ECMWF Newsletter No. 102 – Winter 2004/05 In this issue Editorial Editorial . 1 n page 2 Thierry Lefort (Météo-France) out- Changes to the operational weather prediction system 1 lines the progress that forecasters have made Oin medium-range forecasting for New Cale- METEOROLOGY donia using ECMWF and other forecast products, Starting up medium-range forecasting for New while on page 7 François Lalaurette gives an update on Caledonia in the South-West Pacific Ocean – a not so boring tropical climate. 2 recent developments at ECMWF in predicting tropi- Early medium-range forecasts of tropical cyclones . 7 cal cyclones in the early medium-range, and describes A snowstorm in north-western Turkey some new products available on the web site.The dif- 12–13 February 2004 – ficult practical decisions that forecasters have to make Forecasts, public warnings and lessons learned – . 15 when faced with the threat of severe weather are dis- Planning of adaptive observations during the cussed on page 15 by members of the Turkish Weather Atlantic THORPEX Regional Campaign 2003 . 16 Service.The article on page 16 by Martin Leutbecher Two new cycles of the IFS: 26r3 and 28r1 . 26 and colleagues describes some experiments to assess COMPUTING the usefulness of making additional targeted observa- 25 years since the first operational forecast . 36 tions in areas that are considered likely to be sensitive for the subsequent forecast. Details of the improve- GENERAL ments made to the Integrated Forecasting System in 61st Council session on 13–14 December 2004 . 39 two recent cycles – Cycle 26r3 (operational on 7 ECMWF Calendar 2005. 40 October 2003) and Cycle 28r1 (operational on 9 MARS reaches one Petabyte. 40 March 2004) – are given by Jean-Noël Thépaut and ECMWF publications. 40 colleagues on page 26. Finally, on page 36 John New items on the ECMWF web site . 41 Hennessy looks back into the early history of the Index of past newsletter articles. 42 Centre and describes the problems involved in pro- Useful names and telephone numbers ducing the first operational forecasts 25 years ago. within ECMWF. 44 Peter White European Centre for Medium-Range Weather Forecasts Changes to the operational Shinfield Park, Reading, Berkshire RG2 9AX, UK weather prediction system Fax: . .+44 118 986 9450 Telephone: National . .0118 949 9000 28 September 2004 – Cycle 28r3 was introduced. This International . .+44 118 949 9000 included changes to the: N Physics – Revised numerics of the convection scheme ECMWF Web site . .http://www.ecmwf.int and the calling of the cloud scheme; use of the tangent linear and adjoint of vertical diffusion in the first mini- The ECMWF Newsletter is published quarterly and mization of 4D-Var; reduction of the radiation frequency contains articles about new developments and systems to one hour in the high-resolution forecasts; improved at ECMWF. Articles about uses and applications of numerics of surface tile coupling; post-processing of ECMWF forecasts are also welcome from authors total-column liquid water and ice. working elsewhere (especially those from Member Use of satellite data – RTTOV-8; minor revisions to States and Co-operating States). N ATOVS and AIRS usage; assimilation of MSG clear-sky The ECMWF Newsletter is not a peer-reviewed radiances and GOES-9 BUFR AMVs; assimilation of publication. SCIAMACHY ozone products from KNMI; correction Editor: Peter White of error in AMSU-B usage over land; activation of EARS Typesetting and Graphics:Rob Hine data. N Data assimilation – Blacklist SYNOP humidity data at Front Cover local night time; increased use of radiosonde humidity (using RS90 to -80°C, RS80 to -60°C, other sensors to New Caledonia and (inset) Tropical Cyclone Erica:– see articles on -40°C); proper cycling of the information from the wave pages 2 and 7. © Thierry Lefort. altimeter and the surface data analyses (first-guess fore- casts moved from 00 and 12 UTC to 06 and 18 UTC). 1 METEOROLOGY ECMWF Newsletter No. 102 – Winter 2004/05 N EPS – Gaussian sampling for extratropical singular vectors in the vicinity); orthonormalisation applied to the set of instead of selection and rotation; revision of initial-condi- all tropical singular vectors ). tion perturbations for tropical cyclones (initial condition 7 October 2004 – Monthly forecasts run operationally on perturbations extended to latitude belts 40°S and 40°N a weekly basis (Thursdays). from 25°S and 25°N, and tropical singular vectors are 18 October 2004 – The bias correction for AMSU-A and computed in the subspace orthogonal to the leading 25 AIRS satellites was harmonised. The bias correction for extratropical singular vectors); new algorithm to deter- AMSU-B and HIRS was simplified to use two predictors. mine optimisation regions based on predicted tropical 9 November 2004 – All four BC-project analyses used back- cyclone tracks from previous EPS run (the Caribbean ground fields generated from the latest operational 4D-Var remains an optimisation region if no tropical cyclone is analysis François Lalaurette Starting up medium-range forecasting for New Caledonia in the South-West Pacific Ocean – a not so boring tropical climate ew Caledonia is located 1500km east of Queens- land,Australia, between 18°S and 23°S. Its surface Nis 19,000km2.The main island or Grande Terre is 400km long and 50km wide, with a mountain range roughly parallel to the coasts that reaches 1000 to 1600m in altitude. t r Smaller islands are located to the north and to the south.The o f e 2 L lagoon is 8000km in area and as much as 65km wide. One y r r hundred kilometres to the east lie the Loyauté islands; these e i h T are flat, coral islands (Figure 1). © h p a r g o t o h P New Caledonia is located just north of the Tropic of Capricorn,but its climate is said to be ‘maritime tropical’.The year is divided roughly into two periods.The cool season lasts from May to October: extratropical lows form in the Tasman Sea and we get occasional cold fronts associated with short periods of westerlies.The average low temperature in July is 11.8°C in the western lowlands, so that morning tempera- tures of less than 10°C are frequent.The lowest temperature recorded in flat areas is 2.3°C on the Grande Terre,and 2.7°C on the Loyauté islands! The hot season lasts from November to April; it is also called the tropical-cyclone or rainy season. Since the elongated Grande Terre is roughly perpendicular to the prevailing easterlies, there is a considerable difference between the east coast,which is humid and green,and the west coast, which is much drier in the lee of the range (Figure 2). As New Caledonia is a French Territory, Météo-France New Caledonia is in charge of meteorology.The last decade has seen tremen- dous changes in the working environment of Météo-France forecasters in New Caledonia: the workstation Synergie Koumac was installed in 1994; two precipitation radars are now oper- Poindimie OUANAHAM ating, with a third due to be operating soon; and the forecaster is provided with several global models, including the French model ARPEGE TROPIQUE and the ECMWF T511 fields displayed on a 0.5 degree grid at six-hour intervals. Ensemble 4000 3000 products are also available from the workstation, the Intranet 2000 La Tontouta Fig. 1 (upper) A map of south-west Pacific Ocean. The oval shows 1000 the New Caledonian territory. The square is the domain for tropical 0 Noumea cyclone warnings. The diamond-shape is the domain for marine 0 50 100 km bulletins. (lower) Detail of New Caledonia and the nearby islands. 2 ECMWF Newsletter No. 102 – Winter 2004/05 METEOROLOGY BELEP POINGAM POUM POUEBO TIEBAGHI KOUMAC GOMEN HIENGHENE OULOUP TOUHO MOULI OUANAHAM TEMALA POINDIMIE VOH WE KONE PONERIHOUEN 3750 POUEMBOUT AOUPINIE 3500 HOUAILOU 3250 NEPOUI KOUAOUA 3000 MEA 2750 CANALA LA ROCHE 2500 BOURAIL THIO TADINE 2250 LA FOA 2000 NASSIRAH BORINDI 1750 LA OUINNE 1500 LA TONTOUTA 1250 1000 PAITA YATE 750 NOUMEA 500 0 P.AMEDEE 0 50 100 km MOUE VAO Fig. 2 Reconstruction of the annual precipitation for New Caledonia for the period 1991–2000 using the model Aurelhy (mm). page of the Predictability Section of Météo-France’s Central 80 a A Forecasting Office, and of course the ECMWF website. 70 B 60 What is predictable? C 50 D The current public bulletin issued by Météo-France in New % 40 Caledonia covers the period from Day 2 to Day 5 (Day 1 is 30 the day the bulletin is issued). 20 Over the Atlantic Ocean, Chessa and Lalaurette (2001) have shown that a probabilistic forecast of flow patterns (Ayrault 10 0 et al., 1995) beats climatology,even at Day 9. In 2002, the skill D+4 D+5 D+6 D+7 D+9 of the ECMWF model for the Southern Hemisphere matched the one for the Northern Hemisphere (Hollingsworth et al., 100 b 2003). For the domain covering Australia and New Zealand 90 A+B (including New Caledonia), the Australian Bureau of 80 C+D Meteorology reports that ECMWF produces useful guidance 70 out to at least seven days (François Lalaurette suggested at the 60 1 2004 ECMWF users’ meeting that 7 ⁄2 days was more appro- 50% priate). 40 As for the tropics, Kanamitsu (1985) pointed out that 30 errors grow very fast there during the first 24 hours compared 20 with those in the extratropics, and then grow more slowly. 10 0 He added that the choice of parameters to be verified in the D+4 D+5 D+6 D+7 D+8 D+9 tropics needed some discussion.