R.S Ajaya Mohan and B.N Goswami
Centre for Atmospheric and Oceanic Sciences, Indian Institute of Science, Bangalore, India

How and to what extent the intra-seasonal oscillations (ISO's) influence the seasonal mean and it's inter-annual variability of the Indian summer monsoon is investigated using 42-years (1956-1997) daily circulation data from NCEP/NCAR reanalysis and satellite derived outgoing long wave radiation data for the period 1974-1997. Based on zonal winds at 850 hPa over the Bay of Bengal, a criterion is devised to define 'active' and 'break' monsoon conditions. The underlying spatial structure of a typical ISO cycle in circulation and convection that is invariant over the years is constructed using a composite technique. Typical ISO's have large scale horizontal structure similar to the seasonal mean and intensifies (weakens) the mean flow during it's 'active' ('break') phase. A typical 'active' ('break') phase is also associated with enhanced (decreased) cyclonic low-level vorticity and convection and anomalous upward (downward) motion in the northern position of the tropical convergence zone (TCZ ) and decreased (increased) convection and anomalous downward (upward) motion in the southern position of the TCZ. The cycle evolves with a northward propagation of the TCZ and convection from the southern to the northern position of the TCZ.

It is shown that the intra-seasonal and inter-annual variations are governed by a common mode of spatial variability. The spatial pattern of standard deviation of intra-seasonal and inter-annual variability of low-level vorticity is shown to be similar. The spatial pattern of the dominant mode of ISO variability of the low-level winds is also shown to be similar to that of the inter-annual variability of the seasonal mean winds. The similarity between the spatial patterns of the two variability indicates that higher frequency of occurrence of 'active' ('break') conditions would result in 'stronger' ('weaker') than normal seasonal mean. This possibility is tested by calculating two dimensional probability density function (PDF) of the ISO activity in the low-level vorticity. The PDF estimates for 'strong' monsoon years and 'weak' monsoon years are shown to be asymmetric in both the cases. It is seen that the 'strong' ('weak') monsoon years are associated with higher probability of occurrence of 'active' ('break') conditions. This result is further supported by calculation of PDF of ISO activity from combined vorticity and outgoing long wave radiation. This clear signal, indicates that the frequency of intra-seasonal pattern determine the seasonal mean. As the ISO's are essentially chaotic, it raises an important question on predictability of the Indian summer monsoon.

Bekryaev R.V.
Arctic and Antarctic Research Institute, St.-Petersburg, Russia

Variations of the global air surface temperature are closely connected to the changes of the planetary scale atmospheric circulation. Wallace et co-authors (1995,1996) have found that approximately half of the extratropical mean air surface temperature variation could be connected to the COWL (Cold Ocean Warm Land) patterns. They proposed also that the climate response to anthropogenic forcing should be distinct from the patterns of natural climate variability. Controversially, Corti, Molteni and Palmer (1999) have shown on the basis of studies of nonlinear chaotic models with preferred states or "regimes" that spatial patterns of the response to anthropogenic forcing may in fact project principally onto modes of natural climate variability. Using atmospheric circulation data from the Northern Hemisphere they have shown that recent climate change can be interpreted in terms of changes in the frequency of occurrence of natural atmospheric circulation regimes.

We have investigated the observed air surface temperature in the North Polar Area from 1891 to 1999 and the outputs of the 200 years control run and GHG+A (enhancing green house gases and aerosol content) numerical integration of the Canadian Climate Center Coupled General Circulation Model. Using Wallace at al's (1995) method of analysis and the results of the long-term CCC CGCM integration we have revealed the structures of the air temperature field at the surface level. We have repeated the so-called COWL structures in the temperature field that are well correlated with the global mean air surface temperature (r=0.7). To investigate how long-term variations of atmospheric circulation can be projected onto the modes of natural variability we have repeated our calculations using data of the CO2 + aerosol experiment. We have projected data of the CO2 + aerosol experiment onto the normalized covariation matrix calculated for the control run integration. It was found that most part (about 90%) of the global air surface temperature variability can be associated with the modes of natural variability. It is important that long-term variations connected with anthropogenic forcing projected very well onto these structures. On the other hand we cannon say that anthropogenic forcing is associated with changes in the frequency of occurrence of these regimes - projection of the temperature field of CO2 + aerosol experiment onto the first EOF calculated for control run has experienced very strong enhancing trend.

Spatial patterns of the response to anthropogenic forcing may in fact project principally onto modes of natural climate variability. This response cannot be connected only with changes in the frequency of occurrence of the natural atmospheric circulation regimes. Long-term anthropogenic forcing lead to changes of the phase space of the system of global scale atmospheric circulation.

J.D. Berner (1, 2) and G. Branstator (2)
(1) University of Bonn, Germany
(2) National Center for Atmospheric Re-searc, Boulder, USA.
email: berner@ucar.edu

Although there is agreement that the atmosphere is fundamentally nonlinear, it has been questioned if the amount of data available from observations and GCM integrations is large enough to yield a reliable and robust estimate of regime like behavior of the atmosphere. With this issue in mind, an exceptionally long atmospheric general circulation model integration of NCAR's CCM0 is examined to vigorously test for regime-like behavior and its robustness with regard to sample size.
Two approaches reveal nonlinear, regime-like behavior of the atmosphere. The probability density function (PDF) of projections onto planes defined by EOFs of the GCM's 500-hPa height field indicate significant departures from Gaussianity, though no more than one maximum is found. Regime-like behavior is much more evident in the mean phase space of tendencies where in some planes signatures of more than one equilibrium point are distinct.
Fokker-Planck and Master Equation are used to relate the PDFs to the phase space tendencies of the GCM. They form the framework to investigate to what degree the system behaves like a low-order nonlinear stochastic process. Finally the robustness of using a shorter dataset (like the one available from nature) is tested by applying Monte-Carlo significance testing techniques to 5000-day-segments of the integration, and comparing the results to the total integration of 1 million days.

Roxana Bojariu
National Institute of Meteorology and Hydrology, Sos. Bucuresti-Ploiesti 97, Bucharest, Romania
e-mail: bojariu@meteo.inmh.ro

Many studies demonstrate the existence of extended-range predictability for the particular regions and time scales due to the effects of certain large-scale phenomena. The deterministic forecasts using climate models have been lately improved but they cannot provide accurate details. On the other hand, the use of statistical approach to make regional extended range forecasts is not effective without an understanding of physical mechanisms, which act to configure the statistical relationships. That is why regional extended-range forecast needs a phenomenon-orientated approach. North Atlantic Oscillation (NAO) is the dominant mode of variability in the Northern Hemisphere. The main goal of this paper is to investigate the regional extended range predictability related to NAO.

The data used are monthly means of sea level pressure over the Northern Hemisphere and monthly means of temperature over Europe (from 11°W to 100°E and from 30°N to 90°N) extracted from the NCEP reanalysis. Large scale analyses are performed using canonical correlation analysis (CCA). A source of NAO related predictability has been identified over Central and Eastern European regions. November thermal anomalies over these regions are negatively correlated with the NAO index and could be used to predict the onset of the NAO phase in the following winter. Also, statistically significant correlation coefficients with negative sign have been found between January and next November temperature means suggesting that the mechanism which takes place over the Central and Eastern Europe is somehow related to the NAO interannual persistence. The results shown by NCEP re-analysis are verified using data from 62 Romanian climatological stations in the interval 1961-1990. The January/November signal is also found in the latter data set. This signal seems to be related to the atmospheric processes in the Arctic region. Low frequency phenomena in the Northern Hemisphere such as the 70 day oscillation might interact with the seasonal cycle to develop the November/January signal. The fact that the NAO related pattern is revealed by November thermal anomalies over Central and Eastern Europe does not necessary prove that the temperature anomalies are the cause of the NAO development but rather the effect of an Arctic process enhanced by the seasonal and orographic conditions taking place in the Central and Eastern Europe area.

Linear models have been build using four CCA modes of SLP and temperature anomalies in the interval 1961-1989 to predict November thermal and January SLP anomalies in the interval 1990-2000. The temperature based linear model succeeds in simulate the negative NAO phase occurring in January 1997 but fails to predict the previous return to negative phase of the NAO which occurs in January 1996 after a decadal interval with the atmosphere remaining locked into the positive NAO mode. The other NAO positive phases of the interval 1990-2000 are relatively well predicted although the magnitude of the anomalies are systematically underestimated by the linear model. November thermal anomalies for Central and Eastern Europe are reasonably well predicted from the Northern Hemisphere SLP anomalies occurring in the previous January. Further experiments with global and regional climate models need to be done to clarify this issue. Also, better predictions of NAO phase onset are possible with better tuned statistical models.

S. Corti
CINECA, Inter-University Computing Centre, Bologna, Italy

Diagnostics of the large-scale atmospheric circulation suggest that recent climate change may be interpreted in terms of variations in the frequency of occurrence of preferred modes of natural atmospheric variability, rather than a as simple linear shift in the mean climate superimposed to red-noise fluctuations. This contrasts with the characteristics of the observed changes in temperature-related diagnostics, which appear much closer to the simple linear picture. While not necessarily inconsistent, the implications of these two views on the nature of climate change need to be explored.

In this framework, the main purpose of this study is to explore the vertical structure of circulation regimes using a 50-yr NCEP reanalysis dataset, and (at the same time) illustrate the potential role of weather regimes and non-linearity in the emerging anthropogenic signal. The variables considered in this study are Northern Hemisphere extended-winter (November to April) monthly means of: i) geopotential height at 500 hPa; ii) temperature at 700 hPa; iii) surface temperature; iv) mean sea level pressure. As in the study of Corti et al. (1999) the seasonal cycle has been removed and the data are detrended by taking deviations from a 5-year running mean. An empirical orthogonal function (EOF) analysis is first applied in order to define, for each variable, a reduced phase space based on the leading modes of variability. Then a simultaneous analysis of the four fields is carried out performing a further EOF analysis in the subspace spanned by the 10 leading EOFs of each field. This produced a 4-variable EOF picture of the large-scale vertical (and thermal) structure of the atmosphere.

The PDF in the reduced phase space spanned by the first two 4-variable principal components was computed, and a multimodal distribution with three maxima was found. The 500-hPa height geographical patterns of these density maxima are strongly reminiscent of clusters A, B, and D found by Corti et al. (1999). The fact that the same 500hPa clusters have been found from a multi-variable/multi-level analysis may represent an indication of the existence of a full three-dimensional regime structure. With the three-dimensional structure available, it was possible to assess whether the observed temperature changes are consistent with an increase in certain regime frequencies. Moreover, the hypothesis that regime frequencies change more than regime structures was checked by performing a regime analysis separately on the first and last 25-yr periods.

E. P. Gordov, O. B. Rodimova and A. Z. Fazliev
Institute of Atmospheric Optics SB RAS, Tomsk 634055 Russia

Number of observational and modeling evidences is available now on the existence of non-linear, regime-like behavior in the atmospheric and oceanic circulation and main efforts of the community are addressed to their understanding from a theoretical standpoint. However, there are other physical subsystems participating in climate formation whose nonlinear behavior should be considered while studying the atmospheric response to variations in natural and anthropogenic forcing. In virtue of complexity of atmospheric processes such analysis is to be performed on the base of relevant specially constructed low-order models. Reported are some results of this approach usage to atmospheric ozone photochemistry and altitude behavior of temperature in dependence on absorbing species concentrations. In the both cases relevant low-order models are formed by sets of ordinary nonlinear equations, which allows one to analyze their evolution by means of nonlinear dynamics. Also briefly described is an attempt to fill a gap between low-order model predictions and atmospheric situations on the base of specially designed information-computing systems.

Quest for the atmospheric photochemistry low-order model is performed on the base of qualitative analysis of the ozone cycle kinetic equations. It is shown that the constructed two-variable model mimics basic features of the Chapman oxygen atmosphere, including nontrivial case of two steady states appearance. The model added with one more variable describes tendencies of the oxygen components behavior in the mesosphere oxygen-hydrogen atmosphere. We hope eventually arrive to a low-order model reflecting main nonlinear features of atmospheric photochemistry aimed to determination of states where small variation of some atmospheric species concentrations can lead to significant differences in climate system evolution.

The low-order model for the altitude temperature behavior is constructed on the base of conventional equations for radiation fluxes. Qualitative analysis of resulted set of ordinary differential equations for temperatures of layers determines steady states and their properties at variations of gas species concentrations. It is shown that for a wide range of the variations the only steady state is the stable node. Variations of gas concentrations change the relaxation times and these changes grow with altitude. It should be noted that appearance of temperature dependence in the albedo could lead to instability of altitude temperature profile in definite atmospheric layers.

Low-order models can only indicate domains of possible non-linearity manifestations. To determine whether it will be the case under more realistic treatment of atmospheric processes sophisticated models should be employed at these domains of possible nontrivial behavior. We plan to support such investigations within the framework of currently developed information-computational systems "Atmospheric Chemistry" and " Atmospheric Optics".

Tomas Halenka
Dept. of Meteorology and Environment Protection, Fac. of Mathem. and Physics, Charles University, V Holesovickach 2, 180 00 Prague, Czech Republic

To give an objective characteristics of circulation patterns with respect to the Earth symetry the spectral structure of stratospheric fields is analyzed in terms of spherical harmonics to compare the long-term behaviour and connections to some extra-terrestrial influence and circulation patterns. Two sources of data became available for this study, reanalyses of geopotential and temperature from Free University in Berlin available in 50, 30 and 10 hPa levels covering the period of 1976-96 and huge database of reanalyses from NCEP, where appropriate spectral coefficients are available for period 1948-now four times per day in 28 levels for vorticity, divergence, temperature, etc. To describe the circullation patterns a study for vorticity on selected levels both in troposphere and stratosphere is presented. Temporal analysis of significant spherical harmonics is introduced as well as the comparison of their changes with respect to the changes of different sets of solar, geomagnetic and global circulation indices. Quite strong connections to a set of "extraterrestrial" parameters appear for four trough shape of polar vortex. The natural variability connected to the extraterrestrial influence is studied as well as interannual variability with the emphasis to the QBO and ENSO. The systematic review of the appropriate correlations is presented and regime-like behaviour in time series is discussed. The interactions between stratospheric and tropospheric circulation are also discussed in terms of coefficients of spherical harmonics.

Dörthe Handorf, Antje Weisheimer, Klaus Dethloff
Foundation Alfred Wegener Institute, Research Department Potsdam Telegrafenberg A43, D-14473 Potsdam, Germany
e-mail: dhandorf@awi-potsdam.de

In order to contribute to the understanding of natural climate variations on decadal and longer time scales the low-frequency variability of the atmosphere and of the coupled atmosphere-ocean system under constant external forcing conditions is examined in this study. For this purpose the results of long-term integrations of various climate models of different complexity over 1,000 to 10,000 years have been analysed. The model hierarchy consists of a quasi-geostrophic atmospheric spectral low order model, a coupled atmosphere-ocean climate model of moderate complexity as well as complex coupled atmosphere-ocean circulation models. By means of analyzing the spatio-temporal variability with multivariate methods and time-frequency analysis intermittent, temporary significant signals on annual, decadal up to centennial time scales have been detected for all models at tropospheric as well as stratospheric heights. These signals are governed by the internal nonlinear dynamics of the atmosphere or the atmosphere-ocean system. The dominant spatial patterns are characterized by large-scale circulation structures, e.g. the Arctic oscillation.

The investigations with the low-order model identified orographically and thermally induced instabilities, internal coupling between the waves on different spatial scales and nonlinear chaotic transitions between circulation anomalies as sources for decadal atmospheric variability. Accordingly, it is possible to describe the variability of the more realistic coupled models by changes in the frequency of occurrence of the preferred circulation regimes. In order to identify such circulation regimes, the probability distributions of atmospheric variables in a reduced phase space have been calculated for the simulations with the coupled models. The multi-modality of the different models has been discussed and compared with known analyses of observed data presented in the literature.

Gordana Jovanovic, Irini Reljin, Branislav Reljin
Federal Hydrometeorological Institute, Belgrade, Yugoslavia

The relationship between El Nino and temperatures in Serbia has been proved in this paper. The analysis of frequency spectra derived for temperature series for Nino 3,4 and for the temperatures in Serbia has been performed. A significant correlation among the average monthly temperatures in Serbia and in Nino 3,4 region, with three months delay, has been found. The fractal analysis of these temperature series was performed via R/S statistic indicating to the fractal and multifractal (MF) nature of processes. The multifractal analysis of temperature series was performed via f (?) diagrams for a number of station starting from Nino 3, 4 region going further to the east. The results suggested that the regions being at a greater distance from the huge mass of the Pacific Ocean exhibit much variability in temperature changes so their multifractal spectra become broader.

J. R. Kulkarni, M. Mujumdar, and V. Satyan
Indian Institute of Tropical Meteorology, Dr. Homi Bhabha Road, Pashan, Pune 411008, India

Indian Summer Monsoon (ISM) is a non-linear dynamical system. The simulations of ISM rainfal by General Circulation Models are sensitive to initial conditions. The seasonal ISM rainfall depends on internal variability due to sensitivity to initial conditions and external variability due to boundary forcings. The contributions from internal and external variabilities to the total interannual variability of ISM rainfall have been quantified. Center for Ocean Land Atmosphere General Circulation Model at T30L18 resolution has been integrated for six summer monsoon seasons: (June through September), with observed sea surface boundary conditions. The ensemble integrations with 10 different atmospheric initial conditions from 1 May-10 May were carried out. The study showed that the contributions from external variability are twice the contributions from internal variability on the seasonal scale. On the monthly scale the contributions from external variance dominate for the first three month period. Based on cumulative rainfall variance analysis the contributions of external and internal variabilities to the total variability are found to be ~60% and 40% respectively. The sensitivity of intraseasonal modes to initial conditions was studied by decomposing the time series into constituent modes using wavelet transform. The Haar orthogonal wavelet function is used to decompose the total variance into variances of 7 dyadic scales. It was observed that the intraseasonal modes are very much sensitive to initial conditions.

Sujata K. Mandke
Indian Institute of Tropical Meteorology, Dr. Homi Bhabha Road, Pashan, Pune 411008, India

Indian Summer Monsoon(ISM) is one of the most important features of tropical circulation during northern summer and is one of the most difficult to simulate correctly by General Circulation Models(GCM). Influence of Sea Surface Temperature anomalies specifically in tropical Pacific on ISM is well regonised. The interannual Variability of ISM simulated by Atmospheric GCMs forced with observed SST is met with some success.

The objective of the present study is to find the relative contribution of slowly varying boundary forcing of SST and internal dynamics to interannual variability of ISM simulated by Hadley Centre climate model(HadAM2b). For this purpose, two sets of 17-years(1979-1995) long integrations of HadAM2b model have been carried out. In the first set, the model is forced with monthly observed SST and sea-ice and in the second set with climatological SST and sea-ice.

To study the relative contribution of SST and internal dynamics to interannual variability of ISM, it is worthwhile to study first, how good is the model in simulating ISM and its interannual variability when forced with obsreved SST and sea-ice. For this purpose, the climatology of ISM and its variability based on 17-years simulation by the HadAM2b model forced with observed monthly SST and sea-ice, is compared with observations. Though most of the models simulate gross features of monsoon and its variability well, there are differnces when compared with observations. Wind at 850 hPa for June to September(JJAS) simulated by the model shows that low level westerlies are stronger and extends too far to the east compared to observations(NCEP reanalysis). The model simulates above average rainfall over most of the Asian summer Monsoon region during El Nino years. The spatial distribution of Correlation Coefficient (CC) between model simulated rainfall and observed rainfall(Xie-Arkin data) based on 17-years, shows that over Equatorial oceanic regions, the interannual variability is well simulated. However, over the Asian monsoon region, CC's are very small. The CC between simulated and observed zonal as well as meridional wind also shows some differences, especially over Equatorial Indian and Pacific oceans.

Comparison of the interannual variability simulated by the model in observed SST integration and climatological SST, shows that the interannual variability simulated in the climatological SST integration is much less(Coefficient of variation=4.2) than in the obsreved SST run(Coefficient of variation=7.32). Comparison of the variance of the precipitation as well as the circulation indices for the two sets of integration concludes that about 40% of the interannual variability of ISM simulated by the HadAM2b model may be accounted for by the internal dynamics and 60% by the boundary forcing of SST and sea-ice.

Ute Merkel and Mojib Latif
MPI for Meteorology, Hamburg, Germany
email: merkel@dkrz.de

To address the question of whether an extratropical atmospheric response to ENSO extremes in the North Atlantic/European sector can be detected, we performed several ensemble integrations (120 days) with the ECHAM4/T106 model forced by prescribed SST anomalies. The respective forcing patterns were derived from a regression analysis in which global observed winter (spring) SST anomalies were regressed onto the observed winter Niño3 SST anomaly index. The results of the experiments are compared to a corresponding control integration with climatological SST. The seasonal mean response during both the winter (DJF) and spring (MAM) season is analysed in different atmospheric fields (sea level pressure, temperature, precipitation) to investigate possible simultaneous (DJF) or lagged (MAM) relationships. Since the present study focusses on the North Atlantic/European sector a dynamical understanding of the modelled response requires the consideration of transient eddy fluxes and stationary waves in determining the seasonal mean response. Additionally, a cyclone tracking analysis was performed to better understand the simulated precipitation anomalies over Europe Finally the impact of ENSO extremes on the frequency and intensity of hurricanes over the tropical Atlantic is investigated. A 20-year AMIP-like run with the ECHAM4/T106 model has been conducted to study how the hurricane statistics are affected by ENSO extremes.

Adam Hugh Monahan (1), Lionel Pandolfo (2), and John Fyfe (3)
(1) Institut fuer Mathematik, Humboldt-Universitaet zu Berlin, Germany
(2) Department of Earth and Ocean Sciences, University of British Columbia, Vancouver, BC, Canada
(3) Canadian Centre for Climate Modelling and Analysis Meteorological Service of Canada, Victoria, BC , Canada

A nonlinear generalisation of Principal Component Analysis (PCA) is applied to the 500mb geopotential height field of the Northern Hemisphere extratropical atmosphere. It is found that the low-frequency variability of the mid-troposphere is characterised by three distinct quasi-stationary states. The states are described and compared to those obtained from applications of cluster analyses and linear PCA to the height field. Evidence is provided that modes obtained through PCA (notably the Arctic Oscillation (AO)) are not independent dynamical modes of variability of the Northern Hemisphere extratropics. Rather they arise as the optimal linear compromise between the preferred quasi-stationary states of the circulation.

V. Pavan (1) and F. Molteni (2)
(1) CINECA, Inter-University Computing Centre, Bologna, Italy
(2) International Centre for Theoretical Physics, trieste, Italy

A simplified climate model is used to represent the interannual variability of the Atlantic storm-track. The model consists of a five-level primitive equation dynamical core, joined to a set of simplified physical parametrizations. A description is given of the climatology and of the the main modes of interannual large-scale variability of the atmosphere over the Euro-Atlantic region as represented by the simplified model, paying particular attention to the relation between transients and quasi-stationary features.

The study includes an analysis of the relevance of surface conditions to the determination of mid-latitude interannual variability over this region. This consists of a comparison of multi-decadal integrations of the model using different sea-surface boundary conditions and orographic profiles. First, the influence of imposed sea-surface anomalies over selected, even remote regions, on the intensity of the storm-track is investigated. Secondly, a study is presented on the effect of the Rockies and Greenland orography on the Euro-Atlantic interannual variability.

Vinicio Pelino (1) and Antonello Pasini (2)
(1) Servizio Meteorologico dell'Aeronautica, CNMCA, Roma, Italy
(2) CNR - Istituto sull'Inquinamento Atmosferico, Roma, Italy

Three models, worked out by Lorenz in 1960, 1963 and 1984, respectively, are re-written in a unifying algebraic formalism and analysed as far as their dynamical behaviours are concerned. We obtain clear expressions of dissipation and forcing vectors, usually hidden in the standard formalism, behind a Hamiltonian structure. We give an overview of different dissipation mechanisms and identify the term that is responsible for the birth of chaotic behaviours. Because the underlying Lie-algebra structure of the fluid dynamic equations found in these models clarifies the source of their chaotic behaviour, as a perspective we can search for the same structure in a higher spectral model coming from a bigger group. The possibility of obtaining some useful constraints on the order of truncation of Navier-Stokes equations, based on group theory, is also envisaged.

Weihong Qian
Department of Geophysics, Peking University, Beijing 100081, China

An analysis of the El Niño/Southern Oscillation (ENSO) cycle associated with the vertically-integrated equatorial tropospheric (1000-100hPa) zonal wind anomalies (ZWA) over the Pacific Basin is made in this paper. The data sets used are the monthly US National Center of Environmental Prediction (NCEP) reanalysis wind and the NCEP monthly sea surface temperature (SST) from 1950 to 1998. The wavelet analysis of both ZWA at different regions along the equator and SST anomaly in the equatorial eastern Pacific (EEP) shows that there are various phase-lag relationships between them in the interannual and interdecadal time-scales. In the equatorial western Pacific, the phase of ZWA in most events is earlier than that of EEP SST anomaly (or El Niño event) for about 7-13 months at the interannual timescale. In the equatorial central Pacific, there is little phase-relationship between them. In the equatorial eastern Pacific, the phase of EEP SST anomaly mostly is earlier than that of ZWA for about 4-9 months while the positive ZWA leads positive SSTA about 18 months ahead. The phase of ZWA in the equatorial South America coast (ESAC) leads that of the EEP SST anomaly about 6 months in advance. In the equatorial western Pacific and ESAC, the El Niño event appears basically when the ZWA transforms from westerlies to easterlies while the La Niña event takes place in the reversed situation. The relationships are explained using a simple tropical air-sea coupled model (Gill model), which includes a positive/negative feedback scenario that seems to hold during the ENSO cycle. According to these findings, a strong El Niño is predicted for 2001-2002.

P.M.Ruti (1), A.Sutera (2)
1. Gruppo di Dinamica dell'Atmosfera e dell'Oceano, ENEA, Roma, Italy
2. MEDEA, Un. di Roma, Physics Dep., Italy

Recent works find out that the leading modes of variability of the Northern Hemisphere extratropical circulation are characterized by zonally symmetric or annular structures. This annular mode for the wintertime sea-level pressure field resembles the NAO pattern, but the center of action covers more of the Arctic, giving it a more zonally symmetric signature. This mode has been referred to as the Arctic Oscillation (AO). The AO has been interpreted as the surface signature of modulations in the strength of the polar vortex, hypothesising a stratosphere-troposphere interaction mechanism to explain the NAO. The NAO and the AO have been considered as different interpretations of the same phenomenon. In this work we analysed the interannual variability of several wintertime fields, using 52 years of the NCEP reanalysis data set for the Northern Hemisphere. We applied a principal component analysis (PCA) to sea level pressure, 500 and 50 hPa geopotential heights and tropopause height. These fields allow us also to inspect the vertical signature of the Northern Hemisphere variability. Then following Rossby earlier works we also computed the PCA for the eddy fields. Our analysis confirms the existence of the annular mode, that in the sea level pressure fields represents the AO, and the strong signature of this mode up to the stratosphere. The orthogonality of the empirical modes suggests that the AO and the NAO don't represent the same phenomenon.

Frank Selten
K.N.M.I., AE de Bilt, the Netherlands

A long integration of a million days of the three level quasi-geostrophic model of Molteni is used to assess the structure of the attractor in the space of the dominant circulation structures at the 500 hPa level. The sensitivity of this structure to perturbations to the forcing is assessed by an ensemble of integrations with different types of perturbations added to the time- independent potential vorticity forcing. A method is outlined to find those perturbations to forcing and model parameters that have the greatest impact on the structure of the attractor.

Nityanand Singh
Indian Institute of Tropical Meteorology, Dr. Homi Bhabha Road, Pashan, Pune-8

Most of the rainfall over India occurs during June-December period in association with two planetary scale atmospheric circulations viz. southwest monsoon circulation during June through September and Northeast monsoon circulation during October through December. Traditionally multiple regression models with regional and global parameters, representing regional forcings and global teleconnections respectively, as predictors have been developed for large-scale performance of the monsoons. Reliability of regression models is poor. Further prediction of large-scale performance of the monsoons rainfall is of limited practical value. Predictive model for reliable rainfall for smaller regions of the country does not exist.

Investigation reveals that most of the regional and global circulation parameters that show significant correlation with the southwest monsoon rainfall also show the correlation of the same sign with the northeast monsoon rainfall but magnitude was smaller. Interesting to notice is that large-scale performance of the two monsoons over the country is significantly correlated. From the above analysis we intend to believe that 'rainfall of a particular season depends not only its own preceding years' values but on rainfall of other seasons as well over a long period of time'. In the present study a quantitative technique/method for prediction of seasonal/sub-seasonal/monthly rainfall over smaller regions has been developed by modelling and extrapolating natural regularities (or non-Gaussian features) such as seasonality, long-term/short-term trends, aperiodic oscillation and epochal pattern. There are three main steps of the method: i] harmonic analysis; ii] determination of scaling factor between the one-step-ahead generated value from the combination of selected harmonics and its corresponding actual value; and iii] the final prediction after suitable treatment by the scaling factor of the one-step-ahead generated value in the future period. The method has been applied for the sequential prediction of winter (JF), summer (MAM), June, July+August, September and northeast monsoon (OND) rainfall over the six homogeneous zones of the country. The rainfall series of the period 1848-1997 (150 years) with winter (JF), summer (MAM), June, July+August, September and northeast monsoon (OND) amounts kept in sequence is used in the study.

A Prediction Reliability Index is defined as, (1-rmse/s)*100, where rmse is the root mean square error of independent predictions and s is the error (or standard deviation) if the predictions were mean of interannual series each time, to assess the skill of the present method of prediction. The prediction is categorized as excellent, very good, good, fair, poor and unpredictable for the PRL value between 100% and 80%, 80% and 60%, 60% and 40%, 40% and 20%, 20% and 0% and less than 0% respectively. Based on 10-year independent samples, 1988-1997, prediction of June, July+August, September and post-monsoon rainfalls is found very good to fair over each of the six zones. Prediction of summer rainfall over Northeast India is judged very good and over South Peninsular India as good. In general, winter rainfall is unpredictable.

A modified version of the above technique has been applied for the prediction of number of monsoon storms/depressions over the Indian region, surface air temperature of the Northern hemisphere, annual mean sunspot number, monthly surface air temperature, monthly sea surface temperature, monthly sea level pressure and zonal wind at 50-mb, 30-mb and 10-mb levels over Balboa with excellent results.

Sonechkin D.M.
HydroMeteoCenter of Russia, Moscow, Russia

An important role of the Rossby-Haurwitz velocity in the short-term planetary wave dynamics is well known. Moreover, it has recently shown in our studies that the Rossby-Haurwitz velocity is also an important component of the long-term propagation of planetary waves in the extratropical westerlies. By the reasons of the spherical shape of the Earth, and a prevailence of the zonal propagation of air flows in extratropics the typical reciprocal value of the zonal index (about 20-30 reciprocal days) may be considered as a new small parameter (different from the Kibel-Rossby small parameter) of an expansion of solutions of the equations representing the low-frequency planetary wave dynamics. The terms of the differences between the Rossby-Haurwitz velocities of pairs of the planetary waves with the same zonal wave number turn out to be main contributors in such expansion. Besides, the values of such terms turn out to be commensurate ratios. The fact means that there are some resonances in the low-frequency planetary wave motions. As well-known, the resonant motions usually reveal a phase locking. Therefore, a synchronous propagation of planetary waves in westerlies may be assumed.

At the first time the phenomenon of the planetary wave synchronization has been recognized in runs of some atmospheric toy-models with strange attractors (Sonechkin D.M. "Stochasticity in general atmospheric circulation models", HydroMeteoIzdat, Leningrad, 1984 - in russian). The synchronization reveals itself as a release of enormous instabilities of all nonsynchronous motions that are mathematically possible but practically nonobservable in long runs of the models considered.

Then, the phenomenon was found in the dynamics of the real-world atmosphere. Some fingerprints of the synchronization turned out to be seen even in the day-to-day planetary wave motions as rather essential statistical correlations between the values of the compared daily wave phases and amplitudes. But, the phase synchronization turned out to look like an almost functional relationship (similarly to the quasi-hydrostaticity and quasi-geostrophicity phenomena in shorter time scales) when the 5-day mean geopotential height fields were considered. It needs to be also noted that the synchronized low-frequency motions are almost barotropic, and there exist some preferable geographical localizations of the wave phases of the shorter synoptic (the zonal wave numbers more than 7) waves but not only of the ultra-long waves. The fact of these geographical localizations indicates on an importance of the topographical torque in the low-frequency dynamics.

The synchronization phenomenon has been used to create a special hydrodynamical model called quasi-synchronous (Sonechkin D.M. et al., Monthly Weather Review,1995, vol. 123, n. 8, 2461-2473). A spectral form of the barotropic vorticity equation with an explicit presentation of the coefficients of the nonlinear interactions was used for this purpose. A simple topograpic torque of the planetary waves was also taken into consideration as well as a Newtonian equilibrating the wave amplitudes. This quasi-synchronous model depicts the only synchronized planetary wave motions excluding nonsynchronized ones. Because the latter motions seem to be one of the main sourses of the weekly predictability loss the absence of these motions in the modelled planetary wave dynamics facilitates overcoming this limit of all present-day forecasting models based on the primitive equations.

40-day long runs of this model already are used with a success in HydroMeteoCenter of Russia to create monthly weather forecasts. These runs also are disseminated to several European countries in real time for the same purpose.

David B. Stephenson and Abdel Hannachi
Climate Analysis Group, Department of Meteorology, University of Reading
e-mail: cag@met.reading.ac.uk

Climate is the complex result of non-linear interactions between many degrees of freedom. Various conceptual paradigms can be used to help understand this complexity and these will be discussed. The null hypothesis of multinormality will be presented and an attempt will be made to reject it for various climatic data sets. Some improved techniques for exploring the probability distribution in state space will be illustrated such as mixture modelling and robust normality estimators.

Axel Timmermann
K.N.M.I., AE de Bilt, the Netherlands

The physical mechanism which generates decadal modulations in the amplitude of the El Niño-Southern Oscillation phenomenon (ENSO) is investigated. Our analysis is based on a multicentury present day climate simulation performed with an ENSO-resolving Coupled General Circulation Model. In constistency with observations it is found that ENSO variance undergoes periodic oscillations with a timescale of about 10-20 years. This decadal beat is closely linked to the second dominant pattern of tropical (sub)surface temperature variability. This pattern has been identified in a previous study as the pattern which is associated with the strongest increase of ENSO variability in a greenhouse warming simulation. The dipole-like characteristic of this mode is generated mainly by the interplay of horizontal, vertical advection and mixing. We suggest a nonlinear mechanism which is capable of generating decadal climate anomalies in the tropics as well as decadal ENSO amplitude modulations. This new paradigma is validated using low-dimensional ENSO models. Furthermore, the association between linear (in)stability and changes of ENSO variance will be studied in a greenhouse warming simulation performed with the CGCM ECHAM4/OPYC3.

Alyssa Young (1) and Lionel Pandolfo (2)
(1) Department of Mathematics, University of British Columbia, Vancouver, BC, Canada V6T 1Z2
(2) Department of Earth and Ocean Sciences, University of British Columbia, Vancouver, BC, Canada V6T 1Z4

The dynamics of the extra-tropical atmosphere is comprised of natural climate variability and responses to external forcings. It has been proposed that the responses to forcings proceed through the same dynamical regimes as those describing natural variability. Whether this is always the case for any type of forcings is still an unanswered question. We examine this hypothesis in the case of dynamical forcing caused by the radiative effects of volcanic aerosols. This external agent may force select, natural modes of atmospheric variability or generate new temporary regimes. We use data from satellite observations, reanalyses and general circulation model experiments to investigate differences in the occurrences of regimes of variability between volcanic and non-volcanic years in the fields of temperature and geopotential heights. For the temperature field, the model experiments including volcanic aerosols indicate that the forcing introduces a new mode of variability that is not a leading mode in the control experiments. However, this is not the creation of a new mode so much as the promotion of a previously more insignificant mode. This suggests a regime-like behaviour, where the volcanic temperature anomaly pattern projects onto an existing mode of variability and promotes it temporarily to be a preferred state. The geopotential height field is affected in a different way. The volcanic height anomaly pattern projects onto multiple linear modes. We propose that a non-linear representation is more ideal, where these multiple linear modes are the components of a single non-linear mode. For both temperature and height fields there is also an apparent interaction between the seasonal cycle and the regimes forced by volcanic aerosols. While the non-linear interactions of these various modes are still not well understood, these results present evidence that dynamical forcing through volcanic aerosols has a non-linear effect on atmospheric conditions.

Zavala-Garay J. (1), C. L. Perez (1), A. M. Moore (1) and R. Kleeman (2)
(1) Program in Atmospheric and Oceanic Sciences and Cooperative Institute for Research in Environmental Sciences, University of Colorado at Boulder, USA.
(2) Courant Institute for Mathematical Sciences, New York University, New York, USA.

Atmospheric circulation is composed of motions that are both coupled and uncoupled with the ocean, and interact on multiple timescales. However, numerical models of varying complexity can realistically reproduce some of them. In this work we examine the role that the observed intraseasonal variability can play in controlling the observed ENSO variability. To this end, we force an intermediate ENSO model with observed stochastic forcing which is defined as the part of the atmospheric variability that is uncoupled from the ocean. The stochastic forcing is estimated from 18 years (1982-2000) of NCEP/NCAR reanalysis of surface winds and Reynolds Sea Surface Temperatures (SST) in the Tropical Pacific. The part of the atmospheric variability that is related to variations of SST is estimated using the Singular Value Decomposition of the covariance between wind and SST and is then substracted from the wind data to recover the stochastic component. Principal Component Analysis of the stochastic component shows no preferred mode of variability, exhibits decorrelation times of a few days, and an spectrum with significant power in the intraseasonal band and background indistinguishable from red noise. An 18-year stochastically-forced model integration shows striking similarities with the observed equatorial SST. The robustness of this result is checked by performing different sensitivity experiments. The model's free parameters are chosen so that the coupled system is asymptotically stable (with a decay time of about 3 years). Therefore, these results support the hypothesis that a significant fraction of ENSO variability may be due to stochastic forcing.