4th THORPEX workshop 31 Oct. 2012, Kunming, China A review on recent progresses of THORPEX activities in JMA Masaomi NAKAMURA Typhoon Research Department Meteorological Research Institute / JMA
Contents 1. Tropical cyclone ensemble forecast product using TIGGE CXML data 2. Severe weather potential forecast product using TIGGE data 3. Other THORPEX Related Research
NWP-TCEF Website The Meteorological Research Institute of the Japan Meteorological Agency (MRI/JMA) created a website for the NW Pacific Tropical Cyclone Ensemble Forecast Project (NWP-TCEFP/WWRP/WMO). URL: http://tparc.mri-jma.go.jp/cyclone/ Main Page Accessible by ESCAP/WMO Typhoon Committee members and Researchers worldwide via a password-input
List of available products - TC track - Deterministic forecast Ensemble forecast Strike Probability Point Strike Probability
Deterministic TC track forecast Track forecasts from ECMWF, JMA, MSC and NCEP are available. Blue: 72-96h forecast Red: 0-24h forecast Green: 24-48h forecast Purple: 48-72h forecast Best track Six-hourly plotted
Ensemble TC track forecast -MCGE- Track forecasts from CMA (ensemble size is 15), ECMWF (51), JMA Typhoon EPS (11), JMA One-week EPS (51), KMA (17), MSC (21), NCEP (21), STI (9) and UKMO (24) are available. Ensemble TC track forecast by all ensemble member of all centers. (Ensemble TC track forecast by each center is also available)
Ensemble TC track forecast -Each Center-
Strike probability map Probability that the center of a storm will pass within 120 km of a location during 96 hours is shown. Contour levels shown are 5-20% (green), 20-40% (yellow), 40-60% (orange), 60-80% (red) and 80-100% (purple).
Select the city and cyclone name on the website
Point strike probability map Time series of strike probability at a selected location (city). Y-axis on the left shows the strike probability at city ILAGAN, and the Y-axis on the right shows the distance between ILGAN and a TC center. Time
Severe weather potential forecast Website http://tparc.mri-jma.go.jp/tigge/tigge_swfdp.html MCGE stands for Multi Center Grand Ensemble
Definition of the heavy precipitation The heavy precipitation is defined as precipitation above 90 th, 95 th or 99 th percentile of the climatology. The rate of ensemble members that predict heavy precipitation is plotted The climatology is the model climatology, not the climatology in the real atmosphere, being created for each NWP center using the TIGGE dataset. Note that the users can change the threshold of the percentile (90 th, 95 th, or 99 th ) on the website.
Products on severe weather potential -heavy precipitation- MCGE stands for Multi Center Grand Ensemble Case for Thailand flood in 2011 If If the 4 EPSs (ECMWF, JMA, NCEP and UKMO EPSs) predict heavy precipitation simultaneously, the area is is plotted in red.
Comparison with observations Observation: Precipitation (mm/day) based on SYNOP Reports 10/02 After Harada and Adachi (CPD/JMA) Potential of heavy precipitation was predicted with a lead time of 3 to 4 days
09UTC 29 to 09UTC 30
Work on MRI/JMA TIGGE WEB page Sample scripts for potential TIGGE users uploaded http://tparc.mri-jma.go.jp/tigge/tigge_sample.tar.gz The above file contains the following files: 1. a script to download the TIGGE data, 2. a GrADS script to make a plot of the TIGGE data, (e.g. stamp map, spaghetti map, and probability map) 3.sample TIGGE data (GRIB2), 4. sample plots.
Sample scripts for potential TIGGE users (sample plots)
Probabilistic verification Ranked Probability Skill Score
forecast Obs. J: Number of rank in case of ten members
Forecast verification for the Southern Hemisphere
Verification of severe weather potential forecast product Reliability diagram (DJF) more reliable over-confident over-confident SCE s over-confident forecast improved in MCGE. But samples with high forecast probability decreases in MCGE. over-confident over-confident
New areas in early warning products (polar regions) Requests from the WWRP Polar Prediction Project
Click RAII: Southeast Asia
Click TIGGE Products
Select from the two options
Research on tropical cyclones (2010- ) Study on -the structure and the growth mechanisms of ensemble initial perturbations around Typhoon Sinlaku (2008) using the TIGGE data from the ECMWF, NCEP and JMA EPS: Yamaguchi and Majumdar (Mon.Wea.Rev. 2010). -the basic properties of singular vectors in the vicinity of tropical cyclones using a barotropic model: Yamaguchi et al. (J.Atmos.Sci. 2011) -a verification on tropical cyclone track prediction in the western North Pacific using the TIGGE data.: Yamaguchi et al. (Quart.J.Roy.Meteor.Soc. 2012)
Research on predictability (2010- ) -the predictabilities of extreme Euro-Russian blocking that caused a strong heat wave over Eastern Europe and Western Russian in early August of 2010, using operational mediumrange ensemble forecasts: CMC, ECMWF, JMA, NCEP, and UKMO: Matsueda (Geophys.Res.Lett. 2011) -the forecast performance of operational medium-range ensemble forecasts, regarding the Madden-Julian Oscillation (MJO) for the period 2008-2010: Matsueda and Endo (Geophys.Res.Lett., 2011)
- tropical cyclone track forecasts using JMA model with ECMWF and JMA initial conditions: Yamaguchi et al. (Geophys.Res.Lett., 2012) - a predictability on tropical cyclonegenesis using the TIGGE date set, and verified the genesis of Typhoon Guchol (2012) using the ensembles of JMA, NCEP and UKMO: Nakano et al. (JAMSTEC) Ensemble initial perturbations and their growth for tropical cyclone forecasts Tropical cyclone track forecasts using JMA model with ECMWF and JMA initial conditions
Ensemble initial perturbations and their growth for tropical cyclone forecasts Yamaguchi & Majumdar (2010)
Some contradictions of ensemble spread among EPSs ECMWF (50 members) NCEP (20 members) Sinlaku initiated at 12UTC 10 Sep. 2008 Dolphin initiated at 00UTC 13 Dec. 2008 Black line: Best track Grey lines: Ensemble member Japan Philippines Taiwan
Vertical profile of perturbation Kinetic Energy Perturbation kinetic energy at each vertical level is averaged over the all ensemble members over a 2000 km x 2000 km domain centered on Sinlaku JMA ECMWF NCEP
Decomposition of flows in the vicinity of TCs Total flow Spatial Low-pass filter Cutoff wavelength=1200km Background flows associated with synoptic features Steering vector TC circulation itself Total flow minus Background flow Axisymmetric circulation Asymmetric circulation Asymmetric propagation vector L H
ECMWF NCEP T+48h T+12h T+0h Spread with time Does not spread with time Steering vector Asymmetric propagation vector
The results indicate; 1. Perturbation structure and amplitudes are quite different among the NWP centers 2. Those differences cause the different modification of TC advection flows 3. Baroclinic energy conversion within a vortex leads to the modification of the advection flows 4. Differences of the ensemble spread of tracks among NWP centers are attributed to the growth of the perturbation and the initial amplitude Though the ECMWF initial perturbation amplitudes are small, the growth of the perturbations helps to obtain an appropriately large ensemble spread of tracks. Meanwhile, the relatively large amplitudes of initial perturbations seem to play a role in obtaining the ensemble spread of tracks in NCEP.
Tropical cyclone track forecasts using JMA model with ECMWF and JMA initial conditions From Yamaguchi et al.(2012)
Large improvement with EC initial. Is initial TC position important? Black: Obs. Blue:GSM+JMA initial Red: GSM+ECMWF initial Green: ECMWF No improvement. Model defect? EC and all exp. show large track error.
Research using YOTC Replacing the original initial condition of JMA/GSM with the ECMWF analysis reduces the TC track prediction errors by 5 %, 11 %, 9 %, 11 % and 15 % at 1 to 5 days, respectively (Yamaguchi et al. 2012, GRL) Position error (km) JMA JMA s model + EC s initial 2009.07.22-2009.11.30 16 TCs in the west Pacific EC Forecast hour Black line: JMA s model + JMA s initial condition Red line: JMA s model + EC s initial condition Green line: EC s model + EC s initial condition
It would be of great importance to identify the cause of these events and modify the NWP systems including the EPSs for better probabilistic forecasts. Typhoon Track Prediction by 9 EPSs participating in TIGGE Typhoon Megi initiated at 1200 UTC 25 th Oct. 2010 Typhoon Conson initiated at 1200 UTC 12 th Jul. 2010 Observed track
Plans in future
Feasibility study on tropical cyclogenesis using TIGGE data Purpose of the study To demonstate the skill of tropical cyclonegesis prediction on medium to intraseasonal timescales. Method MRI/JMA plans to explore the predictability of tropical cyclogenesis using the TIGGE grib2 data. For this purpose, we will use a vortex tracker developed by Dr. Frederic Vitart of the ECMWF. TCs over the western North Pacific during the 2009 and 2010 seasons will first be investigated, following a previous study by Tsai et al. (2012). Outreach The verification results will be available on the web site for a WWRP-RDP project, North Western Pacific Tropical Cyclone Typhoon Ensemble Forecast (NWP-TCTEF) Project.
Tropical Cyclogenesis using TIGGE Initial time: 12 UTC 7 Sep 2010 Valid time: 12 UTC 13 Sep 2010 (Time of the genesis of super typhoon MEGI) ECMWF (6 day prediction) JMA (6 day prediction) Red points: tropical cyclone predicted in the model Black point: location of the genesis of MEGI Vortex tracker developed by Dr. Vitart (ECMWF) is used. This algorithm uses vorticity@850hpa, the presence of warm core, the presence of a maximum of thickness, etc.
Tropical Cyclogenesis using TIGGE Initial time: 12 UTC 7 Sep 2010 Valid time: 12 UTC 13 Sep 2010 (Time of the genesis of super typhoon MEGI) ECMWF (6 day prediction) JMA (6 day prediction) Red points: tropical cyclone predicted in the model Black point: location of the genesis of MEGI More TCs are detected for JMA when the criterion used to identify TC vortices are relaxed.
JMA operational EPS At present, 11, 51, and 50 initial conditions are integrated by using a low-resolution version of the JMA global NWP model for producing an ensemble of 132-hour forecasts in the Typhoon EPS, 9-day forecasts in the One-week EPS, and 17/34-day forecasts in the One-month EPS. to assess uncertainties of the forecast targeted on specified phenomena. Week1 Week2 Week3 Week4 Typhoon EPS (TL319L60) up to 5.5 days One-week EPS (TL319L60) up to 9 days One-month EPS (TL159L60) up to 17 days One-month EPS (TL159L60) up to 34 days Hindcast 46
Plan for integrating three EPSs Unifying specifications of these EPSs To increase the resolution of the medium-range EPS model from TL319L60 to TL479L100. To conduct the One-week EPS from once a day to twice a day although the forecast ensemble size is reduced by about half. To increase the ensemble size from 11 to 25 in the Typhoon EPS. Introducing hindcast (reforecast) system to the integrated EPS Week1 Week2 Week3 Week4 FY 2013 Typhoon EPS (TL479L100) up to 5.5 days One-week EPS (TL479L100) up to 11 days One-month EPS (TL319L100) up to 17 days One-month EPS (TL319L100) up to 34 days Hindcast The part is highlighted in red to represent an upgrade point. Future Plan Integrated EPS (TL479L100) up to 17 days Hindcast Integrated EPS (TL319L100) from 17 days up to 34 days Hindcast 47
Thank you for your attention!
What controls the ensemble spread of tracks? 1.Methods: Different methods of creating initial perturbations may control it, resulting in different growth of the perturbations. 2.Amplitudes: Initial amplitudes of the perturbations may affect the size of the ensemble spread, especially in the early forecast stage. ECMWF NCEP JMA Method SV method Ensemble SV method Transform Amplitudes are determined in each NWP center in a statistical way
Steering and asymmetric propagation vectors ECMWF Steering vector NCEP Asymmetric propagation vector Black: Non-perturbed (Control) member Green: Perturbed members