Recent Developments in Predictability and Dynamical Processes (PDP)

Research: A Report by the THORPEX PDP Working Group

Istvan Szunyogh1, Heini Wernli2, Jan Barkmeijer3, Craig H. Bishop4, Edmund Chang5, Patrick Harr6,

Sarah Jones7, Thomas Jung8, Naoko Kitabatake9, Peter Knippertz2, Shuhei Maeda9, Sharanya

Majumdar10, Conny Schwierz11, Olivier Talagrand12, Frederic Vitart8

1 University of Maryland, USA

2 University of Mainz, Germany

3 Royal Netherlands Meteorological Institute, The Netherlands

4 Naval Research Laboratory, USA

5 State University of New York at Stony Brook, USA

6 Naval Postgraduate School, Monterey, USA

7 University of Karlsruhe/ Forschungszentrum Karlsruhe Germany

8 European Centre for Medium Range Weather Forecasts, Reading, UK

9 Japan Meteorological Agency, Japan

10 University of Miami, USA

11 University of Leeds, UK

12 Ecole Normale Superieure, LMD Paris, France

Manuscript submitted to BAMS

April 1, 2008

Corresponding author:

Istvan Szunyogh, Institute for Physical Science and Technology, University of Maryland, Computer

and Space Science Bldg., College Park, Maryland 20742-2431, U.S.A, email: szunyogh@umd.edu

Abstract

THORPEX - A World Weather Research Programme was established in May 2003 by the WMO as a

ten-year international global atmospheric research and development program. THORPEX is a

component of the WMO World Weather Research Programme (WWRP) and it aspires to become a

successor to the fifteen-year Global Atmospheric Research Programme (GARP) that started in 1967.

While GARP aimed at discovering unknown details of atmospheric dynamics and eventually led to

dramatic improvements in the accuracy of weather forecasts, the stated goal of THORPEX is

“Accelerating improvements in the accuracy of one-day to two-week high-impact weather forecasts for

the benefit of society, the economy and the environment”. Entering its implementation phase in 2005,

five working groups have been established within THORPEX, among them the Predictability and

Dynamical Processes (PDP) Working Group. Discussions in this working group, including outreach

activities to the international community with the aid of “interest groups” led to a summary of recent

key achievements in the field of atmospheric dynamics and predictability and identified major

overarching themes of THORPEX PDP research. This paper provides a summary of a selection of

these themes and highlights the need for international research cooperation, including field experiments,

diagnostic and theoretical studies. Research on the key THORPEX PDP research themes aims at

bringing together experts from various sub-disciplines, such as short- and extended-range forecasting,

predictability theory and process studies of hazardous weather, wave and vortex perspectives of

atmospheric flows, and dry dynamics and complex physical processes.

Capsule

The THORPEX Predictability and Dynamical Processes Working Group gives an overview of high

priority overarching research issues that THORPEX is planning to address in the upcoming years.

Introduction

The Fourteenth World Meteorological Congress (Resolution 12) in May 2003 established THORPEX:

A World Weather Research Programme as a ten-year international global atmospheric research and

development program. THORPEX is a component of the WMO World Weather Research Programme

(WWRP) and it aspires to become a successor to the fifteen-year Global Atmospheric Research

Programme (GARP) that started in 1967. While GARP aimed at discovering unknown details of

atmospheric dynamics and eventually led to dramatic improvements in the accuracy of weather

forecasts, the stated goal of THORPEX is “Accelerating improvements in the accuracy of one-day to

two-week high-impact weather forecasts for the benefit of society, the economy and the environment”.

In February 2005, THORPEX entered its implementation phase with the approval of the

THORPEX International Research Implementation Plan (TIP). The TIP called for the establishment of

five working groups and charged these with the implementation of the International Science Plan. The

Predictability and Dynamical Processes Working Group (PDP WG) organizes and supports

international activities that involve collaboration between the research communities working on

numerical weather prediction (NWP) and theoretical meteorology.

The close relationship between theoretical meteorology and the development of operational

NWP models was perhaps best described by Peter Lynch (2006), one of the scientists who made

notable contributions at the intersection of the two areas: “There is a strong symbiosis between

numerical weather prediction and theoretical meteorology. Advances in our understanding of the

physics and dynamics of the atmosphere and ocean are soon exploited in computer models, and these

models themselves provide us with a powerful tool of exploring the behaviour of the real atmosphere

and ocean”. Operational NWP models provide the arguably most consistently accurate solutions of the

equations that govern atmospheric motions. Also, the NCEP/NCAR (Kalnay et al. 1996) and ERA40

(Uppala et al. 2005) reanalysis data sets, prepared by two of the leading operational NWP centers using

their operational analysis/forecast systems, have become fertile breeding grounds for atmospheric

process studies. One of the great potentials of THORPEX is that it provides an international

organizational framework to organize joint activities between the academic and operational

communities to strengthen the “strong symbiosis” for the benefit of society.

The members of the PDP WG are Craig Bishop, Sarah Jones, Thomas Jung, Shuhei Maeda,

Istvan Szunyogh (Co-Chair), Olivier Talagrand and Heini Wernli (Co-Chair). In addition, to broaden

participation, the PDP WG launched 8 Interest Groups (IGs), which are open to all members of the

community and are led by the co-authors of this paper. The topics addressed by the eight IGs can be

found at http://www.ucar.edu/na-thorpex/forum.html. The Interest Groups conduct their discussions

through web-based mailing lists hosted by the National Center for Atmospheric Research (NCAR),

Boulder. An archive of the messages submitted to the IGs can be reached through the links on the IG

website. Those who are interested in joining one or more of the IGs should send a message to Pam

Johnson at UCAR (johnsonp@ucar.edu).

The first task of the IGs was to compile a brief summary of the most important developments in

their respective field in the last five years and a list of specific research issues, which the members of

the given group considered to be the most important unanswered research problems in the given area.

A collection of the full IG reports will be published in a WMO THORPEX publication. In this paper,

we discuss those science issues from the reports that fit into one of the following five overarching

themes:

- Tropical and extratropical cyclones,

- Rossby wavetrains,

- Madden-Julian Oscillation

- The relationship between dry and moist atmospheric dynamics,

- Ensemble-based prediction of forecast uncertainties.

The section “Predictability and Dynamical Processes” briefly describes, in general terms, how

improved understanding of the dynamical processes can lead to better understanding of predictability.

The five sections that follow, discuss the five major themes in the order they are listed above. The box

“Planned Activities” gives a short overview of the relevant THORPEX research activities planned for

the second half of 2008 and 2009.

Predictability and dynamical processes

The atmosphere is a chaotic physical system, that is, solutions of the system of equations that

govern atmospheric motions are sensitive to the initial conditions. Small errors in the initial conditions

lead to a complete loss of predictability after a finite time integration of the equations. An average

predictability time limit can be defined, for instance, by the typical forecast time at which the root-

mean-square error of the ensemble mean forecasts becomes equal to the root-mean square error of the

forecast based on the climatology. While for a state-of-the-art numerical global weather prediction

system the average predictability time limit is about two weeks, the evolution of the errors in the

forecasts exhibits strong spatio-temporal variability.

The spatio-temporal distribution of the magnitude of the forecast errors is determined by three

factors: (i) the spatiotemporal distribution of errors in the initial conditions, (ii) the spatiotemporal

propagation, and amplification or decay, of the initial condition errors in the forecasts, and (iii) the

spatio-temporally varying contribution of model errors. The overall quality of a deterministic forecast

can be improved by reducing the magnitude of the errors in the initial conditions and by improving the

representation of the physical processes of the atmosphere in the numerical models. The overall quality

of an ensemble-based probabilistic forecast can be improved by improving the representation of the

uncertainties in the initial conditions and the forecast model. Since the effect of model errors can also

be taken into account by statistical post-processing techniques, such as calibration of the forecast

probabilities based on re-forecasts (Hamill et al. 2006), the development of improved post-processing

techniques is also essential.

We note that the accuracy of initial conditions can be improved by refining the observing

systems, the observing strategies and the data assimilation systems that provide the initial conditions by

assimilating the observed information. In THORPEX, activities related to these three issues are planned

and carried out primarily by the Data Assimilation and Observing Systems (DAOS) and Observing

Strategies (OS) Working Groups. Here we only note that remarkable advances have been made in these

areas in the last decade. Today forecasts for the Southern Hemisphere are of similar quality to those

for the Northern Hemisphere (Simmons and Hollingsworth 2002; also see Figure 11.8 in Lynch 2006

for an update). The dramatic improvement in the quality of the Southern Hemisphere forecasts reflects

the profound advances made in improving the observational coverage of the globe, developing highly

accurate satellite based observing sensors, and in improving the techniques to assimilate remotely

sensed observations.

Although intrinsically chaotic in nature, the atmosphere is not in a state of fully-developed

turbulence: a large root-mean-square forecast error in the short- or early medium-range (about 1-7

days) over the globe, or over a limited area forecast domain, is typically the result of the erroneous

forecast of a single or very few weather systems, e.g., mid-latitude cyclones. Thus, it is reasonable to

expect that a better understanding of the dynamics associated with the genesis, evolution and

interaction of such weather systems would lead to a better understanding of the limits of their

predictability, to their better prediction with the NWP models, and to a better representation of the

uncertainties associated with their forecasts in the ensemble prediction systems.

Tropical and extratropical cyclones

Tropical Cyclones. The general physical mechanisms leading to tropical cyclone (TC) formation, or

cyclogenesis, are yet to be fully elucidated, and they may differ from case to case. One may surmise

that the predictability of cyclogenesis is controlled by processes on a variety of scales, from convective

to synoptic scales. The nature by which these scale interactions yield a warm-core vortex with a closed

surface wind circulation remains unsolved.

The most significant advances in TC forecasting to date have been on prediction of the TC

location or track, using global models. Track predictions play an important role since their accuracy is

the primary metric for society, tracks can be predicted with some skill by global models, and track

forecasts are straightforward to verify. Over the past half-century, track forecasting has been viewed as

a predominantly synoptic-scale problem in which the ”steering current” can be approximated by some

vertical average of the wind field. The primary synoptic features associated with the steering of the TC,

such as subtropical high-pressure systems and mid-latitude and tropical upper-tropospheric troughs, are

well known (Carr and Elsberry 2000a,b). However, uncertainty remains in the predictability of these

synoptic systems, which can have a major effect on the translational speed and direction of the TC.

Internal dynamics within the TC core are expected to provide a secondary influence on the track in

most situations.

Investigations into predictability and error growth of TC track have been limited to date. A few

studies have used an entropy-based method to yield an e-folding timescale of track error growth. For

example, Aberson and Sampson (2003) used this metric to suggest that 5-day forecasts may exhibit

skill relative to that obtained by climatology and persistence (CLIPER). Concurrently, recent

improvements to track forecasting have led some operational centers to extend the range of track

forecasting from 3 days to 5 days. Recent results have also indicated that TC track forecasts based on a

multi-model consensus are more accurate than those based on single models (Goerss 2006).

Sophisticated multi-model ensembles have been developed to predict TC track and intensity (Weber

2005). Research into the use of ensemble prediction systems for TC forecasts must have high priority

in the future.

In contrast to predictability of TC track, only limited progress has been made on predictability

of TC structure, and consequently intensity, rainfall and wind distribution. A high-resolution, core-

resolving model that captures dynamical and physical processes within the TC is necessary for such

studies, and only one such model has been used to date for operational TC prediction (NOAA GFDL).

However, several high-resolution regional research models demonstrate promise for real-time

forecasting of hurricane structure (e.g., Chen et al. 2007; Davis et al. 2008), and such models will

enable numerous new predictability questions to be addressed over the next decade.

Extratropical transition of tropical cyclones. A TC travelling into the midlatitudes can

transform into a fast moving and intense extratropical cyclone i.e. undergo extratropical transition (ET).

This process is often associated with severe weather conditions and reduced predictability at the

location of the ET and further downstream at later forecast lead times (e.g. Harr et al. 2008). The

changes in the structure of the cyclone during the initial phase of ET are characterized by enhanced

asymmetries in the wind, cloud and precipitation fields (Kitabatake et al. 2007) so that structure

forecasts are important for countries directly affected by ET. ET events are often associated with large

variability in the skill of forecasts initialized from a sequence of analyses (Jones et al 2003). In

addition, forecasts of TC structure have been found to have larger errors during ET than for a mature

TC (Evans et al. 2006). Uncertainty in ET forecasts can arise due to variability in the structure and

location of both the decaying TC (Klein et al. 2002) and the midlatitude trough with which it interacts

(Browning et al. 2000; Ritchie and Elsberry 2007). The relative importance of processes, such as

sensible and latent heat fluxes, latent heat release, frontogenesis, and baroclinic energy conversion

varies between ET events. For example, enhanced upper-level ridging associated with latent heat

release during ET has been shown to result in enhanced precipitation (Atallah and Bosart 2003).

Extratropical Cyclones. Although the first important step toward exploring the processes that

lead to the formation of extratropical cyclones was taken more than a century ago by Margules, who

introduced the concept of available potential energy, systematic studies of extratropical cyclogenesis in

the real atmosphere, based on large statistical samples of events, could start only a few years ago when

the reanalysis data sets became available.

Several recent studies, based on reanalysis data, emphasized the importance of downstream

development, a process that was first described by Simmons and Hoskins (1979). In these studies,

cyclogenesis is triggered by downstream propagating packets of upper-tropospheric Rossby waves

although, unlike in the idealized experiments of Simmons and Hoskins (1979), the upper-tropospheric

waves do not obviously originate from an upstream baroclinic development. Hakim (2003) found that

surface cyclogenesis events over the western Pacific are preceded by wave packets that originate

poleward of the Himalaya Plateau and develop rapidly across the North Pacific to North America.

Chang (2005) examined western Pacific cyclogenesis during mid-winter (December to February) and

found that these events are strongly influenced by pre-existing upstream wave packets, which

propagate along both the northern (over Siberia) and southern (the sub-tropical jet over southern Asia)

waveguide. He also found that some of the cases involve interactions between pre-existing upper-level

wave packets with pre-existing low-level circulation anomalies. Danielson et al. (2004) examined 41

cold-season cyclones that intensified strongly over the eastern North Pacific Ocean and found that

about half of the cyclones are good examples of downstream baroclinic development.

Rossby wavetrains

Synoptic scale Rossby wavetrains play a central role in the downstream propagation of localized

influences in the atmosphere, and they have been shown to be prevalent in the midlatitudes of both

hemispheres during both warm and cool seasons (e.g., Chang and Yu 1999). Due to their dispersive

nature, such wavetrains can propagate much faster than the weather systems that trigger them initially.

Atmospheric wave trains can build causal relationships between weather systems that are distant in

both time and space (Figure 1).

Downstream cyclogenesis, which we discussed earlier, is only one important example for the

atmospheric motions Rossby waves can trigger downstream. Breaking Rossby waves (the final stage of

baroclinic wave development) have been linked to episodes of heavy precipitation over the European

Alps by Martius et al. (2006) and Grazzini (2007). These studies found that in some cases the precursor

waves to these breaking Rossby waves can be tracked up to eight days in advance over the Asian-

western Pacific region during the cool season, as well as from the US Southwest in spring. Grazzini

(2007) also studied the performance of the ECMWF forecasts system for these heavy precipitation

events and found that the forecasts had higher than average skill at the synoptic scales. Wave packets

can also penetrate into the (sub-) tropics, causing extreme precipitation and dust storms (e.g., Knippertz

and Martin 2005). Moreover, breaking Rossby waves can influence low-frequency variability of the

atmosphere (e.g., Nakamura and Fukamachi 2004; Enomoto 2004; Abatzoglou and Magnusdottir

2006).

Rossby wavetrains also play an important role in the downstream propagation of forecast errors:

errors in the forecast of an upstream weather event propagate downstream with increasing forecast

lead-time at the group velocity of the wavetrains. Accordingly, local improvements in the analysis, e.g.,

due to the assimilation of targeted weather observations, can have positive forecast effects far

downstream of the location of the analysis improvement when Rossby wavetrains are present in the

upper-tropospheric flow (e.g., Szunyogh et al. 2002).

In the extratropics, the triggering, propagation and breaking of Rossby wavetrains can be

explained by using a single atmospheric state variable, the isentropic potential vorticity (PV). The PV

distribution in the extratropical tropopause region is typically characterized by zonally oriented bands

of intense PV gradients near the dynamical tropopause (the 2-pvu isoline). These bands are collocated

with the upper tropospheric jet streams, and their length can reach up to several 1000 km. They can be

regarded as temporally evolving “PV waveguides” for the propagation of Rossby wavetrains (e.g.,

Schwierz et al. 2004), constituting a key planetary-scale atmospheric flow element, both for the

evolution of synoptic-scale weather systems and for the error propagation in NWP models. In the last

few years, the concept of waveguides, precursor disturbances, wave propagation and amplification, and

eventually downstream impacts has proven useful for several research issues on predictability and

dynamical processes. It also featured prominently in the THORPEX Science Plan (Shapiro and Thorpe

2004).

Waveguides can be disturbed in many different ways (Figure 1). Common to all these wave-like

disturbances is the generation of positive or negative PV anomalies along the waveguide that can

induce rapid downstream dispersion of the initial wave packet. One example for such a precursor is the

isentropic advection of a high-amplitude stratospheric PV vortex (i.e., a positive PV anomaly) from the

polar region, whose cyclonic circulation will induce a wave-like disturbance along the mid-latitude

waveguide. In many other situations, cloud condensational heating in the lower and middle troposphere

produces a negative PV anomaly with associated anticyclonic circulation at the tropopause level. This

heating can be associated, for instance, with ET, a warm conveyor belt ascending within an

extratropical cyclone, or a mesoscale convective system. PV anomalies can be generated also when the

upper-level divergent outflow from a TC or other organized convective system impinges on the sloping

midlatitude tropopause. Another mechanism for inducing upper-level waveguide disturbances is related

to the flow distortion by elevated large-scale topography (e.g., Greenland and the Himalaya). Chang

(2005) found that existing wave packets are often enhanced by surface cyclogenesis, such increasing

the potential strength of the downstream impact.

Does the existence of a precursor mechanism enhance the predictability of a severe downstream

weather event? In principle, the existence of a coherent upper tropospheric trigger does not guarantee

better predictability, since uncertainties in the prediction of the physical processes that affect the

initiation and propagation of the Rossby wave packet and in the prediction of the processes that

determine the lower-tropospheric conditions at the location of the downstream event can limit

predictability. However, the empirical evidence shown by Grazzini (2007) suggests that in a state-of-

the-art NWP model the representation of the dynamical processes along the waveguides is sufficiently

accurate to ensure higher than average predictability of some of the downstream events.

Madden-Julian Oscillation (MJO)

The stated goal of THORPEX is to improve forecasts up to two weeks lead-time, which is the typical

predictability time limit using a state-of-the-art NWP model. THORPEX recognizes, however, that for

certain atmospheric flow configurations the predictability time limit can be longer than that. An

atmospheric dynamical process that has the potential to lead to extended predictability in the

extratropics is MJO, which is the dominant mode of variability in the Tropics on time scales exceeding

one week and less than a season (e.g., Lau and Waliser 2005). The phrase “potential” is used here,

because global models have major difficulties with simulating and predicting the MJO, and it is not

possible to foresee all consequences of a better representation and prediction of the MJO in the models.

What we know with certainty are that (i) MJO has a significant impact on cyclogenesis in the

eastern North Pacific (Maloney and Hartmann 2000), the Atlantic (e.g., Mo 2000), the western North

Pacific (Sobel and Maloney 2000), the Australian basin (Hall et al. 2001) and the South Indian Ocean

(Bessafi and Wheeler 2006) and that (ii) while statistical predictive models of the MJO display useful

predictive skill out to at least 15-20 days lead time (Waliser et al. 1999; Lo and Hendon 2000; Wheeler

and Weickmann 2001), the predictability time limit of the MJO in NWP models is much shorter than

that (e.g., Vitart 2003). Global circulation models (GCMs) not only have difficulties with predicting the

MJO, but they also struggle to maintain a realistic MJO. Slingo et al. (1996) found that none of the

atmospheric GCMs that took part in the Atmospheric Intercomparison Project were able to capture the

spectral peak associated with the MJO. A recent evaluation of the coupled atmosphere-ocean models

that provided simulation results for the Intergovernmental Panel on Climate Change (IPCC) fourth

Assessment Report showed only slight improvement in the maintenance of the MJO (Lin et al. 2006).

The aforementioned results suggest that model errors are likely to be the main obstacle to

predicting MJO events. In fact, Vitart et al. (2008) found that using an improved representation of the

mixing in the upper ocean and a different cloud parameterization helped maintaining the MJO for 8

more days. More recently, Miura et al. (2007) found that a 7-km resolution atmospheric model running

on the Earth Simulator, a Japanese supercomputer developed for running more realistic global

simulations, was able to maintain an MJO with realistic structure over a period of 1 month. This result

suggests that significantly increasing the resolution of operational weather forecasting systems may

help to reduce model errors that currently hamper more realistic model forecasts of the MJO.

The only study, up to this day, that made an attempt to quantify the potential improvement of

the forecast skill in the extratropics due improved representation of the MJO is Ferranti et al. (1990).

This study found that an improved representation of the MJO in the ECMWF forecast model, achieved

in that case by relaxing the tropical circulation towards the analyses, could lead to a significant increase

of skill in the extratropics for more than 10 days. While this experimental design has the potential to

lead to an overestimation of the potential forecast improvement in the extratropics, as it assumes

perfect predictability in the Tropics, it would be desirable to carry out similar experiments with current

state-of-the-art models.

The role of diabatic processes

Diabatic processes, in particular the release of latent heat due to cloud condensation, but also surface

fluxes of sensible and latent heat, can have a significant impact on the structure, evolution and

predictability of weather systems. In terms of PV, latent heat release is primarily associated with

production of PV in the lower troposphere and destruction of PV above the level of maximum diabatic

heating. This indicates that the primary effect of moist processes is the generation or enhancement of

cyclonic circulations in the lower troposphere and anticyclonic circulations in the upper troposphere.

Diabatic processes have been known for a long time to play a primary role in the development

of such tropical weather systems as organized convection and TCs. In contrast, in the extratropics,

baroclinic instability has been traditionally regarded as the key mechanism for the development of

cyclones and their accompanying frontal structures, with diabatic processes playing a secondary,

typically intensifying, role. A more complex picture started to emerge, however, from a series of recent

studies that provide evidence that in a substantial number of cases diabatic processes (i) have crucial

impact on the formation of intense extratropical weather systems, (ii) have a major influence on the

downstream effects of these weather systems, and (iii) have a major influence on predictability in the

midlatitudes. In what follows, we review some of the most important results of these studies.

The idealized study of Lapeyre and Held (2004) found that while dry and not very moist

atmospheres are primarily dominated by the formation of baroclinic waves, very moist atmospheres are

dominated by finite amplitude vortices. These vortices are reminiscent of diabatic Rossby waves

(DRWs), which are diabatically regenerated, positive, low-level PV anomalies that propagate rapidly

along intense baroclinic zones. (See Moore and Montgomery (2004) for an overview and discussion of

the wave/vortex aspects of the phenomenon.) The significance of these vortices is that they can exist

and propagate without forcing from an upper-level anomaly and may act as precursors for “bottom-up”

cyclone intensification when they interact with an intense upper-level jet (Wernli et al. 2002). Despite

the coherent nature of the phenomenon, current NWP models have serious difficulties with predicting

DRWs. Diabatic processes are also thought to play a potentially important role in creating the initial

perturbations that can lead to rapidly growing “bottom-up” instabilities in regions of enhanced

baroclinicity (Hoskins et al. 2000). These structures can be captured by singular vector calculations and

are often used to identify the optimal locations for the deployment of targeted observations (e.g.,

Langland 2005) and to generate initial ensemble perturbations (e.g. Buizza 2006). The rapid growth of

these perturbations can be explained by PV unshielding, a process in which a positive PV anomaly

becomes unshielded by the eastward and westward displacement of an upper and lower negative PV

anomaly due to the advection by the zonal shear flow (Badger and Hoskins 2001).

Massacand et al. (2001) found the first example for the potentially important downstream

impact of diabatic processes in the extratropics. They showed that the transport of low-PV air into the

upper troposphere by a warm conveyor belt was crucial for downstream Rossby wave breaking, which

in turn led to an event of Alpine heavy precipitation. Knippertz and Martin (2005) observed a similar

link between an upstream conveyor belt and a heavy precipitation event over northern Africa. These

results indicate that diabatic processes can strongly alter the effects of an upstream extratropical

cyclone on the downstream large-scale flow. In such situations, an accurate simulation of the warm

conveyor belt, for instance in the western ocean basins, might be essential for a high-quality medium-

range forecast over the downstream continents. This hypothesis is supported by the study of Danielson

et al. (2004), who found that for those eastern North Pacific cyclones that were influenced by

downstream propagation of eddy energy from cyclones over the western North Pacific, the primary

energy source was a warm conveyor belt type ascent in the warm sector of the upstream cyclones.

Investigating the relationship between predictability and dynamical processes for the “no-

surprise” east-coast snowstorm of January 2000, Zhang et al. (2002) found that forecast errors at the

synoptic scales were preceded by errors associated with moist convection on scales smaller than 100

km. This process of upscale error propagation can be observed in both the models that use

parameterized convection and the models that explicitly resolve convection. Walser et al. (2004),

however, analyzing results from convection-resolving simulations of Alpine heavy precipitation events,

found that topographic triggering of the convection enhanced the predictability.

Ensemble-based prediction of forecast uncertainties

One of the most important post-GARP developments in NWP has been the implementation of

operational ensemble prediction systems, which started in the early 1990s at ECWMF and NCEP (e.g.,

Kalnay 2002; Palmer and Hagedorn 2006). A close collaboration between the operational and the

research centres in the area of ensemble prediction will be facilitated by the creation of the TIGGE

database. TIGGE has been a major investment of the operational centres to support the researchers

community, as it provides scientists unprecedented, nearly real-time, access to ensemble forecast

products of several operational weather prediction centres. (For more information on TIGGE visit

http://tigge.ecmwf.int.)

Despite the more than fifteen years of operational experience and the hundreds of research

papers written on the subject, questions on how ensemble forecasts should be created and used remain

open. A prominent difficulty is that the object of an ensemble-based prediction, basically a probability

or a probability distribution, is not better known a posteriori than it was a priori. In fact, the predicted

object has no objective existence and cannot be possibly observed at all (e.g., Toth et al. 2006). As a

consequence, validation of an ensemble prediction system requires careful statistical analysis of a large

number of events. It is of little use to speak of the quality of ensemble-based predictions of the

probability distribution on a case-to-case basis (e.g., Talagrand et al. 1997).

One potential approach to evaluate the performance of an ensemble is to evaluate the extent to

which the vector sub-space of ensemble perturbations spans the space of forecast error. For instance,

this approach was used to compare the performance of the ensemble prediction systems of NCEP and

ECMWF (Wei and Toth 2003). Kuhl et al. (2007) applied a similar approach in localized regions of the

atmosphere and found that the ensemble captured the space of forecast errors most efficiently in cases

of rapid error growth. We note, however, that this concept leads to a practical measure only if we

assume that the space of uncertainties is linear.

Experience accumulated with the operational ensemble forecasting systems suggest that present

scores saturate for a value of ensemble size N in the range 30-50, independently of the quality of the

score (e.g., Richardson 2001). On the other hand, theory tells us that according to chi-square statistics,

with N=30 and a true variance of 1, the sample variance has a 95% chance of lying between 0.56 and

1.57; i.e. variance estimates are very inaccurate. With N=100, the corresponding 95% confidence

interval (0.74,1.29) is significantly smaller. In the face of this theoretical fact, how can we explain our

aforementioned experience that the forecast scores saturate at N=30-50? There are at least four possible

explanations: (i) Scores have been implemented so far on probabilistic predictions of events or one-

dimensional variables (e.g., temperature at a given point). The situation might be different for

multivariate probability distributions, but for those the large verification sample needed to obtain stable

verification statistics may not be attainable. (ii) Probability distributions in the case of one-dimensional

variables are most often unimodal. The situation might be different for multi-modal probability

distributions, as produced for instance by multi-model ensembles. (iii) Saturation is due to the

characteristics of synoptic-scale atmospheric dynamics. The situation might be different for mesoscale

ensemble prediction. (iv) Lack of direct measure of accuracy of the predicted error variance.

One may also want to have a larger ensemble to better resolve the probability distribution or to

better predict rare events. However, these goals may be elusive: (i) Are there any users who will ever

care whether the probability for rain tomorrow is 123/200 rather than 124/200? (ii) The verifying

sample needed to check the reliability of a large ensemble may be prohibitively large. Assume that the

predicted probability is 1/N for a given event E and that we have one 10-day forecast every day. E has

to occur roughly cN/10 times, where c is of the order of a few units, before reliability can be assessed.

If the event occurs about four times a year, we must wait 10 years for N = 100, and 50 years for N =

500 (c = 4). We can conclude that the reliable probabilistic prediction of an even moderately rare event

is simply impossible. The problem of size of verifying sample will remain, even if it can be mitigated

to some extent by using reanalyses or re-forecasts for validation. More work is clearly needed to

identify the useful size of an ensemble and the practically attainable size of a verification sample.

There is a strong consensus between members the operational and research communities that

the ensemble approach is the way of the future, for both data assimilation and weather prediction. But

no strong views have emerged, yet, on the format of the future ensembles. Must they consist of

mutually equal elements (or of subsets of mutually equal elements, produced for instance by different

models), or must they contain, as is the case now, individual elements of higher statistical quality

(higher resolution, or control forecasts)? The present situation is somewhat hybrid, the predicted

ensemble being a kind of auxiliary to a statistically more accurate higher resolution forecast. Must we

aim at a situation where the predicted object will be a probability distribution?

Activities Planned for 2008 and 2009

- THIS SECTION SHOULD APPEAR IN THE PAPER AS A BOX

THORPEX Pacific Asian Regional Campaign (T-PARC). T-PARC will be the second major, and

so far the largest, field campaign of THORPEX, which follows the Atlantic-THORPEX Regional

Campaign (A-TReC) in 2004. T-PARC will have two major phases: a summer-autumn phase that will

overlap with the 2008 typhoon season and a winter phase that will take place in January and February

of 2009. The first phase will take part in conjunction with the Tropical Cyclone Structure 08 (TCS08)

experiment and will focus on the life cycles of typhoons in the Northwest Pacific. It will include both

basic research components, mainly focusing on dynamical processes associated with tropical

cyclogenesis, ET, and the downstream effects of ET. Analysis/forecast experiments will be carried out

to study the predictability of the track, intensity, and re-curvature of TCs, and to investigate changes in

extratropical predictability due to ET. The winter phase will focus on the genesis and downstream

effects of extratropical cyclones. The detailed science plan, experimental design overview, and the

latest information on T-PARC can be found at http://www.ucar.edu/na-thorpex/PARC.html.

Year of Tropical Convection (YOTC). YOTC is the first planned collaborative activity between

THORPEX and the World Climate Research Program (WCRP). The recognition that poor

representation of tropical convection in the global circulation models is one of the major road blocks to

further improve the quality of both numerical weather predictions and climate simulations motivated

THORPEX and WCRP to organize the Workshop on the Organization and Maintenance of Tropical

Convection and the Madden Julian Oscillation in Trieste, Italy, in March 2006 (Moncrieff et al. 2007).

One of the recommendations made by the workshop was the organization of YOTC, which will bring

together observational information from already existing platforms and analysis and forecast

information from operational centers for a one-year period. YOTC will also take advantage of the

newly available research and modeling framework of explicitly cloud-system resolving models

(CSRMs). CSRMs are expected to lead to improved parameterization for convective organization and

the better representation of upscale effects of organized precipitating tropical convection (Moncrieff

and Liu 2006). A detailed science and implementation plan of YOTC is currently under development

by a science planning group led by Mitch Moncieff of NCAR and Duane Waliser of JPL/CALTECH.

The latest version of this document is available at

http://hydro.jpl.nasa.gov/tmp/WCRP.WWRP.YOTC.scienceplan.pdf.

References

Abatzoglou, J. T., and G. Magnusdottir, 2006: Planetary wave breaking and nonlinear reflection:

Seasonal cycle and interannual variability. J. Climate, 19, 6139-6152.

Aberson, S.D., and C.R. Sampson, 2003: On the Predictability of Tropical Cyclone Tracks in the

Northwest Pacific Basin. Mon. Wea. Rev., 131, 1491-1497.

Atallah, E.H., and L.F. Bosart, 2003: The Extratropical Transition and Precipitation Distribution of

Hurricane Floyd (1999). Mon. Wea. Rev., 131, 1063–1081.

Badger, J. and B. J. Hoskins, 2001: Simple initial value problems and mechanisms for baroclinic

growth. J. Atmos. Sci., 58, 38-49.

Bessafi, M. and M.C. Wheeler, 2006: Modulation of South Indian Ocean Tropical Cyclones by the

Madden-Julian Oscillation and Convectively Coupled Equatorial Waves. Mon. Wea. Rev., 134,

638-656.

Browning, K. A., A. J. Thorpe, A. Montani, D. Parsons, M. Griffiths, P. Panagi, and E. M. Dicks,

2000: Interactions of tropopause depressions with an ex-tropical cyclone and sensitivity of

forecasts to analysis errors. Mon. Wea. Rev.,128, 2734-2755.

Buizza, R., 2006: The ECMWF ensemble prediction system. In Hagedorn and Palmer, 2006. 459-488.

Carr, L.E., and R.L. Elsberry, 2000a: Dynamical Tropical Cyclone Track Forecast Errors. Part I:

Tropical Region Error Sources. Wea. Forecasting, 15, 641-661.

Carr, L.E., and R.L. Elsberry, 2000b: Dynamical Tropical Cyclone Track Forecast Errors. Part II:

Midlatitude Circulation Influences. Wea. Forecasting, 15, 662-681.

Chang, E. K. M., 2005: The impact of wave packets propagating across Asia on Pacific cyclone

development. Mon. Wea. Rev., 133, 1998-2015.

Chang, E. K. M, and D. B. Yu, 1999: Characteristics of wave packets in the upper troposphere. Part I:

Northern Hemisphere winter. J. Atmos. Sci., 56, 1708-1728.

Chen, S.S., J.F. Price, W. Zhao, M.A. Donelan, and E.J. Walsh, 2007: The CBLAST-Hurricane

Program and the Next-Generation Fully Coupled Atmosphere-Wave-Ocean Models for

Hurricane Research and Prediction. Bull. Amer. Meteor. Soc., 88, 311-317.

Danielson, R. E., J. R. Gyakum, and D. N. Straub, 2004: Downstream baroclinic development among

forty-one cold-season eastern North Pacific cyclones. Atmosphere-Ocean, 42, 235-250.

Davis, C. A., Wang, W., Chen, S., Chen, Y., Corbosiero, K., DeMaria, M., Dudhia, J., Holland, G.,

Klemp, J., Michalakes, J., Rotunno, R., Xiao, Q., 2008: Prediction of landfalling hurricanes

with the advanced hurricane WRF model. Monthly Weather Review. (In Press).

Enomoto, T., 2004: Interannual variability of the Bonin High associated with the propagation of

Rossby waves along the Asian Jet. J. Meteor. Soc. Japan, 82, 1019-1034.

Evans, J. L., J. M. Arnott, and F. Chiaromonte, 2006: Evaluation of operational model cyclone

structure forecasts during extratropical transition. Mon. Wea. Rev., 134, 2054-3072.

Ferranti, L., T. N. Palmer, F. Molteni and E. Klinker, 1990: Tropical-extratropical interaction

associated with the 30-60 day oscillation and its impact on medium and extended range

prediction. J. Atmos. Sci., 125, 2177-2199.

Goerss, J. S., 2006: Tropical cyclone track forecasts using an ensemble of dynamical models. Mon.

Wea. Rev., 135, 1985-1993.

Grazzini, F., 2007: Predictability of a large-scale flow conducive to extreme precipitation over the

western Alps. Meteorol. Atmos. Phys., 95, 123-138.

Hakim, G.J., 2003: Developing wave packets in the North Pacific storm track. Mon. Wea. Rev., 131,

2824-2837.

Hall, J. D., Matthews, J. A. and D. J. Karoly, 2000: The Modulation of Tropical Cyclone Activity in

the Australian Region by the Madden-Julian Oscillation. Mon. Wea. Rev., 129, 2970-2982.

Hamill, T. M., J. S. Whitaker, and S. L. Mullen, 2006: Reforecasts: An important dataset for improving

weather predictions. Bull. Amer. Meteor. Soc., 87, 33-46.

Harr, P. A., D. Anwender and S. C. Jones, 2008: Predictability associated with the downstream impacts

of the extratropical transition of tropical cyclones: Methodology and a case study. Mon. Wea.

Rev., in press

Hoskins, B. J., R. Buizza, and J. Badger, 2000: The nature of singular vector growth and structure. Q.

J. Roy. Meteorol. Soc., 126, 1565-1580.

Jones, S. C., P. A. Harr, J. Abraham and Co-authors, 2003: The extratropical transition of tropical

cyclones: Forecastchallenges, current understanding and future directions. Wea. Forecasting,

18, 1052-1092.

Kalnay, E., 2002: Atmospheric modelling, data assimilation and predictability, Cambridge University

Press, Cambridge, 364 pages.

Kalnay, E., M. Kanamitsu, R. Kistler and coauthors, 1996: The NCEP/NCAR 40-year reanalysis

project. Bull. Amer. Meteor. Soc., 77, 437-471.

Kitabatake, N., Hoshino, S., Bessho, K., and Fujibe, F. 2007: Structure and intensity change of

Typhoon Songda (0418) undergoing extratropical transition. Papers in Meteorology and

Geophysics, 58, 135-153.

Klein, P. M., P. A. Harr, and R. L. Elsberry, 2002: Extratropical transition of western North Pacific

tropical cyclones: Midlatitude and tropical cyclone contributions to reintensification. Mon.

Wea. Rev., 130, 2240-2259.

Knippertz, P. and J. E. Martin, 2005: Tropical plumes and extreme precipitation in subtropical and

tropical West Africa. Quart. J. Roy. Meteor. Soc., 131, 2337–2365.

Kuhl., D., I. Szunyogh, E. J. Kostelich, and co-authors, 2007: Assessing predictability with a local

ensemble Kalman filter. J. Atmos. Sci., 64, 1116-1140.

Lapeyre, G. and I. M. Held, 2004: The role of moisture in the dynamics and energetics of turbulent

baroclinic eddies. J. Atmos. Sci., 61, 1693-1710.

Langland, R. H., 1995: Issues in targeted observing. Q. J. Roy. Met. Soc., 131, 3409-3425.

Lau, W. K. M., and D. E. Waliser, eds., 2005: Intraseasonal variability in the atmosphere-ocean

climate system. Springer, Berlin, 436 pages.

Lin, J.-L., and Coauthors, 2006: Tropical intraseasonal variability in 14 IPCC AR4 climate models.

Part I: Convective signals. J. Climate, 19, 2665-2690.

Lo, F., and H. Hendon, 2000: Empirical extended-range prediction of the Madden-Julian oscillation.

Mon. Wea. Rev., 128, 2528-2543.

Lynch, P., 2006: The emergence of numerical weather prediction: Richardson’s Dream. Cambridge

University Press, 279 pages.

Maloney, E. D. and D. L. Hartmann, 2000: Modulation of the Eastern North Pacific hurricanes by the

Madden Julian oscillation. J. Climate, 13, 1451-1460.

Martius, O., E. Zenklusen, C. Schwierz, and H. C. Davies, 2006: Episodes of Alpine heavy

precipitation with an overlying elongated stratospheric intrusion: A climatology. Int. J. of

Climatology, 26, 1149-1164.

Massacand, A. C., H. Wernli and H. C. Davies, 2001: Influence of upstream diabatic heating upon an

Alpine event of heavy precipitation. Mon. Wea. Rev., 129, 2822-2828.

Moncrieff, M.W., M.A. Shapiro, J. M. Slingo and F. Molteni, 2007: Collaborative research at the

intersection of weather and climate, WMO Bulletin, 56, 204-211.

Moncrieff, M. W., and C. Liu, 2006: Representing convective organization in prediction models by a

hybrid strategy. J. Atmos. Sci., 63, 3404-3420.

Mo, K. C., 2000: The association between intraseasonal oscillation and tropical storms in the Atlantic

basin. Mon. Wea. Rev., 128, 4097-4107.

Moore, R. W. and M. T. Montgomery, 2004: Reexamining the dynamics of short-scale, diabatic

Rossby waves and their role in midlatitude moist cyclogenesis. J. Atmos. Sci., 61, 754-768.

Muira et al , 2007: A Madden-Julian Oscillation event realistically simulated by a global cloud-

resolving model, Science, 318, 1763-1769.

Nakamura, H., and T. Fukamachi, 2004: Evolution and dynamics of summertime blocking over the Far

East and the associated surface Okhotsk high. Quart. J. Roy. Meteor. Soc., 130, 1213-1233.

Palmer, T. N., and R. Hagedorn, eds., 2006: Predictability of weather and climate. Cambridge

University Press, 718 pages.

Richardson, D. S., 2001: Measures of skill and value of ensemble prediction systems, their

interrelationship and the effect of ensemble size. Q. J. Roy. Met. Soc., 127, 2473-2489.

Ritchie, E. A. and R. L. Elsberry, 2007: Simulations of the extratropical transition of tropical cyclones:

Phasing between the upper-level trough and tropical cyclones. Mon. Wea. Rev., 135, 862-876.

Schwierz, C., S. Dirren, and H. C. Davies, 2004: Forced waves on a zonally aligned jet stream. J.

Atmos. Sci., 61, 73-87.

Simmons, A. J. and A. Hollingsworth, 2002: Some aspects of the improvement in skill of numerical

weather prediction. Q. J. Roy. Meteor. Soc., 128, 647-677.

Simmons, A., and B. J. Hoskins, 1979: The downstream and upstream development of unstable

baroclinic waves. J. Atmos. Sci., 36, 1239-1254,

Shapiro, M., and A. Thorpe, 2004: THORPEX International Science Plan. WMO, Geneva, 51 pp.

Slingo, J.M., and Coauthors, 1996: Intraseasonal oscillations in 15 atmospheric general circulation

models: Results from an AMIP diagnostic subproject. Clim. Dyn., 12, 325-357.

Sobel, A. H., and E. D. Maloney, 2000: Effect of ENSO and ISO on tropical depressions. Geophys.

Res. Lett., 27, 1739-1742.

Szunyogh, I., Z. Toth, A. V. Zimin, S. J. Majumdar, and A. Persson, 2002: Propagation of the effect of

targeted observations: The 2000 Winter Storm Reconnaissance program. Mon. Wea. Rev., 130,

1144-1165.

Talagrand, O., R. Vautard, and B. Strauss, 1997: Evaluation of probabilistic prediction systems.

Proceedings, ECMWF Workshop on Predictability, ECMWF, 1-25.

Toth, Z., O. Talagrand, and Y. Zhu, 2006: The attributes of forecast systems: a general framework for

the evaluation and calibration of weather forecasts. In Palmer and Hagedorn (2006), 584-595.

Uppala, S. M., P. W. Kallberg, A. J. Simmons and co-authors, 2005: The ERA-40 reanalysis. Q. J. Roy.

Meteor. Soc., 131, 2961-3012.

Vitart, F., 2003: Monthly forecasting system. ECMWF Technical Memorandum 424, 70 pp, available at

http://www.ecmwf.int/publications/library/do/references/list/14

Vitart, F., S. Woolnough, M.A. Balmaseda and A. Tompkins, 2008: Monthly forecast of the Madden-

Julian Oscillation using a coupled GCM. Mon. Wea. Rev., in press.

Waliser, D.E., K.M. Lau, and J.H. Kim, 1999a: A statistical extended-range tropical forecast based on

the slow evolution of the Madden-Julian Oscillation. J. Climate, 12, 1918-1939.

Walser, A., D. Lüthi and C. Schär, 2004: Predictability of precipitation in a cloud-resolving model.

Mon. Wea. Rev., 132, 560-577.

Weber, H. C., 2005: Probabilistic prediction of tropical cyclones. Part II: Intensity. Mon. Wea. Rev.,

133, 1853-1864.

Wei, M. and Toth, Z. 2003: A new measure of ensemble performance: Perturbation versus error

correlation analysis (PECA). Mon. Wea. Rev., 131, 1549-1565.

Wernli, H., S. Dirren, M. A. Liniger and M. Zillig, 2002: Dynamical aspects of the life cycle of the

winter storm “Lothar” (24-26 December 1999). Quart. J. Roy. Meteorol. Soc., 128, 405-429.

Wheeler, M. and K. Weickmann, 2001: Real-time monitoring and prediction of modes of coherent

synoptic to intraseasonal tropical variability. Mon. Wea. Rev., 129, 2677-2694.

Zhang, F., C. Snyder and R. Rotunno, 2002: Mesoscale predictability of the “surprise” snowstorm of

24-25 January 2000. Mon. Wea. Rev., 130, 1617-1632.

Figure captions

Fig. 1. Schematic illustration of the propagation of hydro-dynamical influences along a PV waveguide.

Also shown are the atmospheric processes that can cause the initial upstream disturbances and the

weather events these disturbances can trigger downstream.