Thesis Committee Report 6 Andrew Turner Supervisors: Prof. Julia Slingo, Dr Pete Inness, Dr Franco Molteni (ICTP, Trieste) Thesis Committee: Dr D. Grimes (chair), Prof. A. Illingworth 13 July 2005
ENSO-Monsoon relationships in current and future climate This thesis aims to: a Assess predictability of rainfall anomalies over monsoon affected countries in multi-decadal simulations of current climate, b Assess reliability of tropical rainfall estimates for future climates using current diagnostics and time-slice experiments. Recent work The effect of climate change on the monsoon-enso system Since the last meeting, two further 100-year integrations of the Unified Model have been carried out. These represent future climate conditions (2 CO 2 ) in both the standard version of the model (HadCM3) and the limited-area flux corrected version (HadCM3FA). Both integrations use a vertical resolution of L30 as before. The HadCM3 2 CO 2 experiment was initialised from a previous Hadley Centre run where carbon dioxide was ramped up at a rate of 1% per year, then kept constant. A further period of 10 years was discarded, allowed as a cautious spin-up to 30 vertical levels in the atmosphere, as in previous integrations. This experiment is designed to look at the effect of climate change on the monsoon climate, the ENSO system and the teleconnection between them. That the integration is non-transient allows the teleconnection to be studied in detail. Clearly, if HadCM3 features systematic errors under a 20th century forcing scenario then these errors will also be present under future greenhouse forcing conditions. Making the assumption that the errors will be of the same magnitude under future conditions, the same heat flux adjustments are applied to an integration initiated in the future climate. Although the nonlinearity of the modelled coupled atmosphere-ocean system makes it unlikely that the basic state errors would be of equal magnitude, previous greenhouse forcing experiments with older coupled GCMs used the same seasonal cycle of anomalous fluxes (e.g., Collins (2000) using HadCM2) as in their current climate. In any case the application of this system flux correction to a future climate will serve as a useful experiment and verification of conclusions already drawn in Turner et al. (2005). Adding heat fluxes to an already warmer Pacific ocean proves challenging for the model code and in the first instance, numerical instabilities build up after a period of 35 years. The integration is restarted using slightly different initial conditions, by offsetting the ocean and atmosphere start dumps by exactly five years. A successful 100 year integration then follows, of which the first five are discarded, allowing the coupled system to adjust to the start dump mismatch. Figure 1 shows the effect of climate change (2 CO 2 ) on the mean summer climate in the HadCM3 GCM. First considering surface temperatures (Fig. 1a,d), it is clear that warming has taken place across the whole of the Indo-Pacific sectors. Land regions have warmed by a greater degree than the sea due to the large heat capacity of water. There is also evidence of an El Niño-like warming in the equatorial Pacific, i.e., the east has warmed more than the west. This pattern of warming in the mean climate amounts to a reduction in the zonal equatorial temperature gradient, and may lead to some monsoonal changes, as noted by Meehl et al. (2000) who saw suppression of rainfall in southeast Asia. However, in their review of several models, Collins et al. (2005) found the most likely scenario to be one of no El Niño amplitude change. The surface heating differential has lead to an increase in the land-sea meridional gradient of temperature in the Indian region. This may act to strengthen the monsoon dynamics, as argued by Hu et al. (2000) who found enhanced moisture convergence over land in response to an increased meridional temperature gradient. Meehl and Arblaster (2003) however found SST forcing to be more important. Looking at the lower tropospheric winds (850hPa) in Fig. 1b,e, it is clear that increased greenhouse forcing has affected the mean winds in only a few localities. The Pacific trade winds 1
(a) 2xCO2 HadCM3 surface temperatures (d) 2xCO2 minus control surface temp 30N 15N EQ 15S 30S 0 30E 60E 90E 120E 150E 180 150W 120W 90W 60W 0 30E 60E 90E 120E 150E 180 150W 120W 90W 60W 276 284 292 296 300 304 310-3 -2-1 0 1 2 3 272 280 288 294 298 302 306-3.5-2.5-1.5-0.5 0.5 1.5 2.5 (b) 2xCO2 HadCM3 850mb winds 30N (e) 2xCO2 minus control 850mb winds 15N EQ 15S 2 30S 30E 60E 90E 120E 150E 180 150W 120W 90W 30E 60E 90E 120E 150E 180 150W 120W 90W 20 (c) 2xCO2 HadCM3 precipitation 30N 5 (f) 2xC02 minus control precipitation 15N EQ 15S 30S 0 0 30E 60E 1 90E 120E 5 150E 180 150W 9 120W 90W 15 60W 0 30E -6 60E -4 90E 120E 150E 180-2 -0.5 0.5 150W 2 120W 4 90W 6 60W 0.5 3 7 12-7 -5-3 -1 0 1 3 5 7 Figure 1: HadCM3 2 CO 2 summer (JJAS) climatologies of (a) surface temperature (K), (b) 850mb winds (ms 1 ) and (c) precipitation (mm.day 1 ). Panels (d) to (f) show differences in these fields from HadCM3 under pre-industrial conditions. Unit wind vectors are 20ms 1 and 5ms 1 for the wind climate and differences respectively. Speckling on the precipitation difference (f) indicates significance at the 95% level using a student t-test.
(a) Niño-3 SST vs. DMI (b) Niño-3 SST vs. Indian rainfall Figure 2: Correlation between Niño-3 SST and summer (JJAS) (a) DMI or (b) Indian rainfall plotted against lag time. In (b), AIR refers to the All-India Rainfall precipitation data being correlated with ERA-40 SSTs. Dashed lines are for the 2 CO 2 scenarios. Aug-1 indicates the August Niño-3 anomaly the year before the measured rainfall. Correlations significant in all but 5% of cases are indicated outside the horizontal dashed lines: r > 0.31 for the ERA-40, r > 0.2 for the model integrations. at about 5 N are approximately 1ms 1 weaker across the region 150 E 150 W, likely as a result of the El Niño-like surface warming. Over the Indian Ocean, the flow at lower levels is more easterly just west of Sumatra, i.e., the Somali Jet is weaker in its final push across the south Bay of Bengal. The south westerly monsoon flow is stronger by almost 1ms 1 in the northern Arabian Sea and then on over the north of India. This amounts to a slight northward motion of the monsoon flow structure, which May (2004) attributes to the increased meridional heating gradient. The anomalous easterlies out of Sumatra give rise to a small anomalous anticyclone in the southern hemisphere, leading to a dryer precipitation climate there (up to 3mm.day 1 in Fig. 1f). Significant increases (at the 95% level using a student t-test) in daily precipitation are noted in the main summer ITCZ and SPCZ bands across the Pacific Ocean. The ITCZ is wetter by 2mm.day 1 right across the Pacific basin, with up to 5mm.day 1 extra at 120 W. The SPCZ east of New Guinea is wetter by more than 2mm.day 1. Now considering the effect of increased greenhouse forcing on the Asian summer monsoon, it is clear that daily precipitation during the summer season increases by between 0.5mm.day 1 in the northern states and tip of peninsula India, and 2mm.day 1 in the north-eastern Bay of Bengal, Bangladesh and Myanmar. The South China Sea off Hong Kong and Vietnam also sees significant increases of this magnitude. The pattern of precipitation increase is very similar to that found by Meehl and Arblaster (2003) using the Parallel Coupled Model (PCM), although they found no drying south-west of Sumatra. They attribute the increase in mean south Asian monsoon rainfall to an increased moisture source from the warmer Indian Ocean. The effect of climate change on the flux corrected version of the model is not shown here but amounts to a stronger response of the same pattern of changes seen in Fig. 1. It is thus possible that due to its systematic errors in the tropical oceans, HadCM3 may be underestimating the change due to increased greenhouse gas forcing. Another important effect of climate change may be on the predictability of the monsoon, via its teleconnection to ENSO. Accordingly, Fig. 2 shows the lag-correlation between (a) the DMI (dynamical monsoon index, Webster and Yang (1992)) or (b) AIR (All-India rainfall) and the Niño-3 index. In the case of the monsoon dynamics, the teleconnection is strengthened under climate change in both versions of the model. The timing of the HadCM3 teleconnection is later and compares to present day observations. The suggestion is that monsoons thus become more predictable under 2 CO 2 conditions. Looking at Fig. 2(b), however, it appears that the effect on the teleconnection is less robust. The HadCM3 monsoon-enso teleconnection 3
Figure 3: HadCM3FA 2 CO 2 Niño-3 anomaly index. appears to weaken, and stays broadly the same strength in the HadCM3FA version. However, the HadCM3FA 2 CO 2 teleconnection is noticeably more biennial using both measures of monsoon intensity, marked by the moderate positive correlations at lead times of 8-12 months. That the teleconnection has become more biennial suggests major changes to the monsoon- ENSO coupled system in the flux corrected future climate. Looking at the Niño-3 anomaly index for the HadCM3FA 2 CO 2 integration in Fig. 3, it is clear there are two distinct regimes of ENSO. A period of irregular El Niños is followed by an abrupt change to one of a strongly biennial nature. Later the irregular ENSO returns. The integration can be split into composites of the irregular and biennial regimes which yield quite different characteristics. The irregular regime features many more westerly wind events (WWEs) in the west Pacific region, consistent with stochastic forcing of irregular ENSO oscillators (see, e.g., Jin (1997)). There is also more intraseasonal oscillation activity (MJO) during this regime. The biennial period of the integration locks ENSO much more strongly to the seasonal cycle, and there are fewer cases of stochastic forcing events on synoptic timescales. This relates to the biennial regime being a self-excited limit cycle as in Jin (1997). Wang et al. (1999) describe the monsoon system as having a modulating effect on the intraseasonal forcing, pushing an unstable system towards a biennial oscillation. Wang et al. (1999) also suggest that stochastic shortening of the oscillation period in this way increases the phase locking to the annual cycle. I speculate that in some way the flux adjustments are pushing to favour biennial oscillation, (already suggested in Turner et al. (2005)), and that perhaps an anomalous monsoon aids the regime change. The final part of my thesis will focus on the different behaviour of the two regimes and suggest possible mechanisms for change. A second integration of HadCM3FA 2 CO 2 under different inital conditions is complete and may yield more regime shifts for study. The teleconnection patterns in Fig. 2 raise important questions about interdecadal and longer shifts in ENSO behaviour and their relation with the monsoon-enso teleconnection. If the teleconnection of the HadCM3FA 2 CO 2 integration is broken down into two regimes (not shown), it is noted to be very much stronger during the biennial regime, and much weaker during the irregular period. The DMI and AIR measures of monsoon strength are much more strongly correlated during the biennial era, the DMI timeseries being noticeably biennial also. That the biennial years dominate the teleconnection pattern suggests an inherent danger in using regression models, i.e. using historical data dominated by one mode of ENSO may give an overestimate of the skill to be found during a different regime. Transferable Skills Since the last meeting I have: Given a talk at the joint ECMWF/CGAM workshop on the Tropical Ocean and Atmosphere, Variability and Prediction in the department (April). Given a talk at the joint IPSL/CGAM workshop on the Tropical Climate in the department 4
(May). Presented a departmental lunchtime seminar (June). Given a CHAPA club talk (June). Presented a paper at the 5th GEWEX Conference in Costa Mesa, California (June). Done further computing course demonstrating. I will be presenting a paper at the RMS conference in September. 5
SEP/OCT Thesis Plan Chapter 1: Introduction Chapter 2: Scientific Background JUL/AUG/SEP The original literature review written in 2003 needs updating with information on ENSO theory and climate change influences on the monsoon-enso system. Chapter 3: The role of the Basic State in the ENSO-Monsoon Relationship and Implications for Predictability This chapter has been published as Turner et al. (2005) but needs slightly expanding, to include detail that was left out of the publication. 1 Introduction 2 Model, observed datasets and experimental design 3 The influcence of tropical Pacific SST errors on the GCM basic state 4 The role of the basic state in monsoon variability 5 The impact of the basic state on ENSO variability 6 The basic state and monsoon-enso teleconnections 7 Conclusions Chapter 4: The effect of climate change on the monsoon-enso system JULY Results from the 2 CO 2 integrations of HadCM3 and HadCM3FA will be included here. This chapter is currently being written. 1 Introduction 2 Experimental design 3 The influcence of climate change on the GCM basic state 4 The role of climate change in monsoon variability 5 The impact of climate change on ENSO variability 6 Climate change and monsoon-enso teleconnections 7 Conclusions: implications for future climate monsoon prediction Chapter 5: Different ENSO regimes in a future climate and their effect on monsoon prediction AUGUST Some work is still being carried out on this chapter and it will be written in due course. 6
1 Introduction 2 Two distinct modelled ENSO regimes 3 Characteristics of the irregular regime 4 Characteristics of the biennial regime 5 Causes of regime transition 6 Products of flux correction which favour biennial behaviour 7 Changes to the monsoon-enso teleconnection as ENSO changes 8 Conclusions: implications of ENSO change for monsoon prediction Chapter 6: Conclusions SEP/OCT The implications of this study for monsoon prediction. 7
Bibliography Collins, M. (2000). The El Nino-Southern Oscillation in the Second Hadley Centre Coupled Model and its response to greenhouse warming. Journal of Climate, 13, 1299 1312. Collins, M., (Australia), T. C. M. G. B., (Canada), C., (Japan), C., (France), C., (Australia), C., (Germany), M., (USA), G., (USA), G., (China), I., (Russia), I., (France), L., (Japan), M., (USA), N., (USA), N., (UK), H. C., and Korea)), Y. S. (2005). El Nino- or La Nina-like climate change? Climate Dynamics, 25, 89 104. Hu, Z.-Z., Latif, M., Roeckner, E., and Bengtsson, L. (2000). Intensified Asian summer monsoon and its variability in a coupled model forced by increasing greenhouse gas concentrations. Geophysical Research Letters, 27, 2618 2684. Jin, F.-F. (1997). An equatorial ocean recharge paradigm for ENSO. Part I: conceptual model. Journal of the Atmospheric Sciences, 54, 811 829. May, W. (2004). Potential future chnages in the Indian summer monsoon due to greenhouse warming: analysis of mechanisms in a global time-slice experiment. Climate Dynamics, 22, 389 414. Meehl, G. A. and Arblaster, J. M. (2003). Mechanisms for projected future changes in south Asian monsoon precipitation. Climate Dynamics, 21, 659 675. Meehl, G. A., Collins, W. D., Boville, B. A., Kiehl, J. T., Wigley, T. M. L., and Arblaster, J. M. (2000). Response of the NCAR climate system model to increased co 2 and the role of physical processes. Journal of Climate, 13, 1879 1898. Turner, A. G., Inness, P. M., and Slingo, J. M. (2005). The role of the basic state in the ENSO-monsoon relationship and implications for predictability. Q. J. R. Meteorol. Soc., 131, 781 804. Wang, B., Barcilon, A., and Fang, Z. (1999). Stochastic Dynamics of El Nino-Southern Oscillation. Journal of the Atmospheric Sciences, 56, 5 23. Webster, P. J. and Yang, S. (1992). Monsoon and ENSO: Selectively interactive systems. Quarterly Journal of the Royal Meteorological Society, 118, 877 926. 8