suryachandra a rao
play

Suryachandra A. Rao Project Director Monsoon Mission and High - PowerPoint PPT Presentation

Nov 21, 2023 •317 likes •551 views

Large Scale DNS (Expectations) Suryachandra A. Rao Project Director Monsoon Mission and High Performance Computing Indian Institute of Tropical Meteorology, Pune Workshop on Cloud Dynamics, Micro physics, and Small-Scale Simulatjon

  1. Large Scale DNS (Expectations) Suryachandra A. Rao Project Director Monsoon Mission and High Performance Computing Indian Institute of Tropical Meteorology, Pune Workshop on Cloud Dynamics, Micro physics, and Small-Scale Simulatjon 13-16 August, 2018

  2. Outline • Status of the state-of-the-art predictjon System • Model limitatjons • How DNS/LES can help and what is needed? 2

  3. Potential Predictability VS Actual Prediction Skill of ISMR

  5. The Monsoon Mission Aim: To Improve predictions Skill of South Asian Monsoon  Seasonal and Extended range predictions  Short and Medium range (up to two weeks) prediction • The Mission’s goal is to build a working partnership between the Academic R & D Organizatjons and the Operatjonal Agency to leapfrog in improving monsoon forecast skill (R2O and O2R approach). • Requirement :All research work must be on the Operatjonal Modeling Framework!

  6. Developmental Actjvitjes under Monsoon Mission • High Resolutjon Seasonal/Extended/Short Range predictjon • New LSM development • Modifjed Zhao Carr cloud microphysics/ WSM6 • New Ocean model (MoM5) incorporatjon in CFS • Revised SAS • Super Parametrizatjon-CFS • Multj Cloud multj model stochastjc parametrizatjon in CFS • LETKF coupled data assimilatjon

  7. Improvements in monsoon characteristjcs due to developmental actjvitjes (Parametrizatjon schemes, LSM, Ocean and resolutjon) • Decreased dry bias over Indian Land mass • Decreased cold tropospheric bias • Decreased SST cold bias in tropics • Improved representatjon of snow cover thickness and tjme of meltjng • Improved ENSO characteristjcs and IOD characteristjcs. • Improved teleconnectjons • Betuer representatjon of extratropical and tropical interactjons

  8. IITM CFS Model: Seasonal Prediction Atmospheric Model ATMOSPHERE INITIAL GFS CONDITIONS FROM GSI (NCMRWF) T382 L64 levels Land Model Ice Model NOAH COUPLER Ocean Model OCEAN INITIAL CONDITIONS MOMv4 FROM GODAS (INCOIS/IITM) global 1/2 o x1/2 o (1/4 o in tropics) 40 levels Initjal conditjons for Hindcast runs are obtained from CFSR (Original model is adopted from NCEP)

  9. Potentjal Predictability VS Actual Predictjon Skill of ISMR Present Day Model Potentjal Skill vs Actual Skill 0.7 vs 0.64

  10. Monsoon Mission Model Performance (Prediction Skill as well as interannual variance) is better than other models for Indian Monsoon.

  11. Subdivision Wise Statjstjcs ECMF MME JJAS IITM MME

  12. High Resolutjon global 12.5 km model gives betuer skill (The skill of GFS T574 with 3 day lead is now extended to 5 days with T1534 ~12.5 km global GFS

  13. How Forecast Errors evolve?  Combined efgects of inadequate and incomplete observatjons  Data Assimilatjon methods  Ensemble Predictjon  More relevant and accurate observatjons  Model defjciencies  Numerical methods  Approximatjons made in formulatjon of the model equatjons  Spatjal Resolutjon  Sub-grid Scale Parametrizatjon 13

  14. Systematic Model Biases in CMIP5 models Rainfall bias (Sperber et al., 2012) SST Bias (Wang et al., 2014) 14

  15. What is involved in resolving all scales (Ocean Model)? • Temporal resolutjon : 1 second • Spatjal resolutjon : 10 -3 m • Volume of the Ocean: 1.3 x 10 18 m 3 • Integratjon duratjon : Millennium • Total # tjme steps : 3 x 10 10 • Total # grid cells : 1.3 x 10 27 (Which is 10 4 tjmes more than an Avogadro’s number) A computer to calculate these many integratjons is not available at present. Even if it is made available it is impossible to complete the simulatjons in reasonable tjme. 15

  16. What Parametrizations are used in climate models? • Unresolved Turbulent transfer processes in the boundary layer and the free atmosphere. • Convectjve and stratjform clouds • Precipitatjon • Radiatjve transfer and heatjng • Heat and water storage • Exchanges between land surface and atmosphere 16

  17. Impact of critical relative humidity on High cloud Simulations? Control (CTL) run: value of RHcrit is set as 85, 85, 85% for low, mid, highlevels, respectively (default values specified in CFSv2). CRH90 run: value of RHcrit is set as 90, 90, 90%. CRHvari run: value of RHcrit is set as 88, 90, 89%. 17 Source: Hazra et al., (2015)

  18. on Simulations (Rainfall) ? Impact of entrainment rate Effect of locally-targeted increases in the entrainment rate on JJAS precipitation. Despite both being moist tropical oceans, changing the entrainment rate in the WEIO and WNP has the opposite impact on precipitation. 18 Source: Bush et al., (2013) ; Andrew Turner

  19. Similar attempts at ECMWF Source: Jung et al., (2008)

  20. Response of Precipitation to various developmental 20 activities in ECMWF

  21. Summary Small Scale Simulatjons (both DNS and LES) should be focused toward decreasing the uncertaintjes in the parametrizatjon used in climate models.  Large scale simulatjon group should carryout a detailed process-based analysis of systematjc model biases and identjfy the lacuna of the parametrizatjon scheme used.  Small Scale Simulatjons group take a clue from the large scale simulatjons (partjcularly the systematjc biases in the climate models) and design and carryout the experiments in such a manner which will help large scale groups to fjne tune the parametrizatjon schemes. 21

  22. Thank You 22

More recommend