ideas concerning the mechanisms responsible for climate variations and their relative import and also test (i.e., validate) the behavior of comprehensive climate models like the CCM against that of the observed climate system. Naturally, the aforementioned prediction studies can also be viewed in this latter context. Additionally, several particularly insightful examples of past GDS studies exemplifying these two types of diagnoses are detailed below.

For the most part, investigations of interannual variability concentrate on winter because that is the season of greatest variance.  But, in an ongoing investigation of the seasonality of interannual variability, Branstator (GDS) has found, using National Centers for Environmental Prediction (NCEP)/NCAR reanalysis products, that interannual circulation anomalies are substantial in all seasons while the structure of these anomalies is highly dependent on season.  During the past year Branstator (GDS) has concentrated on determining the mechanisms that produce this seasonality in structure.  One tool that has helped this endeavor is a linear model that is derived in a purely empirical manner.  By design, this model is the most accurate linear model of barotropic atmospheric dynamics that can be constructed.  It was first developed to estimate the response of the atmosphere to external forcing, but when applied to the seasonality problem, it can be used to demonstrate that interannual mean disturbances derive their distinct seasonality from the internal dynamics of the atmosphere and not because of any special organization or seasonality from external stimuli.  Further analysis has revealed that interannual anomalies tend to take on structures that allow them to be slowly evolving even in the absence of anomalous forcing and in such a way that they will damp the climatological waves.  Many of these characteristics appear to have their origins in the simple action of large-scale disturbances conserving absolute vorticity in the presence of the climatological waves.  Feedbacks from momentum fluxes produced by synoptic scale perturbations also play a role in generating the distinctive characteristics of the seasonality.  Because similar behavior is found in CCM3, some of the characteristics of the seasonality have been elucidated, comparing experiments with and without interannually varying SSTs.  These have confirmed the dominant role of internal dynamics rather than organized external forcing in producing the seasonally dependent structures.

With growing interest in decadal-to-centennial climate variability in the ocean, the question arises as to how correlated are the trends in temperature and salinity. The answer to this question has important implications because temperature and salinity contribute to density variations in the ocean and thus determine the ocean circulation. Contributions to density from temperature and salinity anomalies have opposing signs. This means that if temperature and salinity were positively correlated, they would make less of a contribution to density than if they were negatively correlated.

Saravanan (GDS), in collaboration with Gokhan Danabasoglu (Oceanography Section, OS), Scott Doney (OS), and James McWilliams (OS/University of California, Los Angeles), has been carrying out a study of the relationship between temperature and salinity variations on decadal timescales. The primary data set for this research is a long control run of a coupled ocean-atmosphere model with a simplified two-level atmospheric model and an Atlantic-like sector ocean model. Dynamical aspects of the variability in the simplified representation of the climate system were analyzed in an earlier study, which identified oscillations of a decadal timescale.

In the continuation of that study, thermodynamic and tracer-related aspects of the variability have been analyzed. Results show that positive correlations between temperature and salinity are an ubiquitous feature of decadal oceanic variability. Although these correlations are relatively weak at the ocean surface, they increase dramatically with increasing depth. It can be established that these correlations are not due to atmospheric forcing but clearly due to some oceanic mechanisms. The exact details of these mechanisms are still under investigation.

Understanding of the nature of large-scale atmospheric motions is traditionally based on atmospheric models with the assumption of hydrostatic equilibrium, because the hydrostatic assumption is valid for low-frequency, large-scale motions. Moreover, the effect of earth's rotation plays a significant role. On the other hand, research on small-scale motions is primarily done using nonhydrostatic atmospheric models and the role of earth's rotation is often considered as secondary. Acoustics and lee-wave studies are examples of specialized research in the latter category. In fact, there is little interaction between large-scale and small-scale dynamicists, partly due to the adoption of different assumptions and interests in dealing with atmospheric dynamics in the two disciplines.

Availability of global and high-resolution observing systems and continued advancement in the development of high-speed computers enable us to examine local changes of weather and climate in the context of global atmospheric motions. Within this decade it will be possible to analyze and forecast weather and climate with global models having a horizontal grid size of 10 km or less. This will eliminate uncertainty inherent in one-way predictions with regional models embedded in the global domain. For such a very high-resolution global atmospheric model, it is important to consider nonhydrostatic effects that are traditionally neglected in the hydrostatic prediction models.

Short period (< 1 hour) atmospheric gravity waves as they propagate through the earth's upper mesosphere are known to be very important drivers of the mean flow and the thermal structure of the mesospheric region. However, what are the roles of large-scale acoustic waves in this connection, because the frequencies of large-scale acoustic waves are roughly the same as the frequencies of small-scale gravity waves?

Akira Kasahara (GDS) continues his research to explore the roles of nonhydrostatic dynamics in the context of global atmospheric modeling.  His recent study with Jian-Hua (Joshua) Qian (International Research Institute/Lamont Doherty Earth Observatory, Columbia University), which was just published in the October issue of Monthly Weather Review, on the normal modes of a global nonhydrostatic atmospheric model offers an introduction to this subject and provides basic tools to examine the roles of large-scale acoustic motions together with those of gravity waves and planetary scale motions.   For example, the mechanism of hydrostatic adjustment can be investigated, along with global geostrophic adjustment. As prediction models evolve from quasi-geostrophic models to primitive (hydrostatic) equation models, the next level of large-scale atmospheric models calls for inclusion of the roles of nonhydrostatic dynamics. In fact, the long-term effects of the cos (latitude) Coriolis terms neglected in the traditional primitive equation models for weather prediction and climate simulation are totally unknown and should be investigated.

While many of the characteristics of atmospheric variability on monthly and longer timescales can be understood in terms of linear dynamics, there is some evidence that certain properties of this variability are fundamentally nonlinear.  To quantify the importance of these nonlinearities, Branstator (GDS), together with Judith Berner (visitor, University of Bonn and Advanced Study Program) and Claudia Tebaldi (visitor, Geophysical Statistics Project) has continued the investigation of the phase space behavior of extended integrations of CCM0.  Building on their past findings that in a few directions in phase space probability density functions of CCM0 states are distinctly nonGaussian, they have worked on relating this nonGaussianity to the trajectories of model states in phase space.  Plots of mean trajectories indicate that there are multiple stagnation points in the phase space, and these turn out to be places where quasi-stationary episodes are concentrated.  It is useful to characterize the model states as belonging to circulation regimes depending on which stagnation point they are under the influence of, and typically a state remains under the influence of a given point for a few weeks.

To a surprising degree the distribution of states in phase space can be reproduced simply by considering a dynamical system composed of a deterministic term that consists of the observed mean velocities as a function of phase space position and a stochastic term that is position independent.  Nonlinearities in the deterministic term are crucial in reproducing the CCM0 probability density functions (PDF) in those directions where the PDFs are nonGaussian.

Sensitivity, Development, and Assimilation Studies

To assimilate observations of precipitation for the purpose of providing better initial conditions for weather forecast models, it is necessary to consider the error statistics of the model used to relate precipitation to the model prognostic fields of temperature, moisture, pressure, and winds.  These require consideration and estimation of the errors of convective parameterization schemes on the timescales at which the assimilation is performed, usually 1 to12 hours.  Since adequate and accurate precipitation verification data sets do not exist, these are difficult statistics to estimate. To obtain some reasonable bounds on these statistics, Errico (GDS) and Kevin Raeder (GDS) worked with data sets provided by David Stensrud (National Severe Storms Laboratory).  These were 12-hour forecasts produced using identical initial and boundary conditions applied to identical models, except for varying combinations of parameterization schemes for moist convective and the planetary boundary layer.  Statistical distributions of the 1-hour (and longer) accumulations of convective precipitation for the various forecasts were determined along with distributions of the differences between each distinct pair of forecasts. Results revealed that, even for forecasts as short as 3 hours, different convective schemes produce very different precipitation amounts. The ratios of amounts produced by pairs of schemes at the same grid points were approximately logarithmically distributed, with standard deviations corresponding to ratios of 3 or more.  These results suggest that model errors should not be neglected in assimilating precipitation observations.

Investigation of the generalized stability of weather patterns, particularly with regard to estimating our ability to forecast unstable systems, was continued by Errico (GDS) and a number of collaborators.  With Martin Ehrendorfer (visitor, University of Vienna), he has determined spectra of singular values for four widely different synoptic cases using the Mesoscale Adjoint Modeling System (MAMS2). This study compared the effects of moist and dry physics and contrasted norms used to measure perturbation sizes.  It revealed that most singular vectors damp in time and that the effects of moisture on them can be profound, even when moisture itself is not measured by the norms. With Ronald Gelaro and Carolyn Reynolds (both of Naval Research Laboratory), Errico (GDS) also showed that the local growth of Lyapunov vectors is determined primarily by their degree of projection onto the local leading singular vectors.

Two synoptic studies were conducted using MAMS2.  One was performed by John Lewis (Desert Research Institute), Raeder (GDS), and Errico (GDS), demonstrating how to use and interpret adjoint sensitivity fields for synoptic studies.  Specifically, they determined the sensitivity of moisture flux through an atmospheric window located in Texas.  It showed some unsurprising results, like the dominant sensitivity to SST along the trajectory of incoming air, and some more surprising results regarding smaller scale antecedent structures and regions of negative sensitivity to the SST.  Their work also includes a careful addressing of the issue of using a physically appropriate norm to describe the sensitivity.

The second synoptic study was performed by Errico (GDS), Raeder (GDS), and Luc Fillion (visitor, Atmospheric Environmental Service, Canada). They determined sensitivities of barotropic vorticity and convective and non-convective precipitation to antecedent model initial conditions using MAMS2. The study revealed that barotropic vorticity and precipitation rate at the end of the forecast were generally most sensitive to the initial temperature field for both summer and winter forecasts longer than an hour. In many other respects, however, the sensitivities to the different forecast aspects were notably different, including the temporal behavior of the magnitudes of the sensitivities. The results also revealed the importance of the moisture field for determining the vorticity in precipitating cyclones and the least importance of the wind divergence field, when the uncertainty of the fields as well as sensitivities were considered.  This study was primarily motivated by data assimilation issues.  For that purpose, the main result was that temperature adjustments should not be neglected when fitting initial fields to precipitation observations.