The ECMWF Extended range forecasts

Transcription

1 The ECMWF Extended range forecasts ECMWF, Reading, U.K. Slide 1 TC January 2014 Slide 1

2 The operational forecasting system l High resolution forecast: twice per day 16 km 91-level, to 10 days ahead l Ensemble Prediction System (ENS): twice daily 51 members, 30/60 km 62-level, to 15 days ahead l Extended range forecasts /ENS extension: twice a week (Mon/Thursdays) 51 members, 32/64 km 91 levels, to 1 month ahead l Long range forecasts: once a month (coupled to ocean model) 51 members, ~80 km 91 levels, to 7 months ahead Slide 2 TC January 2014 Slide 2

3 Extended range forecast /ENS extension Ens. m. range twice daily TL639 (32Km) ENS Extended range Mon/Thu TL319 (64Km) Init. cond. Day 10 Day 15 Day 32 Atmosphere Initial uncertainties T42L91 SVs+ T399L137 EDA perturbations Model uncertainties Stochastic physics (SPPT and SKEB schemes). The central analysis is the TL1279L137 4DVAR coupled to wave model (WAM) every time step Ocean Slide 3 NEMO (about 1 degree resolution) coupled to IFS every 3 hours. Ocean initial conditions provided by 5-member NEMOVAR analysis

4 Bridging the gap between seasonal forecasting and NWP l Extended-range weather forecasting: Beyond 10 days and up to 30 days description of weather parameters, usually averaged over a period of 5-7 days and expressed as a departure from climate values for that period. l A particularly difficult time range: In fact at this time range is generally too long for the atmosphere to keep a memory of its initial conditions, and too short for the ocean variability to have an impact on the atmospheric circulation. Slide 4

5 The ECMWF extended forecasts consists of 2 elements: Real time forecasts A set of re-forecasts covering the most recent 20 years period the same configuration of the real time forecasts 5-member ensemble integrated at the same day and same month as the real-time time forecast It runs once every week Used to estimate the model drift Slide 5

6 ECMWF Monthly Forecasting System MODEL BIAS: 2m Temperature Forecast start reference is 05/03/ ensemble size = 5 Model Bias: WEEK1-4 <-8K > 8K 16W 12W 8W WEEK1: DAY 5 TO 11 WEEK2: DAY 12 TO 18 4W 2E 4E 6E 8E 12E 16E 16W 12W 8W 4W 2E 4E 6E 8E 12E 16E 8N 8N 7N 7N 6N 6N 5N 5N 4N 4N 3N 3N 2N 2N 1N 1N 1S 1S 2S corrected. 2S 3S 3S 4S 4S 5S 5S 6S 6S 7S 7S 8S Day5-11 8S 16W 12W 8W 4W 2E 4E 6E 8E 12E 16E 16W 12W 8W 4W 2E 4E 6E 8E 12E 16E 8N 8N 7N biases 7N 6N 6N 5N 5N 4N 16W 12W 8W 4W 2E 4E 6E 8E 12E 16E 4N 3N 8N 3N 8N 2N 7N 2N 7N 1N 6N 1N 6N 5N 5N 1S 4N 1S 4N 2S 3N 2S 3N 3S post-processing. 2N 3S 2N 4S 1N 4S 1N 5S 5S 6S 1S 6S 1S 7S 2S 7S 2S 8S 3S 8S 3S 4S 4S 16W 12W 8W 4W 2E 4E 6E 8E 12E 16E 5S 5S 6S 6S 7S 7S 8S 8S 8N 8N 7N 7N 6N 6N 5N 5N 4N 4N 3N 3N 2N 2N 1N 1N 1S 1S 2S 2S 3S 3S 4S 4S 5S 5S 6S 6S 7S 7S 8S Day S 16W 12W 8W 4W 2E 4E 6E 8E 12E 16E 16W 12W 8W 4W 2E 4E 6E 8E 12E 16E 8N 8N 7N 7N 6N 6N 5N 5N 4N 16W 12W 8W 4W 2E 4E 6E 8E 12E 16E 4N 3N 8N 3N 8N 2N 7N 2N 7N 1N 6N 1N 6N 5N 5N 1S 4N 1S 4N 2S 3N 2S 3N 3S 2N 3S Slide 6 2N 4S 1N 4S 1N 5S 5S 6S 1S 6S 1S 7S 2S 7S 2S 8S 3S 8S 3S 4S 4S 16W 12W 8W 4W 2E 4E 6E 8E 12E 16E 5S 5S 6S 6S 7S 7S 8S 8S After 10 days of forecast, model biases cannot be ignored, and the real time forecasts need to be biased The set of re-forecast is used to estimate the model WEEK3: DAY 19 TO 25 WEEK4: DAY 26 TO 32 The bias is removed from the real time forecast during the Day19-25 Day W 12W 8W 4W 2E 4E 6E 8E 12E 16E 16W 12W 8W 4W 2E 4E 6E 8E 12E 16E

7 Weekly mean anomalies Slide 7 Footer-text Slide 7

8 Probabilities for weekly mean anomalies: Slide 8 Footer-text Slide 8

9 Weekly mean multiparameter outlook: Slide 9 Footer-text Slide 9

10 Bridging the gap between seasonal forecasting and NWP Sources of predictability for the extended forecasts : - Land Surface conditions: Snow cover, Soil Moisture - Ocean conditions: Sea surface temperature, Sea ice - Stratospheric Initial conditions - The Madden-Julian oscillation - Atmospheric dynamical processes (Rossby wave propagations, weather regimes ) Slide 10

11 Cold spell over Europe Nov-Dec /11/ /11-2/ / 12 / analysis 5-11d d Slide 11 Footer-text Slide 11

12 Cold Weather over Europe: SSW Index Forecast starting on : 22/11/ /11/2012 Slide 12 12

13 Cold Weather over Europe SSW Index - Forecast starting on 22/11/2012 Strong SSW Weak SSW Slide 13 13

14 Impact of soil moisture: Slide 14 Koster et al, GRL

15 The Madden Julian Oscillation (MJO) MJO life cycle (From NASA) Slide 15 From bmrc/clf

16 MJO impact on European weather: 30 The MJO impact is the strongest about 10 days after 20 the MJO is in the phase with: NAO- NAO suppressed convection over Indian Ocean - enhanced convection over Western Pacific are conducive to negative NAO Cassou (2008) Lin et al (2008) Slide 16 Conv. Over Indian Ocean +10 days Conv. Over Western Pacific+10 days

17 Analysis 10 best MJO forecasts Slide 17 Footer-text Slide worse MJO forecasts

18 MJO Teleconnections Evolution of NAO skill scores day NAO index is computed as projection onto a reference pattern Slide 18 Footer-text Slide 18

19 MJO forecast: Slide 19 Footer-text Slide 19

20 Probabilistic skill scores NDJFMA Reliability Diagram Probability of 2-m temperature in the upper tercile Day N. Extratropics 1 EUROPE obs frequency obs frequency forecast probability MJO in IC 0.1 NO MJO in IC Slide forecast probability

21 Weekly mean Accumulated Cyclone Energy (ACE) Slide 21 Footer-text Slide 21

22 ROC for 2mt in the upper tercile since Oct 2004 Slide 22 Footer-text Slide 22

23 Evolution of skill scores based on the re-forecasts RPSS Probability of 2mt in upper tercile NDJFM All the re-forecasts produced since 2002 have the period in common. RPSS scores have been computed for all the reforecasts produced between April of a given year and March of the following year and covering the period (once a week, 5-member ensemble). Slide 23 Footer-text Slide 23

24 New re-forecast twice a week 11 members Impact on verification T850- Upper terciles Week Slide 24 Footer-text Slide 24

25 Conclusion l SSTs, Soil moisture, stratospheric initial conditions and MJO are source of predictability at the intra-seasonal time scale. In particular the MJO has a significant impact on the forecast skill scores beyond day 20. Model improvements, particularly in simulating the MJO activity are likely to be beneficial for monthly forecasting. l The monthly forecasting system produces forecasts for days that are generally better than climatology and persistence of day Beyond day 20, the monthly forecast is marginally skilful. For some applications and some regions, these forecasts could however be of some interest. Slide 25

26 Extended range ensemble system ENS includes 51 forecasts with resolution: TL639L91 from day 0 to 10 and TL319L91 from day 10 to 15. Atmosphere Initial uncertainties T42L91 SVs+ T399L137 EDA perturbations Model uncertainties Stochastic physics (SPPT and SKEB schemes). The central analysis is the TL1279L137 4DVAR. coupled to wave model (WAM) every time step Ocean NEMO (about 1 degree resolution) coupled to IFS every 3 hours. Ocean initial conditions provided by 5-member Slide 26 NEMOVAR analysis Footer-text Slide 26

27 The ECMWF monthly forecasting system l Atmospheric initial conditions: ECMWF operational analysis l Oceanic initial conditions: Accelerated ocean analysis l Perturbations: - Atmosphere: Singular vectors + stochastic physics - Ocean: Wind stress perturbations during the data assimilation Slide 27

28 Skill of the ECMWF Monthly Forecasting System TO ROC score: 2-meter temperature in the upper tercile Day 5-11 Day TO < > 0.9 < > W 12W 10W 8W 6W 4W 2W 2E 6E 10E 12E 14E 16E 14W 4E 8E 16W 14W 12W 10W 8W 6W 4W 2W 2E 4E 6E 8E 10E 12E 14E 16E 8N 8N 7N 7N 6N 6N 5N 5N 4N 4N 3N 3N 8N 8N 7N 7N 6N 6N 5N 5N 4N 4N 3N 3N 2N 2N 2N 2N 1N 1N 1N 1N 1S 1S 1S 1S 2S 2S 2S 2S 3S 3S 3S 3S 4S 4S 4S 4S 5S 5S 5S 5S 6S 6S 6S 6S 7S 7S 7S 7S 8S 8S 8S 8S 16W 14W 12W 10W 8W 6W 4W 2W 2E 4E 6E 8E 10E 12E 14E 16E 16W 14W TO W 10W 8W 6W 4W 2W 2E 4E 6E 8E 10E 12E 14E 16E Day Day TO < > 0.9 < > W 12W 10W 8W 6W 4W 2W 2E 6E 10E 12E 14E 16E 14W 4E 8E 16W 14W 12W 10W 8W 6W 4W 2W 2E 4E 6E 8E 10E 12E 14E 16E 8N 8N 7N 7N 6N 6N 5N 5N 8N 8N 7N 7N 6N 6N 5N 5N 4N 4N 4N 4N 3N 3N 3N 3N 2N 2N 2N 2N 1N 1N 1S 1S 2S 2S 3S 3S 1N 1N Slide 28 1S 1S 2S 2S 3S 3S 4S 4S 4S 4S 5S 5S 5S 5S 6S 6S 6S 6S 7S 7S 7S 7S 8S 8S 8S 8S 16W 14W 12W 10W 8W 6W 4W 2W 2E 4E 6E 8E 10E 12E 14E 16E 16W 14W 12W 10W 8W 6W 4W 2W 2E 4E 6E 8E 10E 12E 14E 16E

29 Skill of the ECMWF Monthly Forecasting System ROC scores over the Northern extratropics metre temperature ROC score = Mean sea-level pressure ROC score = Precipitation ROC score = hit rate 0.5 hit rate 0.5 hit rate false alarm rate Slide false alarm rate false alarm rate Day 5-11 Day12-18 Day19-25 Day 26-32

30 Monthly Forecast: Northern extratropics ROC score: 2-metre temperature in the upper tercile Monthly Forecast Persistence of day 5-11 Monthly Forecast Persistence of day 5-18 Day Day Slide 30 FD/RD December 2013 Slide 30

31 Madden Julian Oscillation 40R1 40R3 Slide 31 Improvement due to revised organised convective detrainment term Footer-text and the revised convective momentum Slide 31 transport.

32 Slide 32 Footer-text Slide 32

33 Slide 33 Footer-text Slide 33

34 Probabilities (temperature) Slide 34 Footer-text Slide 34

35 Slide 35 Footer-text Slide 35

36 Tropical storm density Slide 36 Footer-text Slide 36