SAPRi SE Project Identifying the role of the

SAPRi. SE Project Identifying the role of the Sun and the El Niño Southern Oscillation on Indian Summer Monsoon Indrani Roy & Mat Collins 25 th June, 2013

Outline I: Observation Role of Sun and ENSO on ISM Ø Background Ø Analyses II : CMIP 5 Model Output Some features of ISM, ENSO in CMIP 5 models ISM and ENSO teleconnection Hydrological Cycle in CMIP 5 models and ISM

Role of sun and ENSO on ISM

Background: General Circulation Walker Cell: Hadley Cell: India Hadley Cell: Thermally driven rising air around ITCZ and falling 30 lat. Walker Cell: EW circulation in tropics and changes direction in warm and cold phases of ENSO

Background : Indian Summer Monsoon(ISM) SLP: Climatology Monsoon means seasonal wind reversals. For ISM, it is from NE ly (Jan) to SW ly (July). January Associated with movement of ITCZ. Heavy rains in summer (JJAS) due to moisture rich air from ocean. Walker circulation and Hadley circulation both play role. ISM has changed in last few decades (IPCC, 2007) July

Background: Major Modes of Variability Tah iti (Australia) Different regions affected by various modes

Why NAO? Positive Phase Negative Phase Anticorrelation between NAO and ISM [Sen Roy, 2011] ISM is strongly modulated by the NAO. [Liu & Yanai, 2001] Relationship bet. temperature W. Eurasia and ISM is stronger, over same period the rel. between the ENSO, ISM weakened. [Chang et al. [2001]

Why Indian Ocean Dipole (IOD) ? Positive Phase Negative Phase Connection: ISM- Australia- E. African rainfall IOD and ENSO have complementarily affected ISM during 1958 -1997 (Ashok et al, 2001)

Atmos-Ocean coupling was different during 1950 s to 1997 Global warming caused weakening of tropical circulation: more in the Walker cell than the Hadley cell Strong decrease in intensity of Walker circulation after 1950 s Modest intensification since 1998 (Vecchi, et al. 2007) [Mc. Phaden and Zhang, 2004] Could climate change during that period have modified ISM?

Solar signal on ISM detected using Solar peak year compositing JJA Max-yr JJA Min-yr Anomaly Van Loon and Meehl (2012) only used solar max-yr compositing on SLP and rainfall and suggested sun enhances ISM Is it true solar signal?

Average SSN Solar Peak years and ENSO (DJF) [Roy and Haigh, 2010] Almost all solar peak years are with –ve ENSO index ENSO signal in peak-yr compositing might be misinterpreted as solar Is any other strong signal also mixed up in compositing method?

Solar compositing on SLP Min r. t. 1956 -1997 Min r. t. 1936 -1975 Max r. t. 1956 -1997 Max r. t. 1936 -1975 Max-yr significant signal around Azore High, min-yr Icelandic Low. Unaffect with change of period of anomaly. SH is mostly affected due to climate change signal during 1956 -1997. During same period, land (+ve)-sea(-ve) SLP contrast favours ISM rainfall

ISM in solar compositing covering India Min r. t. 1956 -1997 Min r. t. 1936 -1975 Max r. t. 1956 -1997 Max r. t. 1936 -1975 Max-yr as well as min-yr compositing suggest similarly around Indian subcontinent. Max-yr suggest stronger effect on ITCZ that min-yr compositing.

Multiple Regression Analysis Trend SSN Volcano ENSO 14

ENSO Signal in JJA 1856 -1955 1956 -1997 ENSO captures SO in SLP, but major changes around Australia in later period. Australia (Darwin), one lobe of SO is also coincidentally one end of IOD. -ve NAO pattern observed in later period. Local N-S Hadley circulation, as manifest as NAO in NH and IOD in SH may have played role in modulating ISM in later period.

Solar Signal in JJA 1856 -2004 Using regression, no significant signal is detected around regions of Indian subcontinent. This is true irrespective of the period considered 1856 -1955 1956 -1997

Some connections between Sun and ISM +ve ENSO index + -ve ENSO index * Rainfall deficit years are usually associated with warm phase of ENSO and vice versa (L. H. S). Some connections, solar cycle and ISM rainfall- different since 1950 s. Decadal solar forcing on trade winds (Meehl, et al 2008, Roy & Haigh , 2012) that acts alongside inter-annual ENSO may be responsible.

Summary I Solar influence on ISM rainfall, using method of solar peak year compositing, may not be robust and can be influenced by factors as ENSO and trends. Compositing suggests SH is mostly affected by climate change signal. Min year detects signal around Icelandic low and max year around Azore high. During 2 nd half of last century, the weakened Walker circulation due to climate change seems to be overtaken by local N-S Hadley circulation, as manifest as NAO in NH and IOD in SH. Some connections between solar cycle and monsoon rainfall, which are different since 1950 s.

ISM and ENSO in CMIP 5 models

Comparison with Observation and another model - last 50 years Observation-CRU : black; Model-all forcing: red Temporal Behaviour (Bollasina, et al 2011, Science) 5 -yr running mean JJAS(relative 1940 - 2005) over central-N India (box) Observation-CRU Model (NOAA GFDL CM 3) all forcing Spatial Behaviour § -ve anomaly in box for both cases.

Spatial Pattern (Historical Run) Precipitation Anomaly w. r. t. (1986 -2005) No consistent pattern – even in box (starting from +ve to –ve shown)

Temporal Pattern (Historial Run)-last 50 years Precipitation Anomaly w. r. t. (1940 -2005) 1 1 1 2 2 2 Better results 11 year running mean Decreasing trend : inmcm 4, Access 1 -0, MIROC-ESM, Nor. ESM 1 -M 1 Increasing : IPSL-CM 5 B-LR, MPI-ESM-P, IPSL-CM 5 A-MR 2

Analysis: Historical +rcp (w. r. t 1985 -2005) Rainfall (box), 11 year running mean, blue rcp scenario. 2 • Most models rising trend in rcp, some overall no trend (e. g. 1 ), some falling trend in rcp scenario (e. g. 2 ). • Similar observation not only in box but also for overall rainfall. 1

SST(Nino 3. 4 in JJAS): rcp (w. r. t 1985 -2005) 3 1 2 • Model with least trend in Nino 1 • Model high variable Nino 2 • All models rising trend in rcp, exception 3

ISM and ENSO Teleconnection Model Correlation : Rainfall vs. Nino 3. 4 Whole India Box I II (His) (rcp) • MIROC-ESM-CHEM -. 15 . 10 . 07 -. 20 • Nor. ESM 1 -M -. 78 - 0. 71 -. 47 -. 40 • MIROC-ESM -. 23 -0. 23 -. 03 -. 10 • Inmcm 4 -. 44 -0. 39 -. 37 -. 27 • ACCESS 1 -0 -. 29 -0. 16 -. 11 -. 15 • CSIRO-Mk 3 -6 -0 -. 26 -0. 30 -. 24 -. 39 • IPSL-CM 5 A-MR -. 66 -0. 54 -. 60 -. 50 • MPI-ESM-LR -. 35 -0. 50 -. 33 -. 52 • Can. ESM 2 -. 49 -0. 65 -. 26 -. 31 • -ve correlation - both in historical (I) and rcp (II)- after removing trend. • True in the box region as well.

Hydrological Cycle in CMIP 5 models and ISM

Global Hydrological Cycle : CMIP 3 (Vecchi and Soden, 2007) Moistening vary model to model, but all models exhibit a nearly linear relationship between column water vapour and surface temperature. The rate of this increase is 7. 5% /K, following Clausius Clapron (C-C) equation.

Precipitation Time Series: CMIP 5 Global Monsoon Global: annual & JJAS resemble. Uncertainty increases with time. Matches CMIP 3 study. Nearterm prediction skill good. Monsoon: large uncertainty throughout. Magnitude-wise much higher.

Annual JJAS Hydrological cycle and ISM: CMIP 5 Atm water and temp Ppt, vert. vel and temp (Globe) Precipitation, vert. vel and temp (India) Although there can be significant regional changes in relative humidity among models, the globalmean behaviour closely resembles that expected from (C–C) arguments also in CMIP 5.

Questions to answer/future work It is usually said that monsoon increases due to increase in water vapour offset by weakening circulation. Hypothesis: Can this explain variations between CMIP 5 models? Does global change reflect regional change? Studies using HADCM 3/ RM 3 with perturbed physics will be carried to understand monsoon dynamics. Emphasis will be on circulation fields.

Summary II ISM - General Features vary model to model in CMIP 5. Model FGOALS-g 2 does not show any trend in nino 3. 4 for historical or rcp scenario. ISM and ENSO teleconnection studied. All models suggest -ve correlation in historical as well as rcp scenario. Hydrological Cycle and ISM were analysed using CMIP 5 models. Hydrological cycle matches to that of earlier CMIP 3 study. Monsoon rainfall suggests large uncertainty in CMIP 5.