MODELS DEVELOMENT IN BMKG Presnted by suratnobmkg go
MODELS DEVELOMENT IN BMKG Presnted by suratno@bmkg. go. id R & D Research Center METEOROLOGY INDONESIA
OULINE q R & D ORGANIZATION AND RESPONSIBILITIES q CURRENT WEATHER PREDICTION q. MODELLING ACTIVITIES Ø SEA STATE ANLYSIS AND FORECASTING Ø Ø WRF-EMS VALIDATION CCAM APLICATION FOR MOS DEVELOPMENT
R & D ORGANIZATION AND RESPONSIBILTIES BMKG DIRECTOR GENERAL R & D Director (Dr. Masturyono) Meterological Division Climatological Divison Geophisical Divison R & D related wetaher and ocean analysis & forecastng R & D related climate analysis and prediction, and air qualty R & D related seismisity, earthquake and Tsunami Administration
CURRENT WEATHER PREDICTION Other Center’s NWP Products TLAPS, Arpege and etc CCAM 27 km outputs run in BMKG (GFS forcing)
MODELLING ACTIVITIES q SEA STATE ANALYSIS AND FORECASTING Existing : WINDWAVES-05 Type : LAM, Deep water, 2 nd generation Boundaries : land =0, no energy trasnfer for open ocean Usage 1 : regular basis wave analisis & forecasting since 2005 & high wave warning since Jun, 2007 Input: GFS 10 wind 0. 5 deg. in resolution Output : 6 hourly forecast up to 168 hour : weekly forecast Usage 2 : Climate Studies, forcing NCEP FNL 1 deg. Input : NCEP FNl 1. deg Output : monthly and seasonal - Average Hs - Average Highest Hs, and highest Hs - Hig waves (Hs > 2 m) frequency
HIGH WAVE VULNERABILITY ASIAN MONSOON DECEMBER JANUARY 0 FEBRUARY 5 10 20 30 40 50 60 (percent) 70 80 90 >
CLIMATE STUDY HIGH WAVE VULNERABILITY TRANSITION TO AUSTRALIAN MONSOON MARCH APRIL MAY 0 5 10 20 30 40 50 60 (percent) 70 80 90 >
HIGH WAVE VULNERABILITY AUSTRALIAN MONSOON JUNE JULY AUGUST 0 5 10 20 30 40 50 60 (percent) 70 80 90 >
CLIMATE STUDY HIGH WAVE VULNERABILITY TRANSITION TO ASIAN MONSOON SEPTEMBER OCTOBER 0 NOVEMBER 5 10 20 30 40 50 60 (percent) 70 80 90 >
MODELLING ACTIVITIES q. SEA STATE ANALYSIS AND FORECASTING CURRENT DEVELOPMENT 2012 – 2014 q. Wave. Watch III Global & Regional domain operate once a day using GFS 0. 5 deg forcing q. MRI III under study q WRF 10 m wind plan to be or regional after validation
MODELLING ACTIVITIES q WRF- EMS (2012 -20014) VALIDATION METODE Boudary and initial condition : GFS 0. 5 Step 1 # : Various convection schemes test for higher resolusion regional Indonesian domain Step 1 # : Each schemes evaluated using rason/ rawind sonde, pibals and ground observation Step 3 # The best Scheme will be selected for operational testing Three schemes has been tested but eavaluation not yet finished
MODELLING ACTIVITIES q CCAM APLICATION FOR MOS DEVELOPMENT (2011 -2014) (Joint reseacrh BMKG & Surabaya Institude Tectonology TARGET AREA : Jabodetabek (Jakarta and serounding) WHY MOS ? MOS Post – Processing NWP Reduce NWP bias Aplicable for prediction of unpredicted vaiable by NWP such as visibilty, thunderstorm
CCAM BIAS CCAM (TMAX, TMIN CCAM) Outputs VS observation CCAM (RH) VS observation
Model Output Statistics (MOS) NWP outputs (X) Observatation (Y) = variable predictant at time t = variables predictor at time t t
Model Output Statistics (MOS) NWP Output High dimensional Curse of Dimentionality BIAS Reduce the dimension of predictor variables Persuit Projection NWP Dimension Reduction Persuit Projection regresion MOS 15
Model Output Statistics (MOS) CCAM AREA OF INTERST
Model Output Statistics (MOS) AREA OF INTEREST
Determining NWP Domain Grid Figure 1. The Position of Observation Station on 3 x 3 Grid
Model Output Statistics (MOS) Daily NWP Output • In Sample (01/01/2009 – 31/10/2010) • Out Sample (01/10/2010 – 31/12/2010) Respon Variables • Tmax-obs • Tmin-obs • Daily average RH-obs Predictor Variables • Tmax-NWP • Tmin-NWP • Daily average RH-NWP Locations • • Maritim Tanjung Priok Cengkareng Curug Dermaga
Results of prediction using out of sample data Fig. 1 Tmax, Tmin and RH Mos prediction for Tanjung Priok versus Observation Fig. 2 Tmax, Tmin and RH Mos prediction for Curug versus Observation
Results of prediction using out of sample data Fig. 3 Tmax, Tmin and RH Mos predictions for Cengkareng versus Observation Fig. 4 Tmax, Tmin and RH Mos prediction for Curug versus Observation
Results of prediction using out of sample data Fig. 3 Tmax, Tmin and RH Mos predictions for Cengkareng versus Observation Fig. 4 Tmax, Tmin and RH Mos prediction for Curug versus Observation
PERCENTAGE IMPROVAL
THANK YOU
- Slides: 24