Contact Information

Tel: 303-497-1366

Fax: 303-497-1700

Postal Address:               P.O. Box 3000          Boulder, CO 80307-3000

Street Address:              1850 Table Mesa Dr.    Boulder, CO 80303

Electronic Mail: dme@ucar.edu

Global Dynamics Section

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 Climate Model (RegCM), and Climate System Model (CSM); and (3) sensitivity analyses of numerical prediction models to atmospheric initial and boundary conditions using variational techniques that will aid in the design of improved methods of data assimilation, for both conventional and non-conventional meteorological data, e.g., precipitation, soil moisture, and sea surface temperature (SST).

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 continue their interest in the inherent predictability of atmospheric phenomena and utilize 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) reinvigorated GDS efforts in the predictability of synoptic time and space scales with particular emphasis on the predictability of precipitation. The effort in GDS will focus on two main topics: the inherent predictability of synoptic scale precipitation and the influence of the loss of predictability in synoptic and larger scales on mesoscale processes. With Ronald Errico (GDS) and Joseph Tribbia (GDS), David Baumhefner (GDS) conducted rigorous experiments with four versions of CCM to evaluate potential differences in synoptic scale predictability error growth (PEG). CCM2 and 3 were compared at two resolutions, T42 and T63. Five 10-member cases were used to evaluate differences in PEG and significant differences were found. These differences ranged from 1.5 to 2.3 days in the doubling times for analysis error sized initial perturbations. In addition, the climate of each model was evaluated with respect to high-frequency variance simulation as an independent check against observed variance. A correlation between weak variance simulation and slow PEG was verified in all cases. CCM3-T42 results showed only 50% of observed variance in the 0- to 2-day frequency band, which corresponded with the slowest PEG of 2.3 doubling. The scale interaction in PEG was examined in a pilot experiment using the CCM2-T63 model. Only the large scales (greater than T30) were perturbed and compared to runs where only the small scales (less than T30) were introduced. Very rapid growth appeared in the unperturbed scales indicating differences from "classical" PEG when instabilities exist. Examination error plots showed that some errors arise from distinctly different scales of initial error. Another important area of applied predictability research is the quantification of the uncertainty in medium-range weather forecasts through ensemble techniques. Baumhefner continues his examination of three ensemble forecasting techniques: the bred vector approach used at the National Centers for Environmental Prediction (NCEP), the singular vector approach used at the European Centre for Medium-range Weather Forecasts (ECMWF), and a random perturbation method designed in GDS using CCM.Baumhefner expanded CCM forecast cases to 61 for CCM3-T42 for direct comparison with NCEP and ECMWF ensemble runs during the 1995-1996 Winter. The ensemble forecast skill and dispersion were evaluated and compared for daily data. CCM dispersion was slower and weaker than the operational counterparts, and the useful forecast skill limits were reached about a day sooner.

The relationship between ensemble dispersion to forecast skill was calculated for day 3 forecasts over the U.S. for the entire Winter of 1995-1996. NCEP and ECMWF both show high correlations (~0.60). Time variations in dispersion also matched well between the two methods of perturbation. Ratio of error variance to spread variance was near one in both operational schemes, indicating an imbalance in either the size or growth of predictability error early in the forecasts.

The early motivation for ensemble prediction studies in GDS is 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. Baumhefner, with Tribbia, continues 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 will be tested for skill in a three-month timeframe. Sixteen winter cases are run with 10-member ensembles using CCM3. These forecasts are all forced by observed sea surface temperatures (SSTs). In the continuation of this work, 16 cases of January-February-March (JFM) seasonal forecasts using observed initial conditions are compared with forced runs using initial conditions with no more than climatological relationship to observations. Ten member ensembles were used in both experiments. The results showed some positive response to initial condition. Weakly forced cases were most likely to show increased skill. While the sample size was not large enough to prove statistical significance, Pacific/North American (PNA) index scores showed an intriguing long-term memory of the initial state.

Research also continues on the predictability of climate variations. The coupled ocean-atmosphere interaction and predictability associated with the tropical El Niņo phenomenon motivated researchers to seek analogous phenomena in the midlatitudes as well. Are there midlatitude coupled ocean-atmosphere modes? Is there significant predictability in the midlatitudes? Ramalingam Saravanan (GDS), Gokhan Danabasoglu (Oceanography Section, OS), Scott Doney (OS), and James McWilliams (OS/University of California, Los Angeles) carried out a study addressing these questions using an idealized model of the ocean-atmosphere system. The atmosphere was represented using a global two-level eddy-resolving primitive equation model with simplified physical parameterizations. The ocean was represented using a state-of-the-art ocean general circulation model (GCM), the NCAR Ocean Model, configured in a simple Atlantic-like sector geometry. In addition to a coupled integration using this model, uncoupled integrations of the component oceanic and atmospheric models were also carried out to elucidate the mechanisms behind midlatitude variability.

The SST in the equilibrium state of the idealized coupled model exhibited two dominant modes of variability: (1) a passive oceanic red noise response to stochastic atmospheric forcing, and (2) an active oceanic mode of variability that was partially excited by atmospheric forcing and was associated with a periodicity of 16 to 20 years. True coupled ocean-atmosphere modes do not appear to play any quantitatively significant role in the midlatitudes, due to the fundamentally different nature of the atmospheric dynamics in the midlatitudes as compared to the tropics. However, coupling to the atmosphere does play an important role in determining the spatial and temporal characteristics of the oceanic variability. In particular, uncoupled oceanic integrations showed that the amplitude and the period of the active oceanic mode were quite sensitive to the surface boundary conditions.

A statistical assessment showed that the midlatitude atmospheric predictability was modest compared to the predictability associated with tropical phenomena such as El Niņo. (This table shows predictable variance fraction F of the projection time series of dominant empirical orthogonal functions (EOFs) of oceanic meridional overturning streamfunction psi and sea surface temperature (SST) computed using first-order and second-order autoregressive (AR(1) and AR(2)) models. 

This predictability arose from the atmospheric response to oceanic modes of variability rather than from coupled modes. There was significant oceanic predictability on interannual timescales but not on decadal timescales.

Climate predictions of a different type are standard in the examination of potential climate change associated with increases in greenhouse gas concentrations. Akira Kasahara (GDS) collaborates with scientists of the Central Research Institute of Electric Power Industry (CRIEPI) in the investigation of the impacts of global warming on various aspects of tropical cyclones (TCs). Last year, when the scientists of NCAR and CRIEPI conducted the 125-year climate simulation under a scenario of increasing CO2 using the CSM, it was thought that we would be able to analyze the results of that experiment to find the impacts of increased atmospheric CO2 on the climatological statistics of TCs. It turned out that the CCM3, which is the atmospheric component of CSM, did not produce TC-like vortices unlike CCM2. Since the standard (T42) CCM2’s ability to produce TC-like storms is well documented, the CRIEPI scientists (Jun-Ichi Tsutsui and Hiromaru Hirakuchi) decided to conduct an experiment with the T42 CCM2 that consists of two long-term simulations: one is a control run using specified climatological SST data, and the other is a 2×CO2 run forced by the SSTs representative of the doubled CO2 state obtained from the 125-year CSM simulation experiment mentioned earlier. The most noticeable change of simulated TC activity in the 2×CO2 run was observed in the  western North Pacific, where the frequency increased by about 30% between 10N and 20N. Large-scale circulation and thermal structure indicated changes favorable to TC development. This propensity, however, was not clear in other ocean basins, such as the South Pacific. Over the eastern North Pacific and North Atlantic, both the frequency and intensity of TCs tend to decrease. Research is underway to investigate why there are such drastic differences in the model’s ability to produce TC-like vortices between the CCM2 and CCM3.

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 to 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.

Investigations of variability on longer than synoptic timescales usually focus on the winter season, when the strongest circulation anomalies occur. Grant Branstator (GDS) developed a three-pronged study that sought to elucidate properties of variability on interannual timescales during all seasons of the year. The first part of this investigation simply quantified the basic properties of interannual variability as a function season. He found that there was a distinct seasonal cycle of amplitude, spatial scale, and structure for monthly and seasonal anomalies. [Not only were summer circulation anomalies weaker than their winter counterparts, they are of distinctly smaller scale, they had a somewhat different geographical concentration, and their phase tilts and aspect ratios were different.] The second component of this study dealt with finding dynamical explanations for the observed features described by the first part of the study. Analysis of CCM3 experiments found that that model does a good job of reproducing most of the attributes observed in nature and that it can do this even if interannual variability in SST anomalies was not included in experiments. This result indicated that the seasonal characteristics of interannual variability were largely the result of intrinsic dynamical processes. Experiments with a stochastically-driven barotropic counterpart to CCM3 suggested that the internal processes of most importance were primarily those associated with the linear influences of the seasonally-dependent climatological state on low-frequency perturbations. [This result was consistent with the observational finding that anomalies in all seasons tend to have structures that can barotropically draw energy from the climatological mean state, even though the structures needed to do this vary with the seasons.] The third part of this study of seasonality (undertaken in conjunction with Jorgen Frederiksen (Commonwealth Scientific and Industrial Research Organization (CSIRO), Division of Atmospheric Research, Australia)) determined to what degree the observed characteristics of low-frequency seasonality can be captured by the eigenmodes of a time-dependent linear operator based on the nondivergent barotropic vorticity equation. It found that when linearized about a state based on the seasonally-varying state observed in nature, the leading eigenvector of such a model had an amplitude and structure that varies with the seasons in much the same fashion that observed anomalies do.

In the past Branstator pursued the notion that for some purposes it may be advantageous to construct dynamical models of the atmosphere using empirical techniques rather than from physical principles. More recently he tested the possibility of hybrid models in which some terms were empirically determined while others were based on physical laws. In one such approach, developed jointly with Ulrich Achatz (Institute for Atmospheric Physics, Germany), the conventional nonlinear terms from a quasigeostrophic system were combined with linear terms determined using observations and an optimization principle. For testing purposes, this semi-empirical approach was applied to GCM-generated data rather than observations. It turns out that even though the model contained none of the explicit physical parameterizations of the GCM, its climate was very similar to that of the GCM. Even sensitive measures like the geographical distribution of transient eddy fluxes and the structure of recurring low-frequency disturbances were similar to corresponding quantities in the GCM. Furthermore, the model’s circulation statistics continued to strongly resemble those of the GCM even when its spatial degrees of freedom were reduced to as few as 20. [This model gives hope that seasonal-to-interannual variability can be largely understood in terms of a highly reduced system. It should also be useful for performing very long experiments (because of its high computational efficiency) and, with its simplified dynamics, may help to identify some of the key processes contributing to prominent intraannual and interannual features.]

One process that strongly influences the behavior of prominent seasonal anomalies is the feedback from fluxes produced by synoptic transients. Branstator is interested in this feedback and its representation in simple dynamical models. As part of this effort, he developed a way to quantify the observed feedback in such a way that it can be compared to parameterized feedbacks. First, by using analogue techniques he begins to determine to what degree transient eddy feedbacks are a function of the time average state. It now seems that (in CCM0, the model whose statistics he is examining) about 50% of the feedback variance can be explained by the time average state. Second, since the parameterization he is developing is linear in the sense that it is a linear function of time-averaged circulation anomalies, he found the observed linear relationship between time-mean states and concurrent transient eddy feedbacks using regression. [This observed linear relationship is the function that he would like to be able to reproduce using a parameterization. It condenses into a single matrix the complex interplay between low and high frequencies one would like a parameterization to capture. (Interestingly, though initially envisioned as a verification tool, the regression-derived relationship by itself may prove to be a very effective parameterization provided one is considering fairly low amplitude climate anomalies. He is currently testing this new concept by inserting the regression parameterization into a conventional linear barotropic planetary wave model.)]

A similar type of parametric relationship is the goal of research focusing on longer timescale, coupled variations in the climate system. On these timescales, the climate system acts as a "heat engine" that transports heat from the warm equator to the cold poles. The two fluid components of the climate system, the atmosphere and the ocean, both contribute significantly to this meridional heat transport. Baroclinic processes in the atmosphere, such as the Hadley circulation, and extratropical transient/stationary waves, play a crucial role in transporting heat meridionally, as does the thermohaline circulation in the ocean. The partitioning of the total poleward heat transport between the atmosphere and the ocean has to be considered an issue of great significance for understanding decadal to centennial scale climate change.

Saravanan and Gudrun Magnusdottir (University of California, Irvine) carried out a study of the relationship between atmospheric meridional heat transport and the meridional gradient of the zonally-averaged SST, using the CCM3 atmospheric GCM. The atmospheric heat transport in CCM3 is computed for five different configurations of implied heat transport in the ocean. The implied oceanic heat transport varied by changing the meridional gradient of SST. Climatological SSTs were employed for the control run. The other runs differed in that a zonally-symmetric component is added to or subtracted from the climatological SST field. The meridional structure of the variation in SST gradient is based on the observed change in zonally-averaged SST over the last century.

The results showed that in the annual mean, the atmosphere adjusted, so that the total meridional heat transport (by atmosphere and ocean) is relatively insensitive to the change in zonally-averaged SST. Interannual variability in the annual mean heat transport
was minor in each of these cases. There was a large degree of compensation, even between the different components of atmospheric heat transport, such that changes in latent heat transport usually go hand in hand with changes in dry static energy transport of an opposite sign. The radiative flux at the top of the atmosphere was effected the most by the change in SST in the tropics, where the shortwave component showed a strong negative feedback and the longwave component showed a weak positive feedback.

Sensitivity, Development, and Assimilation Studies

GDS has been at the forefront of scientific applications of variational techniques in large-scale dynamic meteorology and forecasting. The main applications are in the study of the smooth assimilation of nonstandard initial data into forecast models and to quantitative sensitivity analysis. Adjoint models are the only computationally-reasonable tool for determining the general sensitivity of most measures of aspects of model output with respect to small perturbations of model input. Knowledge of such sensitivity is critical for a wide range of applications, including data assimilation, targeting observations, predictability investigation, and general guidance in determining "what is important." The primary adjoint model used in GDS is version 1C of the Mesoscale Adjoint Modeling System (MAMS) developed by Errico and Kevin Raeder (GDS).

Errico, Luc Fillion (visitor, Atmospheric Environment Service, AES, Canada), Martin Ehrendorfer (visitor, ECMWF), Rolf Langland (Naval Research Laboratory, NRL), and members of NCAR’s Geophysical Statistics Project (GSP) examined the theoretical relationships among singular vectors, Lyapunov vectors, bred modes (as produced at NCEP), and initial condition errors. These pertain to several operational and proposed applications, including ensemble forecasting, data assimilation, and observation targeting. Their analysis examined the theoretical basis of techniques for initializing ensembles, comparisons of "optimal" perturbations produced using different norms to measure optimality, and the dynamic balances of some optimal perturbations.

Another application of adjoint methods is in data assimilation. Errico, Doug Nychka (GSP), Zhan-Qian Lu (GSP), and Fillion examined the statistical aspects associated with the variational assimilation of precipitation data. They used the common nonconvective precipitation model that has a first-order discontinuity at 100% relative humidity. For an assimilation system they considered a general, Bayesian formulation. They showed that precipitation assimilation systems described in the literature mistakenly ignore model error statistics and the constraint that precipitation was non-negative in defining an appropriate measure of fit of observations to predicted values. For the specific problem of the assimilation of precipitation data for analyzing temperature and moisture fields, they focused on non-Gaussian effects caused by the nonlinearity of precipitation models that in turn influence the choice of criteria for selecting a "best" analysis.   More traditional sensitivity studies are also being carried out in GDS, as exemplified by the following examples.

After Tsutsui, Hirakuchi, and Kasahara completed a study on the sensitivity of two cumulus parameterizations on the prediction of TCs using the RegCM, Kasahara recognized the inability of either of the two schemes to adequately simulate both the tropical mean climate and tropical transient variability. Kasahara subsequently examined the history of cumulus parameterization to more fully appreciate the current state-of-the-art and the scientific route taken to reach this state. He wrote an essay on the early history of cumulus parameterization mainly in the 1960s to the early 1970s. The necessity for developing cumulus parameterization became evident to run primitive-equation atmospheric models with moist physical processes without an overwhelming growth of cumulus convection. There are two roots of somewhat independent procedures that achieve stable time integrations. One root is the concept of conditional instability of second kind (CISK) as a cooperative interaction theory between cumulus and large-scale motions, originating from TC modeling. The other root is the practice of convective adjustment to suppress conditional instability of the first kind, originating in general circulation modeling. Akio Arakawa (University of California, Los Angeles) made a notable contribution to unite the two approaches, which led to the so-called Arakawa-Schubert scheme in 1974 as one of the more sophisticated cumulus parameterizations even today. The irony to this history, however, is that even after almost a quarter century since the advent of the Arakawa-Schubert scheme, modelers are still struggling to find a satisfactory cumulus parameterization for climate models. This is because cumulus schemes that are quite suitable for medium-range weather forecasts turn out to be no longer good enough to simulate both the mean and transient aspects of the climate system. How to resolve this difficulty will remain as our challenge in the 21st century.

As can be seen from the above example, oftentimes sensitivity experiments result from a necessity for model development. In line with that, Kasahara and Jian-Hua "Joshua" Qian (GDS) are exploring a numerical framework for very high-resolution climate modeling.

Send mail to dme@ucar.edu with questions or comments about this web site.
Last modified: November 16, 1999