CLIVAR

CLIMATE VARIABILITY AND PREDICTABILITY

International CLIVAR Project Office
National Oceanography Centre
European Way
Southampton, SO14 3ZH, UK
Phone: +44-2380 596777
Fax: +44-2380 596204
Email: icpo@noc.soton.ac.uk

WCRP Workshop on Seasonal Prediction

Barcelona Spain June 4-7, 2007

Abstracts

Poster Session I & II

Day 2 - Tuesday 5th June

POSTER SESSION I - Seasonal Prediction Skill and Coupled Model Performance

Seasonal Prediction Skill in the New ENSO Forecast System at Japan Meteorological Agency
Yuhei Takaya
Global environment and Marine Department, Japan Meteorological Agency
Tamaki Yasuda(MRI/JMA), Satoshi Matsumoto(MRI/JMA), Tosiyuki Nakaegawa(MRI/JMA), and Tomoaki Ose(MRI/JMA)

The New ENSO Forecast System at Japan Meteorological Agency
Tamaki Yasuda
Meteorological Research Institute, Japan Meteorological Agency
Y.Takaya (JMA/CPD), T.Nakaegawa (JMA/MRI), Y.Fujii (JMA/MRI), S.Matsumoto (JMA/MRI), G.Yamanaka (JMA/MRI), M.Kamachi (JMA/MRI), and T.Ose (JMA/MRI)

The Impact of Air-Sea Interaction on Tropical Intraseasonal Variability in the CFS
Kathleen Pegion
George Mason University
B. Kirtman (COLA/GMU) J. Shukla (IGES/GMU)

Historical Forecast Project with four models at the Canadian Meteorological Centre
Juan Sebastian Fontecilla
Canadian Meteorological Centre (CMC)
Benoit Archambault, CMC Marie-France Turcotte, CMC Normand Gagnon, CMC Richard Verret, CMC

Coupled seasonal forecasting at CCCma - Initial results
W. Merryfield
Canadian Centre for Climate Modelling and Analysis
G. Boer, G. Flato, S. Kharin, B. Pal and J. Scinocca

Seasonal forecasts: experience at ECMWF with single and multi-model systems
Tim Stockdale
ECMWF
F. Molteni (ECMWF), M.A. Balmaseda (ECMWF), P. Doblas-Reyes (ECMWF) and L. Ferranti (ECMWF)

GloSea4: the new Met Office seasonal ensemble prediction system
Alberto Arribas
Met Office
S. Cusack (Met Office), A. Maidens (Met Office), M. Gordon (Met Office), A. Scaife (Met Office)

Seasonal forecasting on the basis of SL-AV model at the Hydrometcentre of Russia
Dmitry Kiktev
Hydrometcentre of Russia (HMC)
Mikhail Tolstykh (INM RAS and HMC, presenting author), Radomir Zaripov (HMC), Mikhail Zaichenko (HMC)

Evaluation of summer monsoon intraseasonal variability and predictability in the DEMETER hindcasts
Jean Philippe Duvel
Laboratoire de Météorologie Dynamique. ENS/IPSL/CNRS
P. K. Xavier (LMD) F. J. Doblas-Reyes (ECMWF)

Interannual variations of the boreal summer intraseasonal variability predicted by ten atmosphere-ocean coupled models
Hyemi Kim
Seoul National University
In-Sik Kang(Seoul National Univ), Bin Wang and June-Yi Lee (IPRC)

Characteristics of ENSO and Monsoon Predictability in GCM Forecast
Emilia Jin
George Mason University/COLA
J.Kinter(George Mason University/COLA) B.Wang(University of Hawaii/IPRC) J.Shukla(George Mason University/COLA) I.S.Kang (Seoul National University/CES)

ENSO Influence on the Rotational Flow in the EuroAtlantic Sector: Implications for Predictiability
Javier García Serrano
Departamento de Geofísica y Meteorología, Universidad Complutense Madrid
J. García-Serrano (1), B. Rodríguez-Fonseca (1), I. Bladé (2)
(1) Departamento de Geofísica y Meteorología, UCM, Madrid, Spain (2) Departamento de Astronomía y Meteorología, UB, Barcelona, Spain

Recent Trends in the Connection Between Atlantic and Pacific Ninos
Belèn Rodríguez de Fonseca
Dpto Geofísica y Meteorología.Facultad de Ciencias Físicas. Universidad Complutense de Madrid.
Irene Polo (1), Javier García Serrano(1), Carlos Roberto Mechoso (2)
(1)Dpto Geofísica y Meteorología.Facultad de Ciencias Físicas. Universidad Complutense de Madrid. (2) Dptmt of Atmospheric Sciences.University of California at Los Angeles. USA

The Influence of the Tropical Atmospheric Bridge on the ENSO Extratropical Response in Late Fall/Early Winter
Ileana Bladé
Departament d'Astronomia i Meteorologia. Facultat de Física. Universitat de Barcelona
Matthew Newman (1,2), Michael A. Alexander (2), James D. Scott (1,2)
(1) CIRES Climate Diagnostics Center, University of Colorado, (2) Physical Sciences Division/NOAA Earth System Research Laboratory, Boulder, Colorado.

Interdecadal variability of ENSO properties and prediction lead time
Wenju Cai
CSIRO Marine Atmospheric Research
Tim Cowan (CSIRO Marine and Atmospheric Research)

Realistic greenhouse gas forcing and seasonal forecasts
Mark Liniger
Federal Office of Meteorology and Climatology, MeteoSwiss
H. Mathis (1), C. Appenzeller (1), and F. J. Doblas-Reyes (2)
1. Federal Office of Meteorology and Climatology, MeteoSwiss 2. European Centre for Medium-Range Weather Forecasts (ECMWF)

The International CLIVAR Climate of the 20^th Century Project: Understanding and Attributing Climate Variations of the Past 130 Years
James L. Kinter III
Center for Ocean-Land-Atmosphere Studies, Calverton, MD USA
Chris Folland (Hadley Centre, Met Office, Exeter, Devon UK)

Report on WCRP-WGNE meeting on model systematic errors, held in San Francisco, February 2007
Michel Déqué
Météo-France/CNRM

Improved seasonal forecast skill: The role of model systematic error
Noel Keenlyside
Leibniz Institute of Marine Sciences
Mojib Latif, Leibniz Institute of Marine Sciences Johann Jungclaus, Max-Planck-Institute for Meteorology Erich Roeckner, Max-Planck-Institute for Meteorology

Can multi-models really enhance the seasonal predictions?
Andreas Weigel
Federal Office of Meteorology and Climatology (MeteoSwiss), Zurich, Switzerland
M. Liniger (MeteoSwiss), C. Appenzeller (MeteoSwiss)

Sensitivity of seasonal forecasts to resolution and physics
Jean-François Guérémy
Météo-France/CNRM

Dynamical Subseasonal Forecasting at NCEP: On the importance of atmospheric resolution and initial conditions
Augustin Vintzileos
UCAR - EMC/NCEP/NOAA
Hua-Lu Pan EMC/NCEP/NOAA

Issues in Initial Value Climate Prediction
Alberto Troccoli
ECMWF
Tim Palmer (ECMWF)

Seasonal predictions at INGV-CMCC: sensitivity to improvements in ocean initial conditions.
Andrea Alessandri
National Institute of Geophysics and Volcanology (INGV), Italy
A. F. Carril (INGV), P. Di Pietro (INGV), S. Gualdi (INGV-CMCC), S. Masina (INGV-CMCC), A. Navarra (INGV-CMCC)

Initialization of a global climate model with oceanic reanalysis
Holger Pohlmann
Max Planck Institute for Meteorology, Hamburg, Germany
J. Jungclaus (Max Planck Institute for Meteorology), J. Marotzke (Max Planck Institute for Meteorology)

MERCATOR Global Ocean Data Assimilation System: Presentation and Applications for Coupled Ocean-Atmosphere Seasonal Forecasting
Nicolas Ferry
E. Remy (MERCATOR-OCEAN), M. Drevillon (MERCATOR-OCEAN), B. Tranchant (MERCATOR-OCEAN), C-E Testut (MERCATOR-OCEAN), J-P Piédeliévre (METEO FRANCE)

POSTER SESSION II - Seasonal Prediction Skill and the Coupled System

A New Atmosphere-Land Initialisation (ALI) Scheme for POAMA, the Australian Bureau of Meteorology's Seasonal Forecast System.
Debbie Hudson
Bureau of Meteorology Research Centre (BMRC), Australia
D. Hudson (BMRC), O. Alves (BMRC) and L. Shi (BMRC)

Stratospheric Variability as Predictor of Winter Anomalous Precipitation over Europe
Alvaro de la Camara
Dpto Geofisica y Meteorologia, Universidad Complutense de Madrid
E. Serrano (Dpto Geofisica y Meteorologia, Universidad Complutense de Madrid), C. R. Mechoso (Dpt. Atmospheric and Oceanis Sciences, University of California, LA)

How reliable is Eurasian snow cover as a predictor of Northern Hemisphere winter climate?
Christopher Fletcher
University of Toronto, Toronto, Canada.
P. J. Kushner (University of Toronto), J. Cohen (AER Inc.)

Predictability of Cold Spring Seasons in Europe
Mxolisi E. Shongwe
Royal Netherlands Meteorological Institute
Geert Jan van Oldenborgh (1), Christopher A. T. Ferro (2) and Caio A. S. Coelho (3)
(1) Royal Netherlands Meteorological Institute, (2) National Centre for Atmospheric Science, University of Reading, (3) Instituto Nacional de Pesquisas Espaciais, Cachoeira Paulista, Brazil

Seasonal Prediction Skill in the New ENSO Forecast System at Japan Meteorological Agency
Yuhei Takaya
Global environment and Marine Department, Japan Meteorological Agency
Tamaki Yasuda (MRI/JMA), Satoshi Matsumoto (MRI/JMA), Tosiyuki Nakaegawa (MRI/JMA), and Tomoaki Ose (MRI/JMA)

Seasonal prediction skill is evaluated using the new ENSO forecast system developed at Japan Meteorological Agency/Meteorological Research Institute (for details, refer to the companion poster 'The new ENSO forecast system at Japan Meteorological Agency'). We have carried out the 7-month ensemble forecast initiated at the end of January, April, July and October in the 28 years from 1979 to 2006, as a part of the TFSP retrospective seasonal forecast experiment. From the verification of the preliminary 5-member ensemble forecast experiment, it is found that the anomaly correlation of monthly mean SST in Nino3.4 (Nino3) region is 0.75 (0.69) to 6 months lead. These forecast skills surpass current operational model's skills, and show validity as the ENSO forecast model. SST forecast in the tropical Western Pacific region(130W-150E, 0-15E), in which SSTs influence the northern summer extratropical circulation over East Asia including Japan, is much improved. The atmospheric forecast skills are also verified with JRA-25 reanalysis data and CMAP precipitation data. The atmospheric forecast skills are improved or at least neutral for almost all of the tropical and extratropical region compared with the JMA's 2-tiered operational seasonal prediction model. In particular, the northern summer precipitation forecast in the tropical Western Pacific region shows significant improvement probably due to better representations of the air-sea interactions in the coupled model. The northern summer forecast of 500 hPa geopotential height and temperature at 2 m above ground in East Asia has more predictive skill than the operational model. We convince that the new forecast system will become a first step to the operational seasonal forecast with coupled model in JMA.

The New ENSO Forecast System at Japan Meteorological Agency
Tamaki Yasuda
Meteorological Research Institute, Japan Meteorological Agency
Y.Takaya(JMA/CPD), T.Nakaegawa(JMA/MRI), Y.Fujii(JMA/MRI), S.Matsumoto(JMA/MRI), G.Yamanaka(JMA/MRI), M.Kamachi(JMA/MRI), and T.Ose(JMA/MRI)

Japan Meteorological Agency (JMA) has been running a forecast system for El Nino/Southern Oscillation (ENSO) using a coupled atmosphere-ocean model since 1999. Recently, a new version of the forecast system has been developed at JMA/Meteorological Research Institute (MRI), which will replace the current JMA operational system in 2008. The atmosphere component of the new system is a recent version of the JMA atmospheric general circulation model. The resolution has been increased from T42L40 of the current operational system to TL95L40. The ocean component is the MRI Community Ocean Model (MRI.COM). The horizontal resolution has been also increased 2.5 deg (lon) x 0.5-2 deg (lat) to 1.0 deg (lon) x 0.3-1.0 deg (lat), and the vertical resolution is increased from 20 to 50 levels (24 levels in the upper 200m). The atmospheric initial conditions are derived from the Japanese Re-Analysis 25 years (JRA-25) and the JMA Climate Data Assimilation System (JCDAS). Data assimilation system to create initial conditions for ocean is the Multivariate Ocean Variational Estimation (MOVE) System developed at MRI. The analysis method of MOVE System is a Three-Dimensional Variational (3DVAR) method with coupled temperature-salinity empirical orthogonal function (EOF) modes. The retrospective forecast experiment using this system shows good skill in predicting SST variations associated with ENSO. Comparing anomaly correlation and root mean square error of NINO3.4 sea surface temperature (SST) with the current operational version shows clear progress. Moreover, our new system provides important advance in predicting SST in the western equatorial North Pacific (130-150E, 0-15N) which has a crucial impact on Japan seasonal climate. Using this system as an application for climate research, it is suggested that the Indian Ocean Dipole (IOD) has an important role on the rapid termination of 2006/07 warm event in the Pacific.

The Impact of Air-Sea Interaction on Tropical Intraseasonal Variability in the CFS
Kathleen Pegion
George Mason University
B. Kirtman (COLA/GMU) J. Shukla (IGES/GMU)

The simulation and predictability of tropical intraseasonal variability is compared in long simulations of both coupled and uncoupled integrations of the National Center for Environmental Prediction Climate Forecast System (CFS). The simulation of tropical intraseasonal variability in the coupled and uncoupled models is compared to determine the impact of air-sea interaction. Results indicate that the coupled simulation is better organized, has better eastward propagation than the uncoupled simulation, and correctly simulates the phase relationship between precipitation and sea surface temperature (SST). Potential predictability experiments are performed for both the coupled and uncoupled models. The model is initialized with intraseasonal events selected from the coupled simulation. The uncoupled experiments are forced with 'perfect' SSTs while the coupled experiments are allowed to evolve on their own. Predictability estimates indicate that air-sea coupling provides no additional potential predictability. However, 'perfect' SSTs are an unrealistic case. Therefore, predictability experiments with persisted SST anomalies, forecast SSTs, and climatological SSTs are performed to address the sensitivity of the potential predictability of tropical intraseasonal variability to differences in SSTs.

Since 1995, the Canadian Meteorological Center has been producing seasonal temperature and precipitation anomaly forecasts using objective and statistical methods. Forecast for season 1 is generated by numerical models GEM and GCM2. Deterministic and probabilistic products are generated by this ensemble. Recently, 35 years of seasonal hindcasts has been conducted with four different models: the CCCma models AGCM3 and AGCM2, and the RPN models GEM and SEF. This ensemble of hindcast constitute the second Historical Forecast Project (HFP2) which follows the protocol established by the PCMDI for the SMIP2/HFP. The protocol consists in 4-month forecasts for the period 1969 to 2003 for 12 rolling seasons (January-February-March-April, February-March-April-May, etc.) and 10 different realisations for each season. Climatic sea ice extent were prescribed for subsequent months and the sea surface temperatures (SST) were forecast using persistence of the anomalies calculated for the month preceding the integrations. The use of multimodel ensemble significantly improves the forecast, even if a specified model individually perform less than the others. Scores of different combinations of models, for zero lead-time and 1 month lead time, for years El Nino-La Nina and with no significantly sea surface temperature anomaly will be shown. Deterministic and probabilistic results will be shown.

Coupled seasonal forecasting at CCCma - Initial results
W. Merryfield
Canadian Centre for Climate Modelling and Analysis
G. Boer, G. Flato, S. Kharin, B. Pal and J. Scinocca

CCCma has been producing coupled retrospective ensemble seasonal forecasts in which a very simple ocean data assimilation method, relaxation to observed monthly SST, is applied to the CGCM3.1 coupled model. This effort is serving as a prototype for the Coupled Historical Forecast Project, which will produce a suite of retrospective forecasts from a more recent model version having improved ENSO variability, and using more sophisticated (but still relatively simple) initialization techniques. This project is a component of the Canadian research initiative "Global Ocean-Atmosphere Prediction and Predictability" or GOAPP, which is viewed as providing a basis for bias subtraction and calibration of an eventual operational coupled forecast system, as well as contributing to the Task Force on Seasonal Prediction Experiment. Some skill comparisons between these preliminary coupled forecasts and a soon-to-be-implemented upgrade to Canada's two-tier operational system are reported.

Seasonal forecasts: experience at ECMWF with single and multi-model systems
Tim Stockdale
ECMWF
F. Molteni (ECMWF), M.A. Balmaseda (ECMWF), P. Doblas-Reyes (ECMWF) and L. Ferranti (ECMWF)

ECMWF has been running seasonal forecasts in operational mode since 1997. At present we run a single-tier 41 member ensemble coupled model forecast, with TL159 atmospheric resolution. Re-forecasts for the years 1981-2005 are available for calibration. This model has high levels of skill at predicting Nino SST variability, but its performance at predicting temperature and rainfall fluctuations over land areas is still deficient. In collaboration with the Met Office and Meteo-France, ECMWF organizes an operational multi-model forecast system known as EUROSIP. The advantages and current limitations of this multi-model system will be described. A major factor that limits our ability both to assess seasonal forecast skill, and to make optimal multi-model products, is the very limited period of the past for which re-forecasts are available. Although we may be able to extend this to O(50 years), this would still be insufficient for many purposes. The eventual resolution of this problem may depend on us being able to make far-reaching improvements to the forecast models.

GloSea4: the new Met Office seasonal ensemble prediction system
Alberto Arribas
Met Office
S. Cusack (Met Office), A. Maidens (Met Office), M. Gordon (Met Office), A. Scaife (Met Office)

GloSea4, the new Met Office seasonal ensemble prediction system is expected to become operational in 2009. It is to be developed around the new Hadley Centre model, HadGEM3, and aims to address the main deficiencies of the current system: inappropiate initialisation of land surface conditions; improper representation of uncertainties (for both, model and initial conditions); and the reduction of seasonal forecast's skill as consequence of model biases and drifts. A summary of the scientific and technical plans for the development of the new system will be presented.

The results of SMIP-2 and SMIP-2/HFP experiments using global semi-Lagrangian finite-difference vorticity-divergence SL-AV model (Tolstykh, 2001) are described. A brief description of this model is given. The version of SL-AV model for seasonal prediction has the resolution 1.40625x1.125 degrees lon/lat and 28 levels. Parameterizations are from Meteo-France ARPEGE/IFS model with minor modifications. Initially, the version of SL-AV model for seasonal prediction used very simple parameterization of surface processes. Now the ISBA scheme is implemented along with more recent version of the PBL parameterization (interactive mixing length) and hindcast experiments for SMIP-2 and SMIP-2/HFP protocols are repeated. The sensitivity of results to this change is described. The SL-AV model now produces quasioperational seasonal forecasts and contributes to the APCC multi-model ensemble effort. Some results from these activities are presented.

The intraseasonal variability associated with the Asian summer monsoon as simulated by seven coupled general circulation models (CGCMs) are analyzed and validated against observations. The model hindcasts are produced by the DEMETER project. Each hindcast is an integration of six months starting from 1 February, 1 May, 1 August and 1 November and comprises an ensemble of nine members. All seven models have been run for a common period of 1980-2001. The focus is on the spatial and seasonal variations associated with the summer monsoon intraseasonal oscillations (ISO) of outgoing longwave radiation (OLR), their large-scale organization, propagation characteristics, the air-sea coupling, their 'deterministic' predictability and implications on seasonal predictability. A multi-variate Local Mode Analysis (LMA, essentially a complex EOF analysis over a window of 90 days moving over the 180 days of hindcasts) has been utilized in order to evaluate the above characteristics of ISOs in the hindcasts against observations. Most models have problems in simulating large-scale organized perturbations of the convection. In addition, perturbation patterns are more variable from one intraseasonal event to another compared to observation. However, most models do exhibit some form of northeastward propagation of the perturbations over the Indian Ocean. Models with poor seasonal cycle tends to have larger biases in the northeastward propagation and organization. One possible source of deficiency in organizing intraseasonal large-scale convective perturbation could be the air-sea interaction. The analysis of the nature of coupling in the hindcasts indeed shows that most models simulate too weak SST perturbations and systematic phase quadrature between OLR and SST, indicative of a slab-ocean-like response of the temperature to surface flux perturbations. Simulation done with the same AGCM and different OGCMs tend to have similar coupled perturbations, indicative of the importance of atmospheric processes in defining the nature of the intraseasonal SST perturbation. Mainly because of their relatively coarse vertical resolution, the different OGCMs used are however limited in their ability to represent intraseasonal processes, such as warm layer formation, which are important for the amplitude of the SST perturbations at intraseasonal time-scales and possibly for the triggering of organized convection. Evaluation of the predictability at the ISO time scale (10-50 days) is also performed on the basis of pentad mean OLR maps. Results show a better predictability in the summer (1 May initial conditions) hindcasts compared to the winter hindcasts (1 November initial conditions). This is possibly due to the better predictions and consistency among all ensemble members of the strong seasonal cycle and the embedded ISOs in summer compared to the weaker seasonal cycle in winter. Results show no particular skill to predict the amplitude of the intraseasonal variability of the convection over the Indian Ocean basin for the coming season (after two months of hindcast).

The reproducibility of the boreal summer intraseasonal variability (ISV) and its interannual variation are assessed by diagnosing twenty one-year hindcast outputs from ten state-of-the-art prediction models that include atmosphere-ocean coupled processes. To facilitate the assessment, we define the strength of the ISV activity by the standard deviation of [intraseasonally] filtered precipitation. The observed climatological ISV activity exhibits largest values over the western North Pacific and Indian monsoon regions. The interannual variation of ISV activity is primarily found over the western North Pacific in observation while most models have the largest variability over the central tropical Pacific and exhibit a large range of spatial patterns that are different from the observation. However, the leading EOF modes of the ISV activity in the models are closely linked to the model's El Nino-Southern Oscillation (ENSO), a feature that resembles the observed ISV and ENSO relationship. The ENSO-induced easterly vertical shear anomalies in the western and central tropical Pacific, where the summer mean vertical wind shear is weak, result in the ENSO-related changes in ISV activity in both the observation and the models. Because of the presence of a close relationship on temporal variation in the dominant mode of ISV activity between the observation and models, a statistical correction method based on the singular value decomposition is shown to be able to remove a large portion of the systematic errors. The 21-year-averaged pattern correlation changes from 0.25 to 0.65 over the entire Asian monsoon region after the bias correction in the multi-model ensemble mean.

The limits of predictability of El Nino and the Southern Oscillation (ENSO) and Asian-Australian monsoon in coupled models are investigated based on retrospective forecasts made with 13 different coupled GCMs. The coupled GCM datasets that come from the APCC/CliPAS (APEC Climate Center/Climate Prediction and its Application to Society) and DEMETER (European Multimodel Ensemble system for seasonal to inTERannual prediction) projects are used. The 13 models are fully coupled ocean-land-atmosphere dynamical seasonal prediction systems with 5- to 9-month integrations for 3 to 15 different initial conditions for four seasons in the common 23 years from 1981 to 2003. As a baseline, the dynamic-statistical SST forecast and multiple AGCM forecasts forced by it are also compared. The influence of initial uncertainties and model errors associated with coupled ENSO dynamics on forecast error growth are discussed. Monsoon predictability associated with SST forecast is also investigated. In coupled models, the realism of variability is related to the accurate simulation of the mean state. The predictability with respect to ENSO phase shows that the phase locking of ENSO to the mean annual cycle has an influence on the seasonal dependence of skill, since the growth phase of ENSO events is more predictable than the decay phase. The overall characteristics of predictability in the coupled system are assessed by comparing the forecast error growth and the error growth between two model forecasts whose initial conditions are one month apart. Because the model errors are so systematic, their influence on the forecast skill is investigated by analyzing the erroneous features in a long simulation. The main analysis focuses on the CFS and the SINTEX, SNU and UKMO models are analyzed also, since they provide both more than 50-year control simulations and 23-year forecasts. The systematic errors in the long run are reflected in the forecast skill as a major factor limiting predictability after the influence of initial conditions fades out with respect to lead time.

The study of the Rossby wave activity over the North Atlantic region is analysed in this study describing the Principal Components of the interannual 200hPa streamfunction winter anomalies (y200, NDJF) in the 5S-80N / 90W-40E domain. The analysis reveals that the leading y200-EOF mode does not project on the North Atlantic Oscillation (NAO) and it doesn't have a significant impact on the European anomalous precipitation. The projection of the leading mode of winter y200 onto the global SST anomaly shows a clear El Nino signature and no signal is observed over the Atlantic. When projecting it globally, two wavelike patterns appear over the northern hemisphere: a regional response over the North Atlantic-European region and other over North Pacific-American sector which resembles the Tropical/Northern Hemisphere pattern. The NAO, which is thought to be the leading mode of variability in the North Atlantic region, appears when projecting the second y200 mode. The homologous projection onto the global anomalous SST shows the Tripole pattern and no signal over the Pacific. These results reveal that the leading perturbation of the rotational circulation over the North Atlantic comes from Pacific-El Nino but its impact over the extratropical atmosphere and, in particular, over the European precipitation is weak although statistically robust.

Recent Trends in the Connection Between Atlantic and Pacific Ninos
Belén Rodríguez de Fonseca
Dpto Geofísica y Meteorología.Facultad de Ciencias Físicas. Universidad Complutense de Madrid.
Irene Polo (1), Javier García Serrano(1), Carlos Roberto Mechoso (2)
(1)Dpto Geofísica y Meteorología.Facultad de Ciencias Físicas. Universidad Complutense de Madrid. (2) Dptmt of Atmospheric Sciences.University of California at Los Angeles. USA

The Atlantic Equatorial mode is an important component of the interannual variability of the coupled atmosphere-ocean system in the Tropical Atlantic. The mode, which is also referred to as the Atlantic Nino, impacts regional rainfall, mainly over the Gulf of Guinea. There is evidence that remotely forced variability affects much more the Atlantic Nino than the Pacific Nino. Several authors, using observations from the 50's, have reported connections with a six-month lag between the mode and previous Pacific Nino anomalies. The present work revisits these reported connections by using recent enhanced data over the tropical Atlantic from 1979. For this period, the leading influence of the Equatorial Pacific onto the Atlantic Nino is verified. However, it is also found that the leading role of the Tropical Atlantic on precipitation and SST anomalies in the Pacific is much more significant. The lead-lag correlation scores of the anomalous tropical SST (and precipitation) between these two tropical basins show a 3-4 years oscillation, which is significant from 1979, whilst the same analysis performed from the 1950's shows the same oscilation with no significant scores. Possible explanations for this behaviour include the recent trends in warming rates of the individual ocean basins, which are characterized by stronger warming in the Atlantic from the 80's.

We present some modelling and observational evidence for the hypothesis that the atmospheric bridge from the tropical east Pacific to the tropical west Pacific can modify the extratropical ENSO response in the North Pacific region in late fall/early winter. The model simulations are various sets of regionally-coupled ensemble simulations with the GFDL-R30 model. In these so-called "pacemaker" experiments, observed sea surface temperatures (SST) are prescribed in the tropical eastern Pacific, whereas outside this region a bulk-mixed mixed layer model is coupled to the atmosphere. We also performed simulations with coupling confined to the tropical Indian-Western Pacific (IWP) oceans, which indicate that most of the impact of coupling on the extratropical ENSO response is due to tropical coupling in IWP. First, we show that the observed circulation in the North Pacific in late fall/early winter (November/December) is very sensitive to forcing in the western tropical Pacific, with positive convective anomalies being associated with a deeper Aleutian low; a sensitivity that is well-reproduced in our model simulations. Then we show that, in the simulations, bridge-induced (coupled) warm SST anomalies in the western tropical Pacific tend to increase convection in this region during El Nino events, which in turn leads to a stronger extratropical response. Next, observational analysis are presented indicating that, in nature, El Nino events characterized by warm (cold) SST anomalies in the western Pacific in late fall/early winter also result in a strong (weak) extratropical response. We suggest that this effect might be deterministic and associated with the response of the western tropical Pacific to particular "flavours" of El Nino, which might lead to some predictability of the ENSO response during late fall/early winter.

Interdecadal variability of ENSO properties and prediction lead time
Wenju Cai
CSIRO Marine Atmospheric Research
Tim Cowan (CSIRO Marine and Atmospheric Research)

The predictability of El Nino-Southern Oscillation (ENSO) seems to fluctuate on interdecadal time scales. During 1980-2000, the discharge-recharge paradigm appears to operate well, in which signals in the thermocline provide a precursor to Nino3.4 by some 7 months. Using the Simple Ocean Data Assimilation with the Parallel Ocean Program (SODA-POP) reanalysis, which extends to the pre-1980 period, previous studies have found the lead time for the 20-year period before 1980 is significantly shorter; this seems to be re-occurring for ENSO since 2000. Here we address whether the varying lead time is associated with the varying properties of ENSO. These properties include the strength of ENSO as determined by Nino3.4, the meridional extent of ENSO anomalies, and the degree to which the extra-tropics is involved in discharge-recharge process. We explore the possible linkage between these properties and the lead time in SODA-POP and in a multi-century coupled model control experiment. The implications will be discussed.

Realistic greenhouse gas forcing and seasonal forecasts
Liniger Mark
Federal Office of Meteorology and Climatology, MeteoSwiss
H. Mathis (1), C. Appenzeller (1), and F. J. Doblas-Reyes (2)
1. Federal Office of Meteorology and Climatology, MeteoSwiss 2. European Centre for Medium-Range Weather Forecasts (ECMWF)

This contribution investigates the improvement of seasonal forecasts by including realistically varying greenhouse gas (GHG) concentrations. Forecasts starting every May and November are compared over the period 1958 until 2001. One set has constant GHG concentrations while an other one has a realistic GHG trend. The large scale temperature trends derived at different lead times are compared in between the forecast sets and observations over the entire 44 years. It is shown that after a few months the anthropogenic climate change signal is lost up to 70% although it was present in the initial conditions. The differences in trends vary with lead times, seasons and regions. Strongest effects are found in the Tropics and the Summer Hemispheres, in particular the Northern One. On local scale, the improvement is not widespread in trends and very weak in predicting detrended interannual variability. Both sets exhibit a strong absolute temperature bias.

The International CLIVAR Climate of the 20^th Century Project (C20C; Folland et al., 2002) is focused on simulating and understanding climate variability out to multi-decadal time scales during the 20^th and early 21^st centuries. This includes understanding the impact of changes in sea surface temperature (SST), sea ice, land surface conditions and atmospheric composition and the attribution of the major climate anomalies of the past 130 years, and most recently, the stratosphere. These goals are complementary to those of the WMO/WCRP Working Group on Seasonal-to-Interannual Prediction (WGSIP), which, among other aims, seeks to assess the predictability of the climate system on seasonal to decadal time scales, including understanding and documenting the predictability of climate models over the historical record.

While experiments have previously been conducted primarily by prescribing observed global forcing (i.e., global SST, sea ice and atmospheric composition), a series of regionally coupled or "pacemaker" experiments have been undertaken by C20C groups in which the SST in the eastern tropical Pacific SST (or some other region) is prescribed from observations, but coupled air-sea feedbacks are maintained in the other ocean basins (e.g. Lau and Nath, 2003) in order to maintain an energetically consistent ocean-atmosphere interface. Some coupled model investigations have also been carried out (e.g. Knight et al., 2006) to study impacts of Atlantic Multidecadal Oscillation SST on regional climate.

Another issue is the response of the climate system to the changes in land cover and land use. Over the past 250 years, human settlement has made extensive changes to the landscape, primarily in the form of transforming forests into pastures and croplands. Given current trends, it is expected by 2100 that most of the natural vegetation will disappear in Africa and parts of Asia and that there will continue to be a reintroduction of forests in Europe and North America. The impact of land cover change (LCC) has been explored in a separate set of C20C experiments.

This paper will review the results of pacemaker and LCC experiments recently reported at the 4^thInternational C20C Workshop, held at the Met Office Hadley Centre in Exeter, U.K. in March 2007. A summary of plans for the experiments anticipated in the next two years will also be presented.

References
Folland, C.K., Shukla J., Kinter J.,and M. J. Rodwell, 2002: C20C: The Climate of the Twentieth Century Project. CLIVAR Exchanges, 23, 37-39.
Knight, J.R., Folland, C.K. and A.A. Scaife, 2006: Climatic Impacts of the Atlantic Multidecadal Oscillation. Geophys.Res. Lett., 33, L17706. doi: 10.1029/2006GL026242.
Lau, N-C., and M. J. Nath, 2003: Atmosphere-ocean variations in the Indo-Pacific sector during ENSO episodes. J. Climate, 16, 3-20.

Report on WCRP-WGNE meeting on model systematic errors, held in San Francisco, February 2007
Michel Déqué
Météo-France/CNRM

On 12-16 February, the third WCRP-WGNE meeting on systematic errors in numerical models of the atmosphere and the oceans was held in San Francisco, locally organized by PCMDI. Although the main topic of these days was the IPCC-AR4 simulation, seasonal forecasting activities were addressed, in particular the aspects of ocean-atmosphere coupling. The detrimental role on forecast scores of systematic biases, the tier-2 versus tier-1 approach, and the progresses in 6-month ENSO prediction were discussed. A climate model validation strategy, which is a simplification of the TFSP exercise, was agreed upon.

The seasonal forecasting skill of the IPCC version of the model is investigated in an extensive suite of hindcasts: Four hindcast per year of at least seven months length, with nine ensemble members, extending from 1960 till present. Hindcasts are initialised using a coupled assimilation scheme in which SST data are nudged into the climate model in the tropical band. No other observations are used. The model's seasonal forecast skill is demonstrated to be comparable to that of other state-of-the art GCMs. The hindcast setup adopted here is similar to that used in a previous set of hindcasts with an earlier version of the model. The skill of those hindcasts was markedly poorer than the latest set. The skill improvement is demonstrated to be due to the reduction of model systematic bias, and not to difference in ocean initialisation. As an outlook, skillful forecasts, using a similar setup, of North Atlantic decadal climate variability will be briefly presented.

Probabilistic seasonal forecasts with ensemble prediction systems have found a wide range of applications, and their importance grows continuously. Forecast uncertainties due to model error can thereby be considered in a pragmatic way by combining single model ensembles to a multi-model ensemble. Indeed, in earlier studies it has been shown that seasonal forecasts issued on the basis of such a multi-model on average outperform any single model strategy. However, given that a multi-model contains information of all participating models, including the less skillful ones, the question remains as to why, and under which conditions, a multi-model can have higher skill than if simply the best participating single model is chosen (assuming that the 'best' model can be identified). It is the aim of this study to resolve this supposed paradox by (i) applying a synthetic climate forecast toy model, and (ii) by evaluating the 'structure' of skill gain in real seasonal multi-model forecasts (DEMETER data set). The toy model is designed such that it allows the generation of perfectly calibrated single model ensembles of any ensemble size and prediction skill. Additionally, the degree of ensemble 'overconfidence' (i.e. underdispersion) can be prescribed. Multi-model ensembles are then constructed both from weighted and unweighted averages of these single model ensembles. As a skill metric, we apply the "discrete ranked probability skill score" (RPSSd), which is favorable in the context of multi-model studies since it is insensitive to changing ensemble size. Using this toy model, systematic model-combination experiments are carried out to investigate how multi-model performance depends on ensemble size, prediction skill and overconfidence characteristics of the participating single models. The central conclusion is that multi-model ensembles can indeed outperform a "best model approach", but only under certain conditions concerning the ensemble overconfidence. This conclusion is then substantiated by quantifying the gain in skill in a real seasonal multi-model, using near surface temperature forecasts from the DEMETER data set. It is shown that also in the real case the success of multi-model combination is closely linked to the dispersion characteristics of the participating single model ensembles.

Sensitivity of seasonal forecasts to resolution and physics
Jean-François Guérémy
Météo-France/CNRM

Two kinds of sensitivity experiments regarding coupled seasonal forecasts will be presented. These forecasts have been performed using ARPEGE-Climat coupled to OPA using the OASIS coupler. First, the impact of changing the horizontal resolution of the atmospheric model will be shown. This study has been carried out in the frame of FP6 MERSEA Project. The experimental set-up consists in twin experiments covering the years 1992 to 2201, winter and summer seasons with ensembles of 9 members. The control experiment has been performed using the standard horizontal resolution of 300 km, while the perturbed one has made use of a stretched grid which resolution ranged from 50 to 300 km (the stretched pole being located in the Mediterranean Sea. Second, the impact of changing the physics of the atmospheric model will be shown. This study has been started in the frame of FP6 AMMA Project. The experimental set-up consists in twin experiments covering the years 1958 to 2201, winter and summer seasons with ensembles of 9 members. The control experiment has been performed using the standard physical package, while the perturbed one has made use of a new set of parameterizations including turbulence and convection.

Dynamical Subseasonal Forecasting at NCEP: On the importance of atmospheric resolution and initial conditions
Augustin Vintzileos
UCAR - EMC/NCEP/NOAA
Hua-Lu Pan EMC/NCEP/NOAA

In order to investigate the role of the atmospheric model resolution and initial conditions on subseasonal prediction we performed a series of retrospective forecasts with the NCEP Climate Forecasting System (CFS) under NOAA's Climate Test Bed (CTB). The CFS was run at three different resolutions: T62, T126 and T254 equivalent to 200km x 200km, 100km x 100km and 50km x 50km respectively. The model was initialized by the NCEP operational analysis and by the NCEP Reanalysis-2 every five days apart starting from May 23rd to August 11th for years from 2002 to 2006. Reforecast were run with lead times up to 60 days. We will discuss the impact of the model configuration and its initialization on the boreal summer Tropical Intraseasonal Oscillation and the North American and West African Monsoons.

Issues in Initial Value Climate Prediction
Alberto Troccoli
ECMWF
Tim Palmer (ECMWF)

Sensitivity experiments using a coupled model initialised with real atmospheric and oceanic observations are used to investigate the interannual-to-decadal potential for predictability. The idea is to explore the potential for extending the usual seasonal predictions to longer time scales using the same coupled model configuration and initialisation procedure. It is found that, despite considerable model drift, climatic signals on interannual-to-decadal timescales appear to be detectable once the drift is accounted for. Two climatic states have been chosen, one starting in 1965, i.e. ahead of a period of global cooling, and one in 1994, ahead of a period of global warming. The impact of initial conditions and of the different levels of greenhouse gases (GHGs) is addressed in order to gain insights on the source of predictability. From a regional analysis of these experiments, it emerges that initial conditions are generally short-lived and that GHGs are particularly effective at modifying both temperature and precipitation patterns over specific areas.

Using the version 1 of the INGV-CMCC Seasonal Prediction System (ISPS-v1) developed in the framework of the EU project DEMETER, we investigate the impact of improvements of the ocean initial conditions on hindcast skill. In DEMETER, the ocean initial conditions were produced through off-line integrations in which ERA40 fluxes were imposed as boundary conditions to the oceanic model and by a strong relaxation of the Sea Surface Temperature to observations. Taking as a reference the set-up used in DEMETER, we tried to improve the ocean initial conditions in two different and distinct approaches. In the first approach the horizontal resolution of the off-line ocean model has been increased. In the second one a Reduced Order Optimal Interpolation procedure for the assimilation of temperature and salinity profile data has been used during the off-line ocean model integrations. Different sets of 9 member ensemble hindcasts have been produced for the period 1992-2001 taking the same initial conditions for all the coupled model components but the ocean. The sensitivity of the ISPS-v1 hindcasts skill to Temperature and Salinity assimilation during the ocean initialization has been evaluated with particular detail. The effects on both the ENSO phenomenon representation and regional skill are assessed by comparison with the results obtained with DEMETER-like ocean initial conditions. The present study is part of the research efforts currently going on at INGV-CMCC in the framework of the EU projects ENSEMBLES and MERSEA.

The intention of this project is to improve the qualitative and quantitative climate prediction capabilities. For this purpose the coupled model of the Max Planck Institute for Meteorology consisting of the atmosphere model ECHAM5 and the ocean model MPI-OM is initialized with oceanic reanalysis data from the Estimating the Circulation and Climate of the Ocean (ECCO) project. The role of the initial conditions for seasonal to decadal climate predictions is investigated in hindcast experiments over the second half of the 20th century.

MERCATOR Global Ocean Data Assimilation System: Presentation and Applications for Coupled Ocean-Atmosphere Seasonal Forecasting
Nicolas Ferry
E. Remy (MERCATOR-OCEAN), M. Drevillon (MERCATOR-OCEAN), B. Tranchant (MERCATOR-OCEAN), C-E Testut (MERCATOR-OCEAN), J-P Piédeliévre (METEO FRANCE)

Mercator Ocean is the French operational ocean monitoring and forecasting centre. Its core mission consists in the simulation of the global ocean with primitive-equation high resolution models, assimilating satellite and in situ data, to provide on an operational basis hindcasts and near-real time nowcasts and forecasts of the global ocean circulation, i.e. continuous and well-assessed information on the ocean state from global to regional scales. Mercator Ocean weekly provides estimates of the ocean circulation and thermodynamics at high resolution from global to regional scales. Its products are already used by more than 150 referenced users from various communities: public bodies such as meteorological services and agencies dealing with the ocean and its environment, as well as private bodies that are directly linked with the customers operating in the marine environment. The objective of this poster is twofold. First we describe the Mercator global ocean analysis / forecasting system, more specifically the ocean models used and the assimilation techniques employed to constrain the ocean state. Then we present a particular application of the ocean analyses which is the initialisation of coupled ocean-atmosphere models for seasonal forecasting. At present time, two global ocean configurations are available: a low resolution (~2∞) model, well suited for climate application (i.e. seasonal forecasting) and higher resolution model (1/4∞) describing the ocean at the meso scale. These ocean general circulation models are based on the OPA-NEMO code and use an ORCA grid. Besides, Mercator has developed a series of assimilation techniques in order to optimally combine observations (SLA, SST, in situ Temperature & Salinity profiles) with ocean model simulations. Two different assimilation techniques are currently developed. They both constrain the ocean state in a multivariate way and allow assimilating in situ (temperature and salinity profiles) as well as remotely sensed (SLA, SST) data. The first data assimilation method is based on Reduced Order Kalman Filters using 3D multivariate modal decomposition of the forecast error covariance. The use of 3D modal representation for the error statistics is intended to improve analyses in highly inhomogeneous and anisotropic regions of the ocean. The second assimilation technique is based on a 3D variational approach. We shall review how these different assimilation schemes impact and improve the ocean state estimation. Results from hindcast analyses are presented and discussed. Finally, we show how these global ocean analyses are used in an operational context for coupled seasonal forecasting at Météo France.

This study presents a new Atmosphere and Land Initialisation scheme (ALI) for POAMA (Predictive Ocean Atmosphere Model for Australia). POAMA is a coupled ocean/atmosphere model seasonal forecast system, which has been run operationally by the Bureau of Meteorology since 2002. For atmospheric initialisation, the current system uses data from the NWP forecast system for the real-time forecasts, and data from an AMIP-style simulation for hindcasts. The initial conditions for the hindcasts thus contain observed atmospheric information which is related to sea-surface temperature, but they do not capture the true intra-seasonal state. This is true also for the initialisation of the land surface, which uses an AMIP-style climatology for both the hindcasts and forecasts. The rationale behind developing a new initialisation scheme is that, firstly, real atmospheric initial conditions may be important for intra-seasonal forecasting. Results from the current POAMA system appear to support this hypothesis. In particular, the impact of the Madden-Julian Oscillation (MJO) in generating the spread of 90-member ensemble forecasts in the first four months of the 1997/98 El Nino is discussed. The MJO is important for Australian climate variations primarily through impacts on the onset and breaks of the summer monsoon and tropical cyclone genesis. In addition, there are indications that the MJO may affect the development of ENSO events, interannual variability of monsoon rainfall, as well as rainfall over higher latitudes via teleconnections. Information of the MJO in the initial conditions thus has the potential to improve our ability to forecast these events. Secondly, there may be benefits for intra-seasonal/seasonal forecasting from improving the initialisation of the land surface, primarily due to soil moisture memory in the earth-atmosphere system. ALI uses a forecast-analysis (or nudging) scheme to produce the atmospheric and land surface initial conditions for the hindcasts and real-time forecasts. In this approach, an offline version of the POAMA atmospheric model is nudged towards 'reality', or an 'analysis', provided by ERA-40 for the hindcasts and the NWP forecast system in real-time. The model's forecast of u-wind, v-wind, atmospheric temperature and humidity is compared directly to the analysis at six-hour intervals, and a fractional difference between the forecast and the analysis is added repeatedly to the evolving model atmospheric state. The land surface is initialised indirectly via the nudged atmosphere, such that the soil moisture and temperature evolve to become consistent with the atmospheric forcing. The atmospheric model (forced by observed weekly sea-surface temperatures) is run in forecast-analysis mode for the 25-year hindcast period, allowing an initial spin-up year for the land surface. The output from this simulation is used as initial conditions for the hindcasts. This approach of initialising the land surface does not attempt to correct biases in the land surface model, but it may offer improvements over using climatological land initial conditions. ALI allows consistency between the hindcasts and real-time forecasts, and introduces more realistic atmosphere and land initial conditions into the hindcasts. Since the hindcasts will now use more realistic atmospheric states and will be much more consistent with real-time forecasts, it allows better use of the hindcasts to assess intra-seasonal and seasonal forecast skill. Initial results of the impact of the ALI scheme on forecast skill will be presented.

Encouraged by previous works that show stratospheric variability leading tropospheric variability in the Northern Hemisphere (NH) winter, we have analyzed a possible lagged connection between anomalies in the stratospheric flow and winter climate over Europe. In contrast to other authors, we have based our study on an index of monthly winter precipitation anomalies. This index (pcp_idx) is the standardized expansion coefficient of precipitation of the first SVD mode between winter precipitation over Europe and 1000-hPa geopotential in the Northern Hemisphere. (This first mode accounts for the 38% of the total variance of precipitation). Positive values of pcp_idx are associated both with increased synoptic eddy activity and westerly winds in the North Atlantic (around 60° N). This finding is consistent with an enhanced (decreased) number of cyclones crossings of northern (southern) Europe. We have found that during winter months with pcp_idx values lower than -1 (higher than +1) the stationary planetary wave activity is more (less) likely to propagate from the troposphere into the lower stratosphere, where it tends to decelerate (accelerate) the mean flow. The stratospheric polar night jet weakening (strengthening) appears about three weeks before the beginning of a month with positive (negative) precipitation anomalies over southern Europe. These results support the use of stratospheric anomalies as predictors in the seasonal forecasting of winter anomalous precipitation in Europe.

This paper re-examines the question of whether autumn Eurasian snow cover provides seasonal predictability for Northern Hemisphere winter climate anomalies. Previous observational and modelling studies have shown that predictability from snow cover may arise through excitation of an annular-mode signature in the stratosphere, with subsequent downward propagation from the stratosphere back to the surface on a timescale of several weeks. Robust lower-tropospheric precursors of these stratospheric events are difficult to find because of strong natural variability in the wintertime polar stratosphere. We demonstrate the reliability of a snow-forcing precursor for these events in a large ensemble of integrations using the Geophysical Fluid Dynamics Laboratory Atmosphere/Land model AM2/LM2. These idealised experiments place a useful bound on the predictability that can be expected from using Eurasian snow cover as a predictor, and also provide insight into the dynamics of the large-scale circulation response to snow forcing.

Predictability of Cold Spring Seasons in Europe
Mxolisi E. Shongwe
Royal Netherlands Meteorological Institute
Geert Jan van Oldenborgh (1), Christopher A. T. Ferro (2) and Caio A. S. Coelho (3)
(1) Royal Netherlands Meteorological Institute, (2) National Centre for Atmospheric Science, University of Reading, (3) Instituto Nacional de Pesquisas Espaciais, Cachoeira Paulista, Brazil

The seasonal predictability of cold spring seasons (March-April-May) in Europe from hindcasts/forecasts of three operational coupled general circulation models (CGCMs) is investigated. The models used in the investigation are the UKMO GloSea, ECMWF S2 and the NCEP-CFS. Using the relative operating characteristic score and the Brier skill score the long-term prediction skill for spring 2-m temperature in the lower quintile (20%) is assessed. Over much of central and eastern Europe the predictive skill is found to be high. The skill of the UKMO GloSea and ECMWF S2 models significantly surpasses that of damped persistence over much of Europe but the NCEP-CFS model outperforms this reference forecast only over a small area. The higher potential predictability of cold spring seasons in eastern relative to western Europe can be attributed to snow effects as areas of high skill closely correspond with the climatological snowline, and snow is shown in this paper to be linked to cold spring 2-m temperatures in eastern Europe.

The ability of the models to represent snow cover during the melt season is also investigated. The UKMO GloSea and the ECMWF S2 models are able to accurately mimic the observed pattern of snow cover, but the NCEP-CFS model predicts too short a snow season. Improvements in the snow analysis and land surface parameterizations could increase the skill of seasonal forecasts for cold spring temperatures.

last updated Wed, Sep 2, 2009 by Anna Pirani