UM Model Precipitation for Indian Monsoon: Days to Season Ashis K. Mitra (With Input from Colleagues at NCMRWF & UKMO/HC) National Centre for Medium Range Weather Forecasting (NCMRWF) Ministry of Earth Sciences (MoES), Noida, UP, India ashis.mitra@nic.in NERC-MoES Project ‘SAPRISE CWC’ Meeting, Exeter, 24-26 June 2013
NCMRWF in SAPRISE Goal : Seamless Assessment from hours-to-season Carry out diagnostics from Observations, Globa/Regional UM model output to understanding processes contributing monsoon rainfall at various time scales (from hours to a season variability) Data : (i) Gridded observations including NWP Analysis and Reanalysis data (ii) Satellite estimates for proxy observed fields (iii) UM and other model output from hours to a season NCMRWF focus is on model output from initialised runs both in NWP and Coupled Seasonal runs. First few months of the decadal runs will also be analysed. Model Sensitivity Runs at NCMRWF: NCMRWF UM Global/Regional NWP setup (atmosphere only) NCMRWF UM coupled set up (coarse resolution now)
SAPRISE: Scientific Objectives • To investigate driving processes, variability, predictability and forced changes in South Asian precipitation on multiple time scales . (NCMRWF :Days to a Season for summer Monsoon) • A key focus will be on interactions with the Indian and remote ocean basins and on the local and remote interactions with the dynamic and radiative effects of aerosol.
Advanced Model for Monsoon Simulation/Prediction ‘South Asian Monsoon System’ is a major component t of ‘Earth Climate System’ Simulating the variability of Monsoon is a benchmark for model development Evaluation and Diagnostics of Monsoon is important for model development
Specific Objectives 1. Investigate processes responsible for present day mean, variability and change in South Asia precipitation and test the ability of state-of-the-art climate models to simulate this. Work Package – 1 2. Evaluate the skill of initialized experiments in predicting South Asia precipitation variability and investigate mechanisms for predictability. Work Package – 2 NCMRWF 1. Investigate changes in South Asia precipitation and its drivers and interactions in a changing climate. Work Package – 3 2. Provide a seamless assessment and syntheses of results to advance our understanding of variability, predictability and change in precipitation in South Asia . Work Package - 4
Scientific Plan: Years 1 & 2
Unified Model at NCMRWF NWP set-up up to 10 days Same Model for Global/Regional/Mesoscale & Coupled Model – Seamless System 4 km grid up to 36 hr forecast hourly update 12 km grid up to 48 hr forecast 25 km global grid up to 168 hr forecast Coupled N96L85O1L75
The Unified Model (UM) with NCMRWF Obs. a horizontal resolution of about 25km & 70 levels in the vertical and the associated Obs. Decoding 4D-VAR data assimilation Bufr Database system have been implemented. Real time forecasts are being generated daily since May 2012 Obstore Converter NCMRWF Obstore Converter With data downloaded from UKMO Forecasts with OPS (27.2) Observation NCMRWF data started from obstore Processing December 2012 4D-VAR (27.2) 4D-Variational Assimilation UM (7.9) N512L70 (25 km in Mid lat) Global Model UM (12 & 4 km) The NWP Setup Regional Model
UM Data Assimilation System (real-time) at NCMRWF- Observation Source & Format Observation Type Source Format Aircraft NCMRWF obstore Sonde NCMRWF obstore Satwind NCMRWF obstore Scatwind NCMRWF obstore GPSRO NCMRWF obstore ATOVS NCMRWF BUFR IASI NCMRWF BUFR AIRS NCMRWF BUFR SSMIS UKMO obstore SEVIRI Clear UKMO obstore GOES Clear UKMO obstore
UM Diagnostics/Verification during Monsoon Diagnose the sources and nature of the errors, and providing information that is useful to modellers for targeting model improvements. Numerical experimentation and sensitivity studies of selected rain events (e.g., monsoon onset, monsoon depressions, cyclones).
Verification of GFS, UKMO and NCUM models during Jun-Sep 2012 Root Mean Square Error (RMSE) of 850 hPa winds(m/s) against Radiosondes over India JUN 2012 JUL 2012 8 8 7 7 6 6 5 GFS 5 GFS RMSEV RMSEV 4 UKMO 4 UKMO 3 3 NCUM NCUM 2 2 1 1 0 0 1 2 3 4 5 1 2 3 4 5 FORECAST DAYS FORECAST DAYS AUG 2012 SEP 2012 8 8 7 7 6 6 5 GFS 5 GFS RMSEV RMSEV 4 UKMO 4 UKMO 3 3 NCUM NCUM 2 2 1 1 0 0 1 2 3 4 5 1 2 3 4 5 FORECAST DAYS FORECAST DAYS
Verification of UKMO, NCUM , GFS models during Monsoon 2012 Global Model forecast errors during monsoon over Indian region 1. U.K. Met Office (UKMO) Unified Model forecasts 2. NCMRWF version of the Unified Model (NCUM) over India, 3. Global Forecasting Systems (GFS T574L64) NCMRWF version
Verification against respective analyses
Verification against respective analyses
Verification against respective analyses
Verification against respective analyses
Verification against respective analyses
Verification against respective analyses
Verification against respective analyses
Verification against respective analyses
Monsoon Low Pressure Systems During JJAS 2012 ten low pressure systems formed and none of them intensified to form a monsoon depression. There were four low-pressure systems which lasted long and significantly impacted in terms of heavy rains over the land regions
Low Pressure systems during Monsoon 2012 VECTOR ERROR (Km) IN THE PRECDICTED TRACK OF THE FOUR SYSTEMS DURING JJAS 2012 700 600 T574 500 Error in (Km) UKMO 400 NCUM 300 200 100 0 Day-1 Day-2 Day-3 Day-4 Day-5 Days
Evaluation of rainfall forecasts • Model Forecast Daily Rainfall (cm/day) – 1 st June – 30 th Sept 2012 – T574,UKMO, and NCUM • Observed Daily Rainfall (cm/day) – 1 st June – 30 th Sept 2012 – IMD-NCMRWF [Merged Sat + Gauge] – 0.5° x 0.5° grid resolution. • Focus on Day-1, Day3 and Day5 forecasts • Statistical Verification – 0.5° x 0.5° grids; over Indian land region.
• In T574 the rainfall over central India is overestimated in Day-1 and underestimated in Day-5. • Rainfall northern plains is underestimated in Day-1 and is made up in Day-3 and Day-5. • UKMO and NCUM both show- – Increasing equatorial rain in the western Indian Ocean from Day-1 to Day-5 Increasing rainfall coverage over – Bay of Bengal from Day-1 to Day- 5
• RMSE gives the average magnitude of forecast errors • All three models show higher RMSE over same region of central and eastern India. • RMSE is least over dry regions of peninsula and NW India. • T574 shows higher errors than both UKMO and NCUM. • Over Northeastern India, all models and all forecasts indicate large errors.
Rain-No Rain • ETS tells us how well did the forecast “yes” events correspond to the observed “yes” events (accounting for hits due to chance)? • ETS ranges from -1/3 to 1 • 0 indicates no skill and • 1 meaning perfect score. • All three models show high skill along west coast, some skill over the central India extending east to west. • Poor/no skill in the NE India and eastern peninsula (Day-5) is common in all the forecasts
Rainfall Thresholds • Very low skill for rainfall >2 cm/day • Both UKMO and NCUM show higher skill compared to T574
UKMO Rainfall Forecast Verification (Six Seasons: 2007-2012)
The criteria chosen for daily anomalies computation based on IMD observations and daily climatology. Based on this anomalies dry and wet days during JJAS have been identified
2007
2008
2009
2010
2011
All WET DRY
Recent Very Heavy Rainfall In Uttarakhand and HP: 16-18 June 2013
UM Coupled Model setup at NCMRWF HadGEM3AO [N96L85O1L75] (UM Atmosphere : 1.875 x 1.125 horizontal resolution) Stand alone model/software implementation / test Global NEMO Ocean Model - EU Community Ocean Model CICE - Sea Ice Model , Las Almos lab. USA OASIS Coupler UKMO coupled model Installed: UM 7.6 , NEMO, CICE with OASIS Successful test run of UKMO coupled model including Portability tests 14 years Hindcast 1996 to 2009 completed , JJA period analysed Start date; 09 May , six months integration Observed data: CMAP rainfall , ERA Interim for other parameters Comparisons with NCEP CFSV2 hind-cast data for same period JULES Land surface data assimilation module to be implemented at NCMRWF
UM Coupled Model Set-up
UM Coupled Model Set-up
UM Coupled Model Set-up
WOCE 2009 1 x 1 1955-2006
WOCE 2009 1 x 1 1955-2006
WOCE 2009 1 x 1 1955-2006
Recommend
More recommend
