B. N. Goswami & Prince K. Xavier Centre for Atmospheric and Oceanic Sciences

1. ENSO Control on Indian Summer Monsoon Through Length ofthe RainySeason (LRS) B. N. Goswami & Prince K. Xavier Centre for Atmospheric and Oceanic Sciences Indian Institute of Science, Bangalore.

2. Shall present: • A new mechanism, not recognized so far, through which ENSO induces decreased precipitation over Indian monsoon region during northern summer. • An objective method of delineating the Indian Summer Monsoon Rainy Season.

3. Interannual variation of All India monsoon (JJAS) rainfall (AIR) between 1871-2002.

4. Changing ENSO-Monsoon Relationship (a) 21-year sliding window correlation between AIR and Nino3 SST, (b) lead-lag correlation between AIR and Nino3 SST during the period 1871- 1971 and 1980-2000.

5. How does ENSO induces decreased Indian summer monsoon Prec.? Current paradigm: Large scale circulation changes associated with ENSO introduces inhibition for organized convection over Indian region. Eastward shift of the Walker Circ. With +ve ENSO Decreased low level divergence over the eastern Equatorial IO. Increased subsidence over continental India. Increased convection over the Equatorial IO. Decreased monsoon rainfall over India. Here, we discover, that ENSO can also induce decreased monsoon rainfall through another mechanism!

6. JJAS Composite of Walker circulation {(U,-ω) averaged <5S-5N>} based on 11 El Ninos between 1950 and 2002 (composite of El Nino SST (JJAS) is shown in the horizontal plane (shaded)) JJAS Composite of Monsoon Hadley (MH) circulation {(V,-ω) averaged <70E-100E>} based on 11 El Ninos between 1950 and 2002

7. Implicit in all these is an assumption (blindly!) that the ‘Indian summer monsoon season’ is of fixed duration! The ‘Indian summer monsoon’ is a physical phenomenon driven by large scale heating gradients that vary in intensity and duration from year to year. Therefore, the actual length of the physical monsoon season may vary from year to year. Thus, there is another degree of freedom , namely the length of the rainy season (LRS) that may influence the ENSO-Monsoon relationship.  There is great need for an objective definition to delineate the Indian summer monsoon SEASON.

8. Daily GPCP Precipitation averaged over <70E-90E, 8N-30N> from May 1 till 30 October

9. Many monsoon ‘Onset’ over Kerala (MOK) take place much before June 1 and ‘Withdrawal’ from Kerala also takes place after September 30. • Monsoon rain from spells before June 1 and after Sept. 30 are traditionally not included in the Seasonal mean (JJAS) rainfall! • Could influence the interannual variability of Indian summer monsoon rainfall! • All teleconnections studied so far with JJAS rainfall (e.g. ENSO-monsoon, monsoon-snow etc) may be completely misleading of physical relationships!

10. onset • Define Indian summer monsoon rainy season • Traditionally the Indian summer monsoon season is defined as between June 1 and September 30 (for convenience!). • What really delineates the Indian Summer Monsoon (rainy) Season? • Physically, the rainy season is delineated by Monsoon ‘Onset’ over Kerala and ‘Withdrawal’ from the southern tip (say 10N). withdraw TCZ

11. Whatever controls the MOK and ‘withdrawal’ of Monsoon from southern tip of India (~10oN) , therefore, determines the length of the Indian summer monsoon season or the Length of the Raining Season (LRS). Thus, if we can agree upon an objective definition of MOK and withdrawal of monsoon from the southern tip, we can define LRS or the Monsoon Season. Can we use existing definitions of MOK and withdrawal? Almost all existing definitions of MOK or withdrawal are not physically based and require a ‘magic’ threshold on precipitation and/or low level wind shear! Unsatisfactory. To our knowledge, nobody has attempted to define the Monsoon Season objectively using the physical driving that determine the onset and withdrawal!

12. Summary of some past definitions: Ananthakrishnan et al. (1968, J. Climatol. 8, 283-296; 1983, Curr. Sci. 52, 155-164) Precipitation based for MOK . Transition to sustained heavy rainfall. Based on 70+ raingauge stations over Kerala. Onset is the date when transition from light to heavy rainfall takes place that is sustained for more than 5 days above a threshold of 10 mm/day. MOK by IMD Rainfall criterion like Ananthakrishnan et al. but combined with subjective judgment of forecasters. This includes increase in K.E of the Low Level Jet (LLJ) , low level westerly shear etc. Wang and LinHO 2002, J. Climate, 15, 386-398 Again introduces a rainfall threshold but introduces the seasonality RRi = Ri - RJan Where, RRi is the relative pentad mean rainfall. This measured as specific pentad mean Ri relative to winter mean RJan.. The threshold used is 5 mm/day. They show that this criterion may be useful in defining the ‘onset’ and ‘withdrawal’ of monsoon over south as well as east Asia.

13. Wang and LinHo 2002, J.Climate, 15, 386-398 withdrawal from s. India is too late! Not appropriate for defining ISM season. Because, the rainfall criterion can not distinguish summer and winter monsoon rainfall.

14. Fasullo and Webster, 2003, J.Climate, 16, 3200-3211 HOWI: Vertically integrated moisture transport withdrawal too early! Again can not be used to define ISM season.

15. He et al. 2003, J. Meteorol. Soc. Japan, 81, 1201-1223. Define monsoon ‘onset’ in terms of change in sign of meridional gradient of upper tropospheric temperature (200 hPa-500 hPa) Reg.B <17.5N-25.5N,70-80E> Probably the most physically based definition.

16. All these definitions (except that of He et al. 2003) are based on some criterion related to rainfall and not based on the physical processes that drive the MOK and ‘withdrawal’. Here, we propose to define the rainy season based on the physical process that drives the Indian summer monsoon. To do this we have to start with asking… What drives the Indian summer monsoon?

17. Long term mean JJA precipitation and DJF precipitation Wet summer-dry winter  Major character of monsoon

18. During summer monsoon season, the circulation is characterized by Low level, cross-equatorial flow, south-westeries, westerly jet in Arabian sea Tibetan anticyclone & Upper level easterlies, monsoon easterly jet  Deep baroclinic vertical structure

19. Annual Evolution of the Indian monsoon. Precipitation averaged over 70E-90E (shaded) and KE of the 850 hPa LLJ (50E-65E, 5N-15N) from observations. KE of LLJ Onset

20. The classical land-sea contrast theory is inadequate! Courtesy : JS

21. Courtesy : JS

22. What drives Indian summer monsoon is not north-south contrast of surface temperature but the meridional gradient of Tropospheric Heating!

23. Tropospheric temperature (TT, in oC) averaged over 200 hPa- 700 hPa (shaded) and 850 hPa winds. JJAS average. TT (in oC)averaged over 200 hPa- 700 hPa (shaded) averaged between 70E-100E as a function of time and latitude.

24. Apparent Heat source Q1 and apparent moisture sink Q2 (5) (6)

25. Meridional gradient of TT is closely related to the meridional gradient of tropospheric heating. From Li and Yanai, 1996, J. Climate, 9, 358-375

26. ‘Onset’ and ‘withdrawal’ are also controlled by the heating gradient Annual evolution of rainfall over the monsoon region. Climatological mean daily precipitation averaged over 70E-100E. Annual evolution of TT (200 hPa -700 hPa) over the monsoon region. Climatological mean daily TT averaged over 70E-100E.

27. The real ‘onset’ is followed by sustained northward propagation of TCZ. • Time-latitude section of CMAP anomalies (unfiltered) averaged over 70E-90E. Only +ve anomalies >2m/day is plotted. C.I. is 2 mm/day. • Northward propagation of spells Dashed line  K.E of 850 hPa winds averaged over the LLJ (55E-65E,5N-15N) ONSET  K.E >100 mm2s-2 and P > 6 mm/day

28. JJAS Climatological mean vertical shear of zonal wind (U200 – U850) Large easterly shear is crucial for northward propagation of the TCZ (Jiang et.al. 2003)

29. TT (contour and shaded) Onset reversal of meridional gradient of TT around 10N

30. TT (contour and shaded) Onset reversal of meridional gradient of TT around 10N

31. TT (contour and shaded) and U200 = 0 Onset reversal of meridional gradient of TT around 10N

32. TT (contour and shaded) and U200 = 0 Onset reversal of meridional gradient of TT around 10N

33. Another element of the onset puzzle: Sharpness of the ‘Onset’! Associated with an instability. Hypothesis: Symmetric intertial instability is responsible for it. (Krishnakumar V. and Lau K.M. , 1998, J. Met. Soc. Japan, 76, 363-383 Krishnakumar V. and Lau K. M. , 1997, Tellus, 49A, 228-245, Tomas and Webster 1997, QJRMS, 123, 1445-1482) (also see Review conditional symmetric instability by Schultz & Schumacher, MWR, 1999)

34. If perturbation is in slantwise path (rather than vertical or horizontal), if the mean wind is in x-direction and in thermal wind balance, stability of such motion depends on relative slope of potential temperature Θ-surface and M surface. The resulting circulation is symmetric when viewed along dir. Basic flow.

35. Condition for dry inertial instability is given by: Absolute zonal momentum Where, In terms of Ertel’s potential vorticity P (Charney, 1973), the condition is; Where, In terms of Richardson No. Ri ,the condition is equivalent to Where, Brunt Vaisala frequency

36. 850 hPa ‘Dynamic Equator’ Climatological mean Absolute Vorticity (zeta + f) for JJA , from NCEP Reanalysis

37. Streamlines of climatological mean (-ω,V) averaged between 60E-95E, over 10-day periods from mid-April to mid-June. To note: 1.Northward movement of deep upward motion (TCZ), rapid between last week of May and first week of June. 2.The barrier of massive descending motion is overcome at the time ‘Onset’. 3.The shallow meridional circulation during pre-onset takes north warm moist air near the surface and brings south dry air above PBL

38. Latitude of absolute vorticity =0, averaged over 70E-100E Potential Convective instability index (Θe (700)-Θe (1000)) Precip. Averaged Over 70E-100E,10N-30N

39. Events that lead to the Indian summer monsoon ‘Onset’ (MOK) Surface heating (land-ocean contrast) during pre-monsoon season produces cross-equatorial flow near the surface but is capped by subsidence and a southward flow above the PBL. Builds up potential convective instability, but can not be realized. When tropospheric heating gradient changes sign, primarily due to the influence of the Tibetan Plateau heating, cross equatorial flow and a large scale cyclonic vorticity above the PBL is set up. Zero absolute vorticity line at 850 hPa moves north to about 5N and conditions for dry symmetric inertial instability as well as conditional moist inertial instability is established. Dry inertial instability overcomes the inhibition of subsidence, moist inertial instability takes over and explosive organized convection takes place. Onset has arrived!

40. First EOF of climatological mean TT (shaded). Zero contour around 10N delineates boundary between the heat ‘source’ in north from the heat ‘sink’ in the south. TTn = TT in the north box TTs = TT in the south box TT = TTn - TTs Lat. Of absolute vorticity, η =0. <50E-100E> U200 –U850 <50E-100E,0-15N> PC1  black

41. Weaker meridional migration of the TCZ in the EAM and WPM is due to weaker TT and weaker U200 – U850 in those regions. EAM WPM SAM TT U200 – U850 <respective lon. Belts>

42. Using NCEP/NCAR reanalysis from 1950-2002 onset dates (OD), withdrawal dates (WD) and length of the rainy season (LRS = WD-OD) are calculated. Statistics of Onset dates (OD), withdrawal dates (WD) and length of the rainy season (LRS) in Julian days from NCEP/NCAR reanalysis between 1950-2002. Climatological mean OD  29th May Climatological mean WD 4th October

43. Correlation between OD, WD, LRS and Other climate parameters asignificance at 5% level bsignificance at 1% level

45. Correlation between OD, WD,LRS and Nino4 SST anomalies of each month from NCEP/NCAR reanalysis (NC) as well as ERA.

46. Actual dates of OD, WD and LRS from NC and ERA from 1950 onwards. (in Julian days)

47. Actual dates of OD, WD and LRS from NC and ERA from 1950 onwards. (in Julian days)

48. Composite of Prec. • 3 pentad before OD-OD Prec. • 2 pentad before OD- OD Prec. • 1 pentad before OD- OD Prec. • on OD Prec. • 1 pentad after OD- OD Prec. • 2 pentad after OD- OD Prec. • 3 pentad after OD- OD Prec. Composite P from CMAP during Onset (OD) based on deltaTT

49. Composite of 850 hPa winds • 3 pentad before OD-OD winds • 2 pentad before OD- OD winds • 1 pentad before OD- OD winds • on OD winds • 1 pentad after OD- OD winds • 2 pentad after OD- OD winds • 3 pentad after OD- OD winds

50. Correlation between (a) OD and TT during 15May-15June (b) WD and TT during 15Sep-15Oct. (c) LRS and TT during 15Sep.-15Oct.