Скачать 39.39 Kb.
OPERATIONAL PROCEDURES FOR DISSEMINATION AND EXCHANGE OF PRODUCTS, TERMS AND CONDITIONS
Status of GPCs and their nomination
(Submitted by Dr Richard GRAHAM)
Summary and Purpose of Document
This document presents the Long range Forecasting Progress Report for 2005 in the U.K. Meteorological Office.
The Meeting is invited to take into account this information for its discussions and work.
Met Office, UK
Annual Research Progress Report on Long-Range Forecasting
January 2004 - December 2005
R.J. Graham, B.D. Becker, A. Brookshaw, A.W. Colman, M.K. Davey,
C.K. Folland, M. Gordon, M. Huddleston, S. Ineson, B. Ingleby, M. MacVean, P.J. McLean and
Met Office, FitzRoy Road, Exeter, UK
The current status of current long-range forecasting activities at the Met Office is detailed in the following sections 1 and 2. In this section a summary of progress is provided with regard to the GPC designation criteria (a) to (e).
A fixed monthly production cycle is operated. The precise time of issuance is under review. Currently forecasts are issued in the last week of each calendar month.
The majority of products listed in the manual are provided (Type B products (probabilities for tercile categories) have been provided since July 2004).
c) Provide verifications as per the WMO SVS for LRF.
SVS verification diagnostics are provided on the Met Office website and provision of these to the Lead Centre for SVSLRF has now started.
e) Provide up-to-date information on methodology used by the GPC
This information is provided on the Met Office website and has been submitted to the Lead Centre of SVSLRF.
f) Make products accessible through the GPC website and/or disseminated through the GTS and/or internet.
Products are made available (since January 1998) mainly via the Met Office website.
1. Dynamical prediction systems and products
1.1 Dynamical monthly prediction
Monthly forecast services to a range of users have continued as in previous years. Forecasts are generated using operational output from the ECMWF coupled ocean-atmosphere 51-member monthly-range ensemble system (Vitart, 2003). The model is run weekly from initial conditions at 00GMT Thursday. A hindcast dataset, with the same start time and valid period as the forecast, is available ahead of each forecast using a 5-member ensemble. For forecast calibration, the Met Office’s post-processing uses a rolling 12-year hindcast period, ending with the year prior to the forecast year.
Met Office post-processing is performed for mean, maximum and minimum temperature, precipitation and sunshine amount averaged/accumulated over three forecast periods; days 5-11 ahead, days 12-18 ahead and days 19-32 ahead (for the UK region, forecasts for the 5-11 day period are generated using the ECMWF 10-day EPS). Forecast products include probability forecasts for various regions of the globe with additional focus on the 10 UK climate districts. Global probability products are provided in the form of 1) probability maps for tercile and outer-quintile categories of temperature and precipitation, and 2) for specific regions, probability histograms for quintile categories (well-below, below, near, above, and well-above the climate normal for the region and time of year). For the 10 UK climate districts temperature and rainfall forecasts are generated in terms of quintile categories. Tercile categories are used for sunshine. The UK forecasts are expressed both in terms of the probability of each category, and a deterministic forecast based on either the ensemble mean or the most probable quantile.
1.2 Dynamical seasonal prediction
As in previous years, seasonal forecasts to 6-months ahead have been generated each month using the Met Office’s 41-ensemble coupled ocean-atmosphere global seasonal prediction system (known as GloSea). GloSea is based on the HadCM3 climate model. A performance assessment of the GloSea system is provided by Graham et al., 2005. Operational forecasts are initialised with ocean and atmosphere conditions valid for the first day of the current month. Perturbations to the initial conditions are applied to the ocean component only and are based on 5 parallel ocean assimilations, generated through application of perturbed windstress. Additional instantaneous SST perturbations are applied at initial time to generate the 41 starting states required for the ensemble. The forecasts run on the ECMWF computing facility in parallel configuration with the ECMWF system2 seasonal prediction model as part of a developing European multi-model system (the European Seasonal to Interannual Prediction Project – Euro-SIP).
GloSea forecasts are expressed relative to a model climatology defined for each month of the year from a set of 15-member ensemble integrations initialised at the beginning of each month over the 15-year period 1987-2001. A range of forecast products are made available to NMSs, Regional Climate Outlook Fora, UK government agencies, the public and commercial companies. In 2005, a major upgrade of the Met Office seasonal forecasting web pages was released. Products now available include the following. Forecasts for anomalies in 3-month-average 2-metre temperature and precipitation, at one-, two- and three-month leads - corresponding to months 2-4, 3-5 and 4-6 of the integration. A probabilistic format is used giving probabilities for equi-probable tercile categories and also for two outer-quintile categories (20th and 80th percentiles). In addition to these probability products, maps indicating the most probable tercile category are also provided. Forecast products for monthly-mean Sea Surface Temperature anomalies in the tropical Pacific are also made available. Products may be viewed at www.metoffice.gov.uk/research/seasonal. Verification information indicating forecast performance has been generated, using WMO guidelines, and is available on the website. Verification diagnostics used include ROC curves, ROC score maps and reliability diagrams. On the website, forecasts from the GloSea system may be compared with corresponding forecasts generated using output from the Euro-SIP multi-model forecast database, which currently includes forecast ensembles from the Met Office, ECMWF and Météo-France seasonal systems. Currently Met Office products derived from Euro-SIP comprise an unweighted combination of the Met Office GloSea forecast ensemble and the ECMWF system2 seasonal ensemble.
1.3 Empirical and hybrid empirical/dynamical real-time seasonal prediction
As in previous years, hybrid statistical and dynamical prediction schemes were used to make seasonal forecasts for selected regions of special interest including the East Africa Short rains season (October-December), the west African Monsoon season (July-September) and the NE Brazil wet season (March-May) . The statistical schemes use historical relationships between key sea surface temperature patterns and surface meteorological conditions. For these regions, information from both statistical models and the GloSea coupled model ensemble was combined to obtain best-estimate seasonal rainfall forecasts, which were contributed to Regional Climate Outlook Fora and distributed to National Meteorological Services and drought monitoring agencies in the target regions. Forecasts were also contributed to Regional Climate Outlook Fora and published in the Experimental Long-Lead Forecast Bulletin. Statistical forecasts of July-August temperature tercile probabilities for western Europe were issued on the Met Office website in March and updated in July.
Warm summers were correctly predicted for NW Europe in 2004 and 2005 and a drier than average short rains season was correctly predicted for East Africa in 2005. Our forecasts for the Sahel have indicated above average rainfall. Whilst observed rainfall has been generally higher here than in the drought years of the 1980s and early 1990s, it has not been as wet as predicted. An explanation for this could be that the relationship between inter-hemispheric contrast in SST and rainfall, which is a major contributor to our statistical forecasts, has changed. Possible reasons for such a change include the impact of climate change and altered ecology of the Sahel region related to trends in land use. Improvements to the forecast system to take account of such changes are being investigated. Our GLOSEA model correctly predicted a slightly drier than average season in NE Brazil in 2005, but unpredicted mid season changes in South Atlantic SST anomalies were the likely cause of a poor forecast for 2004.
Forecasts have also been issued, on a monthly basis, for rainfall in the Volta river catchment in West Africa and for inflow into Lake Volta, Ghana. This forecast application was developed with the Volta River Authority Ghana, and forecasts are used to assist management of hydro-electric power generation. A particular challenge in this region is that rainfall anomalies of opposing signs are often observed between the north and south of the catchment, and this was a particularly common feature in 2004 and 2005. The total rainfall, and consequent lake inflow, is thus dependent on the positioning and relative strengths of each pole of the dipole. Slightly below average rainfall and inflow for the 2005 peak season (July-October) was successfully predicted but, except for one longer-lead forecast, predictions for 2004 were too dry.
The Met Office has issued, in June each year, a long range forecast of the state of the winter North Atlantic Oscillation (NAO) averaged over the December to February winter period. In June 2004 the forecast was for a weakly positive NAO of +0.5 s.d. and this was realised in the 2004/5 winter when the NAO index was +0.1 s.d.
This year the forecast was also used to infer a north European winter temperature anomaly. This is possible because of the strong influence of the NAO on winter temperature over the European region (Scaife et al., 2005). The method used is based on the previous May’s Atlantic sea surface temperature (Rodwell and Folland, 2002, 2003), giving a forecast with a lead time of about six months. By projecting the monthly mean SST fields for May onto a predefined North Atlantic tripole pattern, the method correctly predicts the sign of the following winter NAO in two out of three cases. We have also now established that the method predicts the correct sign of winter central England temperature anomalies and Northern European winter temperature anomalies in 2 out of 3 years. Furthermore, new ocean heat content analyses show that some of the years in which hindcasts by this method have failed can be explained by the occurrence of ENSO events which produce additional European signals, confounding the forecast. The statistics of the method are therefore improved in non-ENSO years. Finally, the method has also been shown to be more skillful than dynamical model predictions for the European region in winter. For winter 2005/6 the method indicated colder than average conditions over Northern Europe (10W-50E, 50-70N) and a strongly negative NAO of -1.1 s.d. Details are available at the following web site: http://www.metoffice.com/research/seasonal/regional/nao/index.html.
Output from the method was combined with dynamical model forecasts and observations of the evolving North Atlantic surface and sub-surface temperature anomalies to generate Met Office statements on the prospects for winter 2005/6, first issued in August 2005 and updated monthly. The forecast stated a 2 in 3 chance of a colder-than-average winter for much of Europe and that, if this were to hold true, parts of the UK - especially southern regions – would have temperatures below normal. The forecast and its likely impacts were communicated to users in the Utilities, Finance and Insurance, Defence, Aviation and Transport sectors as well as to local authorities and regional resilience planners and charities, and gained widespread media attention.
1.4 Forecast of annual mean global surface temperature in 2004 and 2005
Each December the Met Office issues a forecast of the global mean surface temperature anomaly (i.e. a combination of global land surface air temperature and global sea surface temperature) for the year ahead. The forecast uses a statistical method that includes a variety of natural and anthropogenic forcing factors, the state of the Atlantic Multidecadal Oscillation (Knight et al, 2005) and a coupled model forecast of the state of El Nino in the first part of the year ahead. In December 2003 we issued a forecast for a global surface temperature anomaly for 2004 of 0.50+-0.12oC (Folland and Colman, 2003) where the uncertainty encompasses the 95% confidence range. The observed global temperature anomaly was 0.44+-0.06 oC (95% confidence range estimate) calculated using an optimum average (Folland et al, 2001) of the HadCRUT2v data set (Jones and Moberg, 2003). So the outcome of this forecast was well within its uncertainty range. In December 2004 the forecast for 2005 was for 0.51+- 0.12 oC (Folland and Colman, 2004). The observed optimally averaged global temperature anomaly for 2005 is still being assessed, but is clearly warmer than 2004.
2. Dynamical prediction studies, model calibration and validation
Validation studies have found that, in common with other models, GloSea forecast probabilities for outer-tercile, outer-quintile and outer-decile events exhibit a bias relative to the observed frequency of the events. In general the bias indicates ‘over confidence’, such that when the chance of observing the event is relatively high, forecast probabilities are too large, and when the chance of observing the event is relatively low forecast probabilities are too small. Such biases may be corrected by calibration techniques that use the statistics of past performance. The merits of a number of such techniques have been contrasted, and the best overall method of those tested found to be discriminant analysis. Discriminant analysis may also be used to combine, in an optimal way, ensemble output from the different CGCM components of the Euro-SIP multi-model system. Investigations of (discriminant) calibrated multi-model products show the main potential for unbiased probabilistic prediction of outer-quintile and outer-decile events is in tropical regions. It is planned to release calibrated probability forecast products on the seasonal forecast website in 2006.
Further improvements were made to the statistical downscaling methods developed to improve long-range forecast skill for UK climate districts. It has been shown that the strategy of using dynamical model forecast data from non-local grid points produces significant gains in skill for monthly-average temperature forecasts at ranges of a few months. Work to perform additional assessments for categorical probability forecasts is in progress.
Investigations into dynamical seasonal predictability, with an extension to inter-annual and decadal timescales are continuing as part of the ENSEMBLES project. ENSEMBLES is an Integrated Project under the 6th Framework Programme (FP6) of the European Union (EU), coordinated by the Met Office. The central project theme is the development of multi-model ensemble-based probabilistic prediction of climate and its impacts from seasonal to decadal and longer (century) timescales. The project started on 1st September 2004 and will run for 5 years. The seasonal to decadal component of ENSEMBLES builds on the FP5 project DEMETER, which demonstrated the superior skill available at seasonal timescales from representation of uncertainties in model formulation through use of a multi-model ensemble comprising 7 European CGCMs. DEMETER also pioneered the integration and testing of user application/impact models with ensemble CGCM output, specifically for crop yield and health applications.
In addition to other improvements, ENSEMBLES models will include, for the first time, realistic concentrations of green house gases, solar forcing and (at initialisation time) volcanic dust. The Met Office DePreSys system, designed specifically for decadal prediction, will also be included in the ENSEMBLES multi-model. In addition to the multi-model approach, alternative (or complimentary) techniques for representing model uncertainties will be investigated. In this respect the Met Office is investigating the benefits of a perturbed parameter technique in which an ensemble is generated by using perturbed versions of the CGCM physics to generate each member. The perturbed model versions are constructed by using different settings (from within a plausible range) for a number of tuneable physics parameters. The method has been previously developed and used to generate ensemble-based probabilistic predictions of climate change (Murphy et al., 2004). Initial conditions for the ENSEMBLES multi-model will be generated using improved ocean analysis techniques and observation datasets formulated as part of the FP5 ENACT project and further developed in ENSEMBLES. Techniques for representing initial condition uncertainty will also be compared. As part of the development and assessment phase of the ENSEMBLES system the GloSea model has been updated to allow realistic green house gas concentrations and volcanic dust (at initialization time), and hindcasts have been run for the 11-year period 1991-2001. Experiments using both the operational method of ensemble initialization (see Section 1.2) and a lagged start method have been conducted. The integrations are made in 9-member ensembles from May and November start dates out to at least 12 months ahead. Runs from May 1965 and 1994 have been integrated to 10years ahead. An extended1960-2001 hindcast set will be employed in final hindcast production starting in 2007.
Folland, C.K. and A. Colman, 2003: Empirical prediction of the global temperature anomaly for 2004. COLA, Experimental Long Lead Forecast Bulletin, 12, (on line). Dec 2003
Folland, C.K., Rayner, N.A., Brown, S.J. Smith, T.M. Shen, S.S. Parker, D.E., Macadam, I., Jones, P.D., Jones, R.N., Nicholls, N. and Sexton, D.M.H., 2001: Global temperature change and its uncertainties since 1861. Geophysical. Research Letters, 106, 2621-2624.
Jones, P.D. and A. Moberg, 2003: Hemispheric and large-scale surface air temperature variations: an extensive revision and an update to 2001. J. Climate, 16, pp.206-223.
Murphy, J.M., Sexton, D.M.H., Barnett, D.N., Jones, G.S., Webb, M.J., Collins, M., Stainforth, D.A. 2004: Quantification of modelling uncertainties in a large ensemble of climate change simulations. Nature 430, 768-772.
Rodwell, M.R. and Folland, C.K., 2002: Quarterly Journal of the Royal Meteorological Society, 128, 1413-1443.
Rodwell, M.R. and Folland, C.K., 2003: Annals of Geophysics, 46, 47-56.
Vitart, F., 2003: Monthly forecasting system. ECMWF Technical memorandum No. 424. 68pp.
Internal and External Publications in 2004/05 (non-Met Office authors indicated ‘*’):
Colman, A.W., 2004: Combined Statistical/Dynamical Forecast of 2004 Season Rainfall in the Sahel and Other Regions of Tropical North Africa using Observed Ocean and Atmosphere Information from up to June 2004. Experimental Long-Lead Forecast Bulletin 13 No. 2 (2004).
Colman, A.W., 2004: Forecast of East African Rainfall for October-December 2004 using Dynamical and Statistical Methods. Experimental Long-Lead Forecast Bulletin 13 No. 3 (2004).
Colman, A.W. and Graham R.J., 2004: Forecast of North East Brazil Seasonal Rainfall for February to May 2005 using Empirical and Dynamical Methods and Atmosphere and Ocean Data up to November 2004. Experimental Long-Lead Forecast Bulletin 13 No. 4 (2004). http://grads.iges.org/ellfb/Dec04/colman/colman.htm
Davey, M., Huddleston, M., Ingleby, N., Ineson, S. and *project partners, 2005: ENACT project: final report to the European Commission
Folland, C.K. and A. Colman, 2004: Empirical prediction of the global temperature anomaly for 2005. COLA, Experimental Long Lead Forecast Bulletin, 13, (on line). Dec 2004
Graham, R.J., Gordon, M., McLean, P.J., Ineson, S., Huddleston, M.R., Davey, M.K., Brookshaw, A. and Barnes, R.T.H. 2005: A performance comparison of coupled and uncoupled versions of the Met Office seasonal prediction General Circulation Model. Tellus, 57A, 320-339.
Huddleston, M.R., Bell, M.J., Martin, M.J. and Nichols, N.K., 2004: Assessment of wind stress errors using bias corrected ocean data assimiliation. Q.J.R. Meteorol. Soc., 130, 853-871.
Huddleston, M. and Ingleby, B., 2005: ENACT: a 43-year ocean analysis set. Met Office internal (GMR) report 2005.
Ineson, S. and Dong, B., 2005: Report on HadGEM’s suitability for long-range forecasting. Met Office internal (GMR) report 2005, V.C.1(04/05).
Ingleby, B. and Huddleston, M.: Quality control of ocean temperature and salinity profiles - historical and real-time data; paper submitted to Journal of Marine Systems
Knight, J., Allan R.J., Folland, C.K., Vellinga, M. and M.E. Mann*, 2005: Natural Variations in the thermohaline circulation and future surface temperature. Geophysical. Research Letters, 32, L20708, doi: 1029/2005GL024233.
*Palmer, T. N., *Alessandri, A., *Andersen, U., *Cantelaube, P., Davey, M., *Delecluse, P., *Deque, M., *Diez, E., *Doblas-Reyes, F., *Feddersen, H., Graham, R., *Gualdi, S., *Gueremy, J.-F., *Hagedorn, R., *Hoshen, M., *Keenlyside, N., *Latif, M., *Lazar, A., *Maisonnave, E., *Marletto, V., *Morse, A., *Orfila, B., *Rohel, P., *Terres, J.-M., and *Thomson, M., 2004: Development of a European multi-model ensemble system for seasonal to interannual prediction (DEMETER). Bull. Amer. Met. Soc., 85, 853-872.
Scaife, A.A., Knight, J.R., *Vallis G.K. and Folland C.K., 2005: Geophysical Research Letters, 32, L18715.
Smith D.M., A.W. Colman, S. Cusack, C.K. Folland, S. Ineson and J.M. Murphy: Predicting surface temperature for the coming decade using a global climate model. Submitted.
|This is a preliminary draft made available for the purpose of allowing industry to see the flow and logic of the document. This document is not content||Meeting of expert team on extended and long-range forecasting|
|The purpose of this proposed research is to improve cloud simulation in the Weather Research and Forecast (wrf) meteorological model by assimilating||Note: This document was created from a non-spell-checked ocr scanned version of the original document. Therefore, it is likely that you will find spelling|
|Cost action 289 Spectrum and Power Efficient Broadband Communications progress report||Cost action 289 Spectrum and Power Efficient Broadband Communications progress report|
|Joint wmo technical progress report on the global data processing and forecasting system and numerical weather prediction research activities for 2009||Joint wmo technical progress report on the global data processing and forecasting system and numerical weather prediction research activities for 2011|
|Joint wmo technical progress report on the global data processing and forecasting system and numerical weather prediction research activities for 2008||1About This Document 1Scope of the Document|