The objective of the Climate Dynamics and Predictability (CDP) Section  is to further develop the scientific understanding of the dynamics and predictability of large-scale atmospheric variability and coupled variability on time scales 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. CDP 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 Atmosphere Model (CAM) and Community Climate System Model (CCSM), and (3) sensitivity analyses of numerical prediction models to atmospheric initial and boundary conditions using variational and ensemble techniques that will aid in the design of improved methods of data assimilation, for conventional and non-conventional meteorological data, e.g., precipitation and sea surface temperature (SST).

The studies described below are highlights of the research in CDP 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. CDP 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 about the goals of the U. S. Navy predictability initiative has reinvigorated CDP efforts in the predictability of synoptic time and space scales with particular emphasis on the predictability of nonlinear events. Nonlinearity of climate models is necessary if models are to capture sensitivity seen in nature, and yet models are frequently run in a 'deterministic' mode. Exploitation of (nonlinear) model sensitivity to initial conditions may enable a model to demonstrate behavior observed in nature but strongly in a single, deterministic run; further, information from such ensembles of runs can help in distinguishing system sensitivity and model error.

Dave Baumhefner (CDP), Joseph Tribbia (CDP), and Ron Errico (NASA Global Modeling and Assimilation Office) addressed several basic questions regarding the sensitivity of different perturbation techniques to estimates of predictability error growth in ensemble forecast systems. The main goal was to compare the differences among two dynamically initiated ensembles and one generated from background statistics. They documented the initial structure of three different schemes and examined how the different perturbations evolve in time and compared their behavior.

The following questions were addressed in their study:

1. How different are the initial perturbations, and do they resemble measurements of analysis error?

2. Are the evolved estimates of predictability error growth (variance of the ensemble) influenced by the differences in initial perturbation techniques, and, if so, at what timescales?

3. Is the rate of error growth and its temporal behavior different among the three schemes?

4. Is there any consistency in the three dispersion estimates, especially locally, and what is the relationship to the synoptic circumstances?

In answering these questions, Baumhefner, Tribbia, and Errico used forecasts from the Winter of 1995-1996 from two operational centers (National Centers for Environmental Prediction, (NCEP) and European Centre for Medium-range Weather Forecasts, (ECMWF)) and the NCAR Community Climate Model 3 (CCM3) in ensemble prediction mode.

Surface boundary conditions also play an important role in determining atmospheric predictability. Foreknowledge of the SST or the sea ice distribution can help predict the evolution of atmospheric flow on timescales of months to years. Understanding the role of boundary conditions in predictability is an important area of research in CDP. Currently this research is focused on tropical Atlantic variability and on long-term rainfall trends in the Sahel region of Africa.

Alessandra Giannini (CDP and Advanced Study Program, ASP) and Ramalingam Saravanan (CDP) have analyzed an ensemble of integrations of an atmospheric general circulation model (the NASA Seasonal to Interannual Prediction Project (NSIPP) model) forced only by the observed record of SST over the time period 1930-2000. They have compared the long-term climate variability in the Sahel rainfall in observations and in the model. The correlation between the two time series of July-September rainfall, characterized by a distinct negative trend between the wetter than average 1950s and the progressively drier decades of the 1960s, 1970s, and 1980s, is 0.60. The fact that the historical climatic progression can be reproduced using SSTs as the only external forcing strongly suggests that the secular change in Sahel rainfall during the past century was not a direct consequence of regional environmental change, anthropogenic in nature or otherwise. By analyzing additional integrations in which the land-atmosphere interaction is disabled, it is seen that land-atmosphere feedback acts to amplify the ocean-forced precipitation signal. The recent drying trend in the semi-arid Sahel is attributable to warmer-than-average low latitude waters around Africa, which, by favoring the establishment of deep convection over the ocean, weaken the continental convergence associated with the monsoon and engender widespread drought from Senegal to Ethiopia.     

Just as the seasonal forecast community uses concepts and methodologies developed by the climate dynamics community to diagnose, interpret, and further develop its prediction capabilities, there is the potential for climate change researchers to apply these same concepts and techniques to climate change scenarios. Conversely, climate change scenarios provide a fertile testing ground for ideas developed by the climate dynamics community. As a step toward realizing this potential, Grant Branstator (CDP) has begun applying climate dynamics methods to a unique climate change dataset generated as part of the Dutch Challenge Project. In this project a version of CCSM1 was integrated on a Dutch supercomputer to produce an ensemble of 62 climate scenario realizations, each designed to simulate the years 1940-2080. Realizations differed from each other only in their initial state. The purpose of the experiment was to produce a large enough ensemble that the occurrence of extreme events could be studied. But, the large sample also makes the experiment ideal for analyzing the dynamical processes that are affecting the model climate. Branstator's analysis is aimed at analyzing the evolving experimental climate from the standpoint of regional changes in the mean state and changes in the characteristics of interannual variability. Preliminary indications are that a key to understanding these properties is the concept of planetary wave propagation, a process that has been studied extensively in climate dynamics investigations. Major features in the evolving mean state and evolving interannual variability are not local but rather are dynamically linked together to form a global pattern that is recognizable from previous theoretical and numerical studies of the dynamics of the atmosphere. Recognition of this fact should help determine what factors influence the regional character of the simulated climate change and help in the assessment of their realism.

Within CDP 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 NCAR CCSM against that of the observed climate system. Naturally, the aforementioned prediction studies can also be viewed in this latter context. Additionally, several particularly insightful samples of past CDP studies exemplifying these two types of diagnoses are detailed below.

Saravanan and Ping Chang (Texas A&M) have been studying aspects of ocean-atmosphere interaction in the tropical Atlantic. Coupling between the atmosphere and the ocean involves the exchange of both momentum and heat. Dynamic coupling, which relates to the momentum exchange, is believed to play the dominant role in the tropics, especially in phenomena such as the El Niño-Southern Oscillation (ENSO) in the tropical Pacific. However, thermodynamic heat exchange between the atmosphere and the ocean can also play a significant role in air-sea coupling. This is especially true in the tropical Atlantic, where dynamic coupling may be of secondary importance. Saravanan and Chang have studied the role of this thermodynamic air-sea coupling using an atmospheric general circulation model (CCM3) coupled to a slab ocean model. They find that thermodynamic coupling leads to amplification and increased persistence of surface wind variability in the deep tropical Atlantic region. This effect is anisotropic, being stronger in the meridional component than in the zonal component of the surface wind, which suggests a role for the wind-evaporation-SST feedback in this region.

Predictability/prediction experiments using observed December SST initial conditions were also carried out. These showed that thermodynamic coupling can lead to forecasts of north tropical Atlantic SSTs that are significantly better than persistence forecasts during the boreal spring. These results mean that thermodynamic coupling certainly leads to a richer, more complex set of interactions than a local, Hasselmann-type of red-noise model would imply.

Model comparisons with observations and amongst models from other research centers are often useful in gauging confidence in mechanisms discovered in numerical model integrations. Such a strategy is being used in answering the question as to why some El Niño and La Niña events have global impacts while others do not. An investigation being carried out by Branstator suggests that the answer may depend on a) the exact position of the tropical rainfall induced by an El Niño or La Niña event and b) dynamical consequences of the climatological wintertime subtropical jet. Experiments with NCAR's CCM and NASA's NSIPP GCM indicate that tropical Pacific heating can excite a pattern of variability that extends completely around the Northern Hemisphere, but that this is more likely to happen if the heating is located to the west of the dateline than if it is located east of the dateline. The hemispheric-wide pattern of interannual variability was first identified in an earlier study by Branstator and found to result from properties of the subtropical jet that extends around much of the Northern Hemisphere. An interpretation of the new GCM experiments is that this waveguide is more readily excited from western Pacific than from eastern Pacific forcing. Because El Niño and La Niña events occur near the region where the transition between global and sector responses occurs, exact details of the structure of the rainfall associated with a particular event can be crucial to the extent of their impact. Analysis of the observational record tends to support these findings from theory and GCM experimentation. If past El Niño and La Niña events are subdivided according to their longitudinal position, those that occurred farther to the west in the Pacific have tended to affect the entire Northern Hemisphere midlatitudes, including Asia, while those centered more to the east have tended to affect only the North Pacific and North America.

Model studies of unforced variability are also necessary to understand the nature of free climate fluctuations that are intrinsic solely to the atmosphere. As part of an effort to characterize the dynamical processes that shape climate variability, Branstator has continued a project that is analyzing the behavior of the most prominent circulation patterns of natural atmospheric variability. Working together with Judith Berner (ASP and University of Bonn) and Frank Selten (Royal Netherlands Meteorological Institute), he has considered two aspects of these patterns. One of these is the distribution of pattern amplitudes, a quantity of interest because, if it can be shown that there are preferred amplitudes, this fact could indicate an enhanced predictability of the system. Other investigators have studied pattern amplitudes in observations from nature and concluded that there are preferred values, thus supporting the commonly made conjecture that the climate system has multiple modes that result from multiple equilibria. The ongoing investigation at NCAR analyzes extremely long integrations of a GCM and concludes that past studies have been detrimentally affected by sampling errors. The new results indicate that the more likely situation is that the distribution of state amplitudes does not have multiple modes, though it is distinctly non-Gaussian. The non-Gaussianity is especially pronounced when variability on seasonal and longer timescales is considered. The second focus of the analysis considers a more direct indicator of the dynamics of prominent circulation patterns, namely, their evolution with time. It turns out that, though to a large degree many patterns evolve in an approximately linear fashion, distinct signatures of nonlinear behavior can be detected by finding the most common trajectories that the system takes through a phase space defined by the leading circulation patterns. Stochastic models indicate that these nonlinearities are responsible for the non-Gaussian distribution of pattern amplitudes. Thus, although traditional linear dynamical concepts are likely to be adequate for understanding much of the atmosphere's behavior on these longer timescales, the effects of nonlinearity and the enhanced predictability that may result cannot be ruled out. Furthermore, these effects appear to be most important for the most prominent and highest amplitude patterns used by the system, the very patterns that are most likely to come into play in problems ranging from seasonal forecasting to climate change.

Tribbia, Saravanan, and Jeff Lee (all of CDP) worked with a Significant Opportunities in Atmospheric Research and Science (SOARS) protégé, Erik Noble (Pennsylvania State University) to examine the impact of SST variations in the eastern tropical Atlantic on the prediction of El Niño events. The rationale for the study was some preliminary results that indicated a significant impact of the SST off the west coast of Africa on the tropical Pacific. The questions studied were the nature of this remote interaction and whether this remote impact of the Atlantic on the Pacific occurred rapidly enough to be useful in ENSO prediction. The preliminary results indicate that a full six months are necessary for the remote influence to impact ENSO predictions.

Data assimilation studies have been a component of CDP research for many years. In the past year a more coordinated NCAR-wide activity has been developed with CDP a hub of activities within CGD. Jeffrey Anderson (CDP/Mesoscale and Microscale Meteorology Division (MMM)), Kevin Raeder (CDP), and Hui Liu (CDP) have been teaming with Alain Caya, Chris Snyder, Dale Barker, Syed Rizvi, (all of MMM), Tim Hoar (GSP), Doug Nychka (GSP), and Tribbia (CDP) in an NCAR strategic initiative. The NCAR Data Assimilation Initiative (DAI) is creating and leading a research community for data assimilation where individuals benefit from sharing ideas, methodologies, and software tools as well as access to a data assimilation research testbed (DART). The DART facility is used to provide focus to both internal and external collaborations related to data assimilation. An initial version of DART was completed during Fiscal Year 03 and provided the exercises for the ASP summer colloquium that was co-sponsored by DAI. DART is also being used to build ensemble filter assimilation systems for the CAM, MMM's Weather Research and Forecast (WRF) regional model, Geophysical Fluid Dynamics Laboratory's (GFDL) Flexible Modeling System (FMS) global atmospheric model, and NCEP's operational global prediction system (GFS). Observing system simulation experiments have been performed for CAM, WRF, and FMS models this year revealing a variety of new insights into the models' capabilities and shortcomings and into the information content associated with different types of observations. For instance, Figure 2 shows an evaluation of the information content available using only observations of surface pressure with CGD's CAM model.

Figure 2: Results from synthetic observation (perfect model) assimilation experiments with the standard resolution (T42 L26) version of CAM used for climate prediction applications. The red curve shows the global mean of the climatological standard deviation of the U wind component as a function of model level from the surface to the model top. The other curves show the global mean RMS error from assimilations conducted using the DART system. The green curve comes from a set of 300 randomly located simulated radiosondes observed every 12 hours. The blue curve is from 1800 randomly located surface pressure observations available every 12 hours. Both cases produce analyses in which the error is reduced far below the models climatological standard deviation. The surface pressure observations (blue) are somewhat better near the surface while the radiosondes, although less dense horizontally, produce better results aloft.

Efforts are underway to allow real observations to be used in the DART framework. Initial tests with real data assimilation in CAM have recently begun, and a DART ensemble filter is being used with complete global observations in parallel tests with NCEP's GFS. At the same time, DAI is beginning efforts to evaluate the characteristics of new observational sets including GPS radio occultation observations by developing forward observation operators and an assimilation testing capability.

DAI is a leader in fundamental research on ensemble filter assimilation techniques. New methods for dealing with model and observational error and for facilitating the efficient use of small ensembles for assimilation in large models have been developed this year. These should provide the initial components of more general and flexible ensemble filter assimilation systems that are under development.

In addition to advances in assimilation, the next generation of atmospheric model dynamical cores will, in all likelihood, span a range of scales that will include those for which the hydrostatic approximation is questionable. This will require new understanding of global non-hydrostatic effects. In search of a more refined formulation of the dynamical models for global weather prediction and climate projection, Akira Kasahara (CDP) continues his research to understand the dynamics of a compressible, stratified, nonhydrostatic, deep and global atmospheric model that is free from traditional shallowness and hydrostatic approximations which are adopted mostly in present-day general circulation models. The main goal of his research is to explore under what conditions, or for what phenomena such additional dynamical refinements become necessary, considering the range of parameters relevant to other planetary atmospheres as well as our own. The most basic characteristics of dynamical models are revealed from the modes of oscillations as the solutions of dynamical systems linearized around the state of rest.

For these studies, Kasahara developed a method of calculating the modes of small-amplitude free oscillations of a nonhydrostatic deep atmosphere confined between two concentric spheres, including a complete representation of Coriolis force. Unlike the Laplace-Taylor problem for the hydrostatic primitive equations that can be solved by applying the separation of variables in the vertical and horizontal directions, this problem is non-separable and complicated. Nevertheless, its solutions can be obtained numerically using the combination of spectral expansion and finite-difference discretization by setting up a block eigenvalue-eigenfunction matrix form. A technical report is being prepared to describe details of the method of solution and a test of numerical schemes for an isothermal basic state as the first step toward extending the present approach to a more general basic state, including the variation of gravity acceleration with altitude.

Tribbia also has been investigating the limitations of the hydrostatic balance approximation in a different context—that of limited area modeling. With Roger Temam (Indiana University) and Antoine Rousseau (Universit'e Paris-Sud), he has been continuing the examination of approximate equations that break the strong constraint of hydrostatic balance. The reason for their interest is the well-known deficiency of the hydrostatic primitive equations, ill-posedness as an initial-boundary value problem. The ill-posedness of the system imposes severe restrictions on the applicability of the system for limited area regional climate modeling and the use of adaptive mesh methods. Temam and Tribbia developed a simple alternative, designated the delta model. In the extension of this work with Rousseau, they have examined the limit in which delta, which regularizes the initial-boundary value problem, approaches zero to better understand the nature of the boundary layer behavior that ensues. The end result is transparent boundary conditions in characteristic variables.