The Impacts of GPSRO Data Assimilation and Four Ices Microphysics Scheme on Simulation of heavy rainfall Events over Taiwan during 10-12 June 2012 Pay-Liam LIN, Y.-J. Chen, B.-Y. Lu, C.-K. WANG, C.-S. CHEN, Y.-L. CHEN C.-Y. Ke, T.-C. Chen Wang and Wei-Kuo TAO Department of Atmospheric Sciences National Central University
Outline Torrential Rainfall Occurred over Taiwan during 12 June 2012 Introduction Model Configuration Experiment Design Results & Discussion
Mesoscale Convective System 3
Introduction Heavy rainfall occurred over northern Taiwan, central Taiwan and southwestern Taiwan on June 12, 2012. Picture from: http://codweb.net/blog/article/760 410553942/
East Asia Study Case 2012/6/11 2000LST EC Analysis Mei-Yu Front L Taiwan Southwesterly Flow
10 11 12 Yangmei Max:123 mm/hr For 10 hours, 30 mm/hr Total:479.5 mm /10hr CWB:15 mm/hr (Heavy Rain) 350mm/24hr (Extremely Torrential Rain) 6
10 11 12 06/11 --- 1200UTC (2000LST) Con. Div. surface 850 mb 300 mb 7
Introduction- Analysis 2012/06/11 00Z 2012/06/11 06Z 2012/06/11 12Z Taiwan 2012/06/11 18Z 2012/06/12 00Z Mei-yu front near Taiwan SW wind prevails over Taiwan
JMA Analysis (850 hpa) 2012/6/12 00UTC Taiwan is over the wet Area (shading: T-Td<3 o C) Taiwan
JMA Analysis (300 hpa) 2012/6/12 00UTC the upper-level trough, which is favorable for the development of heavy rainfall Taiwan
Observation Summary This is a linear MCS along Mei-Yu front that bring more than 400mm of rain in the evening. Favorable environmental conditions for the the system east of short trough low-level convergence upper-level divergence It s difficult to forecast mesoscale systems evolution. 11
Model Configuration WRF V3.5.1 NCEP GFS MP: Goddard GCE scheme CU: Kain-Fritsch scheme (Domain 1 and 2) BL: YSU scheme SF: Unified Noah landsurface model 27/9/3 km
Experiment Design(1) COLD Without data assimilation CTRL Data assimilation(cycling) twice With GTS_GPS GTS_ONLY Cycling four times GTS_GPS Cycling four times GTS_ONLY GTS_GPS 06/10 12Z Goal: 1. Assimilate observational data to Improve rainfall prediction 2. Test GPS impact on rainfall distribution COLD CTRL 06/11 00Z 06/11 00Z 06Z 12Z Use WRF 3DVAR 18Z 06/11 00Z 06Z 12Z
Data Assimilation- data location (06/11 12Z) SYNOP SHIP SATEM BUOY AIREP METAR PILOT GPSREF SOUND
Results-radar reflectivities 2012/06/11 23Z LST COLD GTS_ONLY CTRL GTS_GPS
Accumulated precipitation COLD 06/12 00 06/13 00 LST CTRL GTS_GPS GTS_ONLY
Results- ETS score (ctl1112)ctrl (cold00)cold (fg1112)gts_gps (fo1112)gts_only
Experiment Design(2) CTRL Data assimilation(cycling) twice With GTS_GPS EC FNL GFS DA (GPS+GTS) Cycling four times Use WRF 3DVAR EC FNL GFS 1) 4ice 2) 3ice 1) 4ice 2) 3ice Goal: 1. Impact of DA Cycling Times 2. Test different boundary and initial condition (EC FNL GFS) 3.Test 4ice and 3ice scheme CTRL DA 06/10 12Z 18Z 06/11 00Z 06Z 12Z 06/11 00Z 06Z 12Z
Why do we need to have the 4-ICE scheme? Observation 3ICE-Hail 3ICE-Graupel Almost all microphysics schemes are 3-ICE (cloud ice, snow and graupel). Very few 3ICE schemes have the option to have hail processes (cloud ice, snow, graupel or hail) Both hail and/or graupel can occur in real weather events simultaneously, therefore a 4ICE scheme (cloud ice, snow, graupel and hail) is required for real time forecasts (especially for high-resolution prediction of severe local thunderstorms, midlatitude squall lines and tornadoes) Current and future global high-resolution cloud-resolving models need the ability to predict/simulate a variety of weather systems from weak to intense (i.e., tropical cyclones, thunderstorms) over the globe; this requires the use of a 4ICE scheme 19
Results-accumulated precipitation (EC) 06/12 00 06/13 00 LST CTRL- 3ice DA-3ice CTRL- 4ice DA-4ice mm
Results-max DBZ (EC) CTRL-3ice CTRL-4ice DB Z 2012/06/11 23Z LST DA- 3ice DA-4ice
Results-accumulated precipitation (FNL) 06/12 00 06/13 00 LST CTRL-3ice DA-3ice CTRL-4ice DA-4ice mm
Results-max DBZ (FNL) CTRL-3ice CTRL-4ice DB Z 2012/06/11 23Z LST DA-3ice DA-4ice
Results-accumulated precipitation (GFS) 06/12 00 06/13 00 LST mm CTRL-3ice DA-3ice CTRL-4ice DA-4ice
Results-max DBZ (GFS) CTRL-3ice CTRL-4ice DB Z 012/06/11 23Z LST DA-3ice DA-4ice
Discussion- I.C. B.C. Model starts from 06/11 00 UTC (08LST) 06/12 00 06/13 00 LST rainfall accumulation 3ice_3km_EC 3ice_3km_NCEP GFS Model initialized with EC reanalysis data show better rainfall simulation than the model results initialized with GFS forecast data.
Discussion- I.C. and B.C. Time: 2012/06/11 12 Z Field: RH Diff. (925hPa:left & 850hPa:right) EC - GFS FNL - GFS EC - GFS FNL - GFS
2015.06.11 12 Simulation Domain and initial time setup
Experiment Design GCM: FNL 0.5 vs ERA interim 0.75 Domain: different size and location on d02 Time: 6/10 08:00 vs 6/11 08:00 LST
2015.06.12 Accumulated Precipitation
Conclusion The frontal in the run with bigger domain on d02 can move faster than the run with smaller domain and land on time though the unideal frontal position makes an unideal precipitation pattern. Although the frontal in the run with earlier initial time lands late, the ideal structure of frontal can make an ideal pattern of precipitation.
2015.06.11 12 Simulation Microphysical Scheme
Experiment Design Domain: 9km 3km Landuse update Microphysics 4Ice JP WSM5 WSM6 Radiation Cumulus PBL Land-surface Surface-layer RRTM & Goddard previous Kain-Fritsch YSU Noah Monin-Obukhov Reference from MEFSEA
2015.06.12 Accumulated Precipitation
The experiment design for Mei-Yu case in Hybrid system: period1: 10 June 2012 06Z ~ 11 June 2012 06Z period2: 11 June 2012 06Z ~ 12 June 2012 06Z
Hybrid Day1 (20120610 06Z~0611 06Z)
Hybrid Day2 (20120611 06Z~0612 06Z)
Conclusion Initial conditions (ICs.) and boundary conditions (BCs.) are important for the model prediction. For 12 June 2012 case, EC reanalysis data are better ICs. and BCs. than GFS data resulting in better rainfall prediction. To improve the model run with GFS data as ICs. and BCs., data assimilation (WRF 3DVAR) has applied. With four cycles run, rainfall over northern Taiwan and southwestern Taiwan has improved as compared to COLD START run and CTRL run. Increasing DA times improves precipitation time lag in all simulations and precipitation pattern. With GPS data assimilation could be more realistic to simulate heavy rainfall. IC and BC input will result in different precipitation pattern and value over Taiwan, especially over northern Taiwan.