Mechanistic links between the tropical Atlantic and the Indian monsoon in the absence of El Nino Southern Oscillation events Vijay Pottapinjara 1*, Roxy Mathew Koll2, Raghu Murtugudde3, Girish Kumar M S1, Subimal Ghosh4, Ashok Karumuri5 and M Ravichandran1,6 Indian National Center for Ocean Information Services, Hyderabad, India 2 Indian Institute of Tropical Meteorology, Pune, India 3 Earth System Science Interdisciplinary Center, University of Maryland, College Park, MD, USA 4 Department of Civil Engineering, Indian Institute of Technology, Bombay, India 5 University Centre for Earth System Science, University of Hyderabad, Hyderabad, India 6 National Centre for Antarctic Ocean Research, Goa, India 1 Sixth WMO-International Workshop on Monsoons at Singapore during 12-17 Nov 2017
The Atlantic Zonal Mode (AZM) or Atlantic Nino seems to be a dominant phenomenon influencing the ISMR, especially in the absence of an ENSO event After removing the ENSO effect on ISMR, linearly Kucharski et al., 2009
Relation between the Atlantic Zonal Mode (AZM) and summer monsoon rainfall over India Cold AZM Warm AZM During a cold AZM, enhanced rainfall occurs over certain parts of India because of more monsoon depressions with long clustered tracks forming in the Bay of Bengal due to strengthened mean monsoon flow, increased moisture availability in the midtroposphere and reduced vertical wind shear; vice versa for a warm AZM
Physical mechanism relating the Atlantic Zonal Mode and Indian summer monsoon rainfall A schematic showing the effect of a warm AZM The anomalies of wind in the western and heat content in the eastern equatorial Atlantic associated with the Atlantic Zonal Mode which is akin to ENSO, may offer some predictability for the Indian summer monsoon one season in advance via the mechanism outlined above, especially in the absence of ENSO (Pottapinjara et al., 2014; 2015).
Hypothesis: That the AZM influences the ISMR by perturbing the tropospheric temperature gradient between over the land and Indian Ocean through changes in the Kelvin like waves propagating in the troposphere (Gill type response). This may be prominent especially in the absence of an ENSO event. Testing the hypothesis using a coupled model: The model For our study we used, Coupled Forecast System (CFS) version 2. The CFSv2 model simulates the AZM, mean monsoon and its interannual variability reasonably well. CFSv2 = Ocean (MOM, 0.25-0.50 horizontal, 40 vertical) + Atmosphere (GFS, T126 horizontal, 64 hybrid vertical) No flux correction between the ocean and atmosphere. Saha et al. 2013 The test See if the model simulates the AZM and its relationship with ISM If it is doing well, go for the sensitivity experiments to test the hypothesis
Despited the delayed and weak cold tongue, AZM is simulated reasonably well in the model Observations CFS Cold AZM Warm AZM Cold AZM Warm AZM In CFS, cold or warm anomaly over the seasonal cold tongue region is not as strong and extensive as it is in the observation. The already noted delayed development of seasonal cold tongue in CFS is reflected in the persistence of SST anomalies into September
In principle, the propagation of AZM effect through TT is captured in CFS Lead-lag correlations between the Atl3 index and integrated tropospheric temperature (TT) in CFS (removing ENSO effect over TT) TT leads by 2 months TT leads by 1 month Simultaneous Atl3 leads by 1 month Atl3 leads by 2 months
Relation between the AZM and ISMR in observations and in CFS is good enough over the Western Ghats but not so over Central India for several reasons Observations (GPCP) CFS a) ONI vs Rainfall c) Atl3 vs Rainfall b) ONI vs Rainfall_noenso d) Atl3 vs Rainfall_noenso a) ONI vs Rainfall b) ONI vs Rainfall_noenso c) Atl3 vs Rainfall d) Atl3 vs Rainfall_noenso In each panel, (top row) JJA(S) correlations between the ONI index and rainfall before (a) and after (b) removing the ENSO effect over rainfall. (bottom row) similar to the top but the correlations are between the Atl3 index and rainfall From the observations, the AZM seems to reduce the rainfall over a band extending from the central to northwest India and also along the western Ghats while enhancing the rainfall over northeast India. With an exception of along the western Ghats, the CFS fails to capture the relationship.
The experiment(s) Constant SSTA experiment: the seasonal mean response of the monsoon to the AZM. It means, whenever the AZM SST anomalies are imposed in the Atlantic their magnitude and the pattern will be the same throughout JJA(S) season. Another set of experiments may be conducted suppressing ENSO; a different set by imposing the cold anomaly, if possible. SSTA pattern can be evolved at a later stage. Response = Average ( Sensivity run Freerun ); Experiment run for 20 years May Freerun JJA SSTA pattern (Obs.) May Jan Jan Dec Dec IC + pattern IC + pattern JJASO JJASO
The pattern in the Atlantic is similar to what is imposed but with some differences that might be important Seasonal (JJAS) average of Surface Temperature (Atmosphere) The red box indicates the area where SSTA were added
On the seasonal mean, the circulation pattern and associated rainfall response shifted to the south Seasonal (JJAS) average of Precipitation (shaded; mm/day), OLR (contour in magenta) and Winds at 850 hpa (vectors). The climatological southwesterlies seem to be weekended by the wind flow. Anticyclonic flow over the Bay of Bengal and adjacent region can be noted.
On the seasonal mean, the circulation pattern and associated rainfall response shifted to the south So far, the model response is as expected as per our hypothesis. However, the precipitation (and OLR) response seems to be opposite to what is expected (see the circles). Interestingly, regions identified by red circles are almost the same (central India and Western Ghats) as in Observations where the AZM effect was found to be significant. Question: Is the precipitation response consistent with that of the winds, physically? Not in the seasonal average but it is in the monthly means. Seasonal (JJAS) average of Precipitation (shaded; mm/day), OLR ( contour in magenta) and Winds at 850 hpa (vectors)
The circulation response extends to the western equatorial Pacific and tropospheric temperature seems to support that ( wave propagation is not clear) JJAS Average of the model response in the low level wind and mid-level Air Temperature
The circulation response in the seasonal mean is dominated by that in August (favorable); but shifted response Monthly (Jun-Sep) averages of low level wind Seasonal average wind response is dominated by that in the month of Aug. Note the presence of an anti-cyclone in the BoB in Aug. Also note the alternating direction of flow in the Arabian sea over to the Western Ghats
The seasonal mean precipitation is dominated by that in September (unfavorable); response in August is most desired Monthly (Jun-Sep) averages of Precipitation The response in precipitation is consistent with the wind response. Seasonal average precipitation response is dominated by that in the month of Sep. That may explain why there seems to be an inconsistence between the response in the wind and precipitation in the seasonal average
The heating over equatorial Indian Ocean in August is consistent with precipitation; Hadley circulation over IO longitudes Monthly (Jun-Sep) averages of tropospheric temperature
Conclusion and the way forward CFSv2 is able to reproduce the AZM and its relation with the ISMR reasonably well The means response is different in different months and it is most desired in the month of August The effective warming seen by the atmosphere in the Atlantic is not what it is supposed to be. It is mostly to the south of equator. This may need to be taken care of. Improving the design of the experiment may bring out the signal more clearly Suppression of ENSO in future experiments may also improve the signal clarity Given that there is a warm bias in the Atlantic cold tongue, imposing a cold anomaly might also help References: Kucharski et al (2009), Quart. J Royal MetSoc., 135, 569-579 Pottapinjara et al (2014), Journal of Geophysical Research-Atm., 119(11), 6456-6469 Pottapinjara et al (2015), International Journal of Climatology, 36(6), 2469-2480 Saha et al (2013), J. Climate 27, 2185 2208 (2013)
Thank you
Additional Information
Delayed development of the cold tongue and hence the warm bias Difference of monthly climatologies (Jan Dec) of CFSv2 and observations (modelobservations) SST overlaid by winds Atl3 region: 3S 3N & 20W 0W Corresponding month is indicated on each panel 10 m/s
JJA(S) wind composites w.r.t. AZM in observations (NCEP-2) JJA(S) wind composites w.r.t. AZM in observations (CFSv2) From the observations, during a cold (warm) AZM, the climatological southwesterlies are strengthened (weakened). They bring more (less) moisture that may increase (reduce) the rainfall over India. Note that similar composites in CFSv2 capture the essence as in observations, although the magnitude and location may be different. Interestingly, this AZMwinds relation directly reflects in the rainfall along the western Ghats. However, the rainfall over central India may be sensitive to the location of these wind anomalies along with other factors. Pottapinjara et al 2014 show that the monsoon depressions in the Bay of Bengal contribute to the rainfall over central India and the AZM influences the factors (positive vorticity, less wind shear and availability of moisture in the midtroposphere) conducive for the formation of the depressions. This angle is examined below.
There is a disconnect between the SST and Surface Temperature in other ocean basins; might have consequences 20-year seasonal (JJAS) average of SST The red box indicates the region we should be looking at for the response The Indian Ocean, especially the Arabian sea seems to cool down. The central and eastern Pacific appears to cool down as well (La Nina?). Interestingly, there is also meridional SST gradient in the Pacific.