Why is ENSO influencing Indian northeast monsoon in the recent decades?

Authors


Abstract

This study examines decadal changes of the El Niño/Southern Oscillation (ENSO) influence on the interannual variability (IAV) of Indian northeast monsoon rainfall (NEMR). In recent decades, the ENSO influence has strengthened over the NEMR variability. During El Niño years, the subsidence anomaly over the warm-pool region of Indonesia was much stronger in the recent decades as compared to the earlier decades. The zonally elongated warm sea surface temperatures (SSTs) of the warm-pool region along the tropical Indo-Pacific Ocean were much warmer than the earlier decades, which had intensified the deep convections over the Indian Ocean. The northwestward propagating trailing Rossby-gyres, formed in the western portion of the equatorial deep convection and stratiform heating, intensified as tropical storms and cyclones over north Indian Ocean. These storms and cyclones strike southern peninsular India and hence flood northeast monsoon (NEM) years. During the recent decades in the La Niña years, the contracted warm SSTs over the warm-pool region were much warmer than the earlier decades which intensified the deep convection and stratiform heating over there. The tropical storms and cyclones that formed due to Rossby-gyre dynamics had to travel a long distance towards the northwestward direction over the Bay of Bengal and misses southern peninsular India, and hence, the drought NEM years. Copyright © 2011 Royal Meteorological Society

1. Introduction

The Indian subcontinent has two principal monsoon seasons: summer and winter. The summer monsoon season spans four months from June through September, and the rainfall during the season contributes about 80% of total annual rainfall of the country. While, the winter monsoon season spans three months from October through December, and the rainfall during the season contributes about 50% of the annual rainfall over the southeastern tip of the Indian peninsula (Rao Krishna and Jagannathan, 1953; Kumar et al., 2007). The region is sheltered from the rain-bearing southwesterlies of the summer monsoon by the Western Ghats (orographic barrier along the west coast of India). Therefore, there is not much rain over this region during the summer monsoon. Temperature, pressure, and wind patterns change from summer to winter. The temperature is lower, the pressure decreases from north to south and is generally higher than in summer over central India. Airflow reverses and the wind blows northeasterly across the Indian Ocean, therefore, the season is called northeast monsoon (NEM). Clear skies, fine weather, light winds, low humidity, and large daytime variations of temperature are the normal features of the winter season in India. This is the driest season for the country as a whole, except in the extreme southeast of Peninsular India where the NEM brings heavy rainfall.

The Indian northeast monsoon rainfall (NEMR) displays considerable interannual variability (IAV), significantly affecting the agricultural activity in the region (Kumar et al., 2007). Compared to the Indian summer monsoon (ISM), relatively very little work has been done to explore the NEMR variability and its teleconnections (Dhar and Rakhecha, 1983; De and Mukhopadhyay, 1999; Singh and Sontakke, 1999; Raj, 2003; Kripalani and Kumar, 2004; Kumar et al., 2007). The association of Indian climate and El Niño/Southern Oscillation (ENSO) is well known. The IAV of ISM is anti-correlated with ENSO (Yadav, 2009a, 2009b), while the northwest India winter precipitation (NWIWP) is positively correlated (Dimri, 2005, 2006; Yadav et al., 2009a, 2009b, 2010). Also, the IAV of NEMR is well known to be significantly influenced by the positive phase of ENSO (De and Mukhopadhyay, 1999; Kumar et al., 2007). The ISM relationship with ENSO has decreased (Krishna Kumar et al., 1999; Yadav, 2009a, 2009b) while, the NWIWP and NEMR relationship has increased (Kumar et al., 2007; Yadav et al., 2009a, 2009b, 2010), in the recent decades.

Earlier studies by Kripalani and Kumar (2004) and Kumar et al. (2007) using 130 years of data have shown that the positive phase of Indian Ocean Dipole Mode (IODM) and the warm episodes of ENSO, respectively, are associated with the large positive rainfall anomalies over peninsular India. Kumar et al. (2007) have focused the homogeneous relationship between NEMR and ENSO, and the associated changes in the circulation regimes. They observed the recent strengthening of relationship between NEMR and ENSO. They studied the cyclones track occurring over the Bay of Bengal (BoB) and Arabian Sea. They found that, in the recent period with the strengthening ENSO relationship, most of the cyclones formed in the BoB have moved westwards/northwestwards and crossed the east coast of the south peninsular India region. In contrast, in the earlier period, double the amount of cyclones formed in the BoB. Only few cyclones have crossed the south peninsular India, and some of them have recurved to move northeastwards. Kripalani and Kumar (2004) described the moisture convergence and divergence during the positive and negative phases of IODM, and hence flood and drought NEMR years, respectively. However, both the studies have failed to produce the proper dynamical reason for the changes in the circulation regimes.

This paper focuses on the physical mechanisms of the homogeneous relationship between NEMR and ENSO, and the associated changes in the circulation regimes and the tracks of tropical cyclones during the warm and cold episodes of ENSO. The data used in the present study and the analyses procedures are briefly described in Section 2. The results pertaining to the characteristic features of impact of ENSO on NEMR and the associated changes in circulation patterns are discussed in Section 3. A physical mechanism proposed is provided in Section 4, and the main summary and discussion is in Section 5.

2. Data and methodology

An updated version of the high-resolution India meteorological department (IMD) 1° × 1° latitude/longitude gridded rainfall data prepared by Rajeevan et al. (2008) was used (hereafter: IMD gridded data). The original dataset was developed by Rajeevan et al. (2006) for the period 1951–2003 using 1803 stations. Later, the grid-point values for the period 1951–2007 were recalculated by Rajeevan et al. (2008) by using the same technique as used earlier (Rajeevan et al., 2006), but by using 2140 stations. The simultaneous correlation for the period 1971–2000 was calculated between Niño-3.4 index (representing the averaged SST anomalies over the equatorial central Pacific box (6°S–6°N, 170°W–120°W), commonly referred to as Niño-3.4 index, has been extracted from ERSST data) and IMD gridded data for the season October through December (OND). The result is shown in Figure 1. The correlation pattern shows positive anomalies over peninsular India and northwest India, and negative anomaly over northeast (NE) India.

Figure 1.

Simultaneous correlation between IMD 1 × 1 gridded data and Niño-3.4 index for the season OND (period: 1971–2000). The thick black contour is the 0 correlation line. The light (dark) shaded areas are positive (negative) correlations. Negative (positive) anomaly is indicated by thin white dotted (continuous) contours (mm)

Monthly rainfall data, for the meteorological subdivisions of peninsular India, for the period 1949–2008 (60 years), have been obtained from the Indian Institute of Tropical Meteorology (IITM) rainfall records (Parthasarathy et al., 1995; available at www.tropmet.res. in). The subdivisional averages were computed based on station data acquired from the IMD. The present study, however, considers rainfall variations only during the Indian northeast monsoon (NEM) season i.e. October through December. Time series of the mean northeast monsoon rainfall (NEMR) for the country has been prepared for the largest possible spatially coherent area comprising six meteorological subdivisions of peninsular India (Figure 2(a)), showing positive correlation with Niño-3.4 index (Figure 1). Also, these subdivisions have been identified based on similarity in precipitation characteristics and associations with regional/global circulation parameters (Parthasarathy et al., 1993).

Figure 2.

(a) Geographical pattern for different meteorological subdivisions of India. Shaded regions are 6 meteorological subdivisions of the northeast monsoon (NEM) region. (b) The time-series of northeast monsoon rainfall (NEMR) for the period 1949–2008 expressed as the rainfall anomaly of area-weighted seasonal rainfall from the long period normal (1949–2008). The horizontal dash lines are + 1 and − 1 standard deviation (SD). The straight line is the trend line and the curve is the 11-year running mean. (c) 21-year sliding correlations of NEMR with Niño-3.4 index. The horizontal dash line is 95% significant level

To examine the observed behaviour of anomalous NEM circulation over India, the monthly mean NCEP/NCAR global atmospheric reanalysis dataset (Kalnay et al., 1996) for the period 1949–2008 have been used. The sea surface temperature (SST) dataset used is the monthly extended reconstructed sea surface temperature (ERSST version 3) dataset produced by National Climate Data Center (NCDC) with a resolution of 2° × 2° grids (Smith et al., 2007; Xue et al., 2003). The interpolated outgoing longwave radiation (OLR) data have been provided by the NOAA-CIRES Climate Diagnostics Center, Boulder, Colorado, USA. The data are on a 2.5° × 2.5° grid and cover the period from June 1974 to December 2008. Here, due to nonavailability of data from March to December 1978, we have used data for the period 1979–2008. Daily OLR data have been used to study the westward movement of the Rossby-gyres, from the NOAA NCEP CPC (Liebmann and Smith, 1996). The data for the track of tropical cyclones over north Indian Ocean has been taken from ‘Tropical cyclone best track data site'maintained by Joint Typhoon Warning Centre (JTWC), USA.

For understanding the dominant modes of variability, simultaneous correlations, regressions and composite anomalies patterns of SST, MSLP, OLR, 200-hPa velocity potential, 200- and 850-hPa wind fields, and omega (vertical velocity) and zonal wind at 1000- to 100-hPa levels have been examined and their statistical significance estimated.

3. Results and discussion

3.1. Interannual variability (IAV) of northeast monsoon rainfall (NEMR)

The Indian northeast monsoon rainfall (NEMR) time series have been prepared as the area-weighted seasonal (OND) rainfall over 6 meteorological sub-divisions of peninsular India (Figure 2(a)) for the period 1949–2008, expressed as the rainfall anomaly from the long period normal (1949–2008) in Figure 2(b). The time series shows significant year-to-year variation. In this study, NEMR is considered to be normal if the rainfall anomaly is within + /− 1 standard deviation (SD). The flood and drought years have been defined if the rainfall anomalies are greater than + 1 SD and less than − 1 SD, respectively. There are 10 flood years (1956, 1966, 1969, 1972, 1977, 1987, 1993, 1994, 1997, and 2005) and 9 drought years (1949, 1950, 1951, 1965, 1974, 1984, 1988, 1989, and 2000) in the 60-year period from 1949–2008. The average rainfall is 345.1 mm with a standard deviation (SD) of 83.6 mm and a coefficient of variation of 24.2%. The average rainfall is less than and CV is greater than the all-India summer monsoon rainfall (AISMR) series (i.e. for AISMR, Mean = 850 mm, S.D. = 85 mm and CV = 10%). The trend line and 11-year running mean curve are superimposed on the NEMR series to show the increase in mean and IAV of NEMR in the recent decades. The 11-year running mean curve shows considerable low and high from the mean value in the recent decades than the earlier decades. The trend line shows the rate of increase of 1.05 mm of rainfall per year. Decadal changes in the precipitation series have also been noted for northwest India winter season (Yadav et al., 2009a, 2009b, 2010).

3.2. Homogeneous relationships between ENSO and NEMR

ENSO is the dominant mode of climate variability on instrumental records. Typically, El Niño develops during boreal summer (JJA), peaks during early winter (NDJ), and decays the following spring (MAM). Therefore, the analysis has been started with Niño-3.4 index. The simultaneous correlation of the Niño-3.4 index with the NEMR series, for the period 1949–2008 (60 years), is 0.3, which is statistically significant at the 98% significance level. To investigate the homogeneous relationship, sliding correlations on a 21-year moving window between Niño-3.4 and NEMR have been computed. The effective degree of freedom (EDOF) has been calculated for all central years (Figure 1(c)). The sliding correlations indicate strengthening of relationship in the recent decades after 1978 of the central year. Also, significant long-term changes (climate shift) in the distribution of tropical Pacific and Indian Oceans SST around the mid-1970s have been noted (Nitta and Yamada, 1989; Graham, 1994; Trenberth and Hurrell, 1994). Therefore, there could be some connection with the recent changes in the ENSO–NEMR relationships and the mid-1970s climate shift. As the ENSO starts developing in summer and gets mature in the month of December, it has strong predictive relationship a season in advance (i.e. JJAS ENSO) (Kumar et al., 2007). The Indian Ocean dipole (IOD) (Saji et al., 1999), which is in phase, most of the years, with ENSO, develops rapidly in boreal summer and matures in October. Kripalani and Kumar (2004) have found simultaneous positive relationship between IOD and NEMR.

3.3. NEMR teleconnections pre- and post-1978

In the previous subsections, we have seen that the relationship between ENSO and NEMR have increased after 1978. Therefore, to examine the differences in circulation features associated with the homogeneous relationship prior to 1978 and after that, two equal periods of concurrent regressions for the time slices 1949–1978 and 1979–2008 have been constructed, hereafter referred to as Period 1 and Period 2, respectively. The distinctive features in the regression patterns of different parameters relevant to the NEMR variability in these two periods are expected to provide further insight into the homogeneous changes observed in the relationships. Only areas with regression patterns significant at the 95% confidence level are shown and discussed.

Figure 3(a) and (b) show the regression of SST, mean sea level pressure (MSLP) and 850-hPa wind for the Period 1 and Period 2 onto the NEMR, respectively. The SST is shown as shaded and thin white contours and MSLP pattern is shown as thick black contours. The regressed anomalies of wind pattern are shown as arrows. During Period 1 (Figure 3(a)), there is no spatially wide significant anomaly observed over tropical SST. MSLP shows significant negative anomaly over the north Pacific and significant positive anomalies over west USA, NE India, and eastern BoB extending NE up to the South China Sea. The 850-hPa wind patterns show significant cyclonic circulation over the north Pacific, easterlies/northeasterlies over BoB. The high-pressure anomalies over NE India and Thailand associated with easterlies/northeasterlies anomalies over BoB intensifies the climatological background northeasterlies (trade winds), which converges moisture over southern peninsular India and hence flood NEM years and vice versa for drought NEM years, as reported by Doraiswamy (1946), Rao Krishna and Jagannathan (1953), Rao (1963), Rao and Raghavendra (1971), Ramaswamy (1972), Srinivasan and Ramamurthy (1973), Dhar and Rakhecha (1983), Krishnan (1984), Raj and Jamadar (1990), Sridharan and Muthusamy (1990), and Singh and Sontakke (1999). In Period 2 (Figure 3(b)), the SST shows significant positive anomalies over tropical central and eastern Pacific and negative anomalies over north, south, and tropical west Pacific (horseshoe pattern). MSLP shows significant positive anomalies over a wide region of tropical west and northwest Pacific, and west of Australia, and significant negative anomaly over tropical northeast Pacific. No significant high-pressure anomaly is observed over NE India or Thailand as observed during Period 1. The significant positive and negative anomalies over tropical west and east Pacific, respectively, resembles the negative phase of southern oscillation (SO). The 850-hPa wind pattern shows a significant westerly anomaly over the tropical Pacific. The warm SST and westerly wind anomalies over the tropical central Pacific resembles positive phase of ENSO (El Niño) phenomenon.

Figure 3.

Regression of SST, MSLP, and 850-hPa wind onto NEMR for (a) Period 1 (1949–1978) and (b) Period 2 (1979–2008). Areas of 95% significance level for SSTs are shaded and thin white contours ( °C), MSLP are thick black contours (hPa), and winds are arrows (ms−1). Negative (positive) anomaly is indicated by dotted (continuous) contours. The thin white contours interval is 0.1

During Period 1, the IAV of NEMR over India was associated with the variation of high surface pressure anomaly over Thailand associated with northeasterlies/easterlies winds over BoB. During Period 2, the remote forcing from ENSO was responsible for the IAV of NEMR. It is interesting to note that in Figure 3(b), the SST anomalies over the equatorial eastern Pacific do not show significant positive correlation. The asymmetric transition processes between El Niño and La Niña events, does not clearly define the SST anomalies over the equatorial eastern Pacific during their mature phase. The atmospheric response for the El Niño causes a rapid reduction of the equatorial Pacific westerlies in winter, which play a role in hastening the following El Niño to La Niña transition through the generation of upwelling oceanic Kelvin waves. The turnabout of the thermocline anomalies is recognized in its mature phase. However, the anomalous equatorial Pacific easterlies during La Niña persist to the subsequent spring, which tends to counteract the turnabout from the La Niña to El Niño (Ohba and Ueda, 2009).

3.4. ENSO response pre- and post-1978

In the previous subsections, we have seen the increasing relationship of NEMR with ENSO. To assess why ENSO is influencing NEMR variability in the recent decades, a similar regression analysis for Period 1 (1949–1978) and Period 2 (1979–2008) with the Niño-3.4 index have been carried out. Regressions significant at the 95% level are only shown (Figures 4,5 and 6). Figure 4(a) and (b) show the regression of SST, MSLP and 850-hPa wind onto Niño-3.4 index for Period 1 and Period 2, respectively. The SST pattern is shown as shaded, and thin white contours and MSLP pattern is shown as thick black contours. The regressed anomalies of wind pattern are shown as arrows. The basic features are the same for both the cases: Warm SST anomalies over tropical Indian Ocean, and central and eastern Pacific and cold SST anomalies over north, south, and western Pacific. High- and low-pressure anomalies over the warm-pool of Indonesia (Wang and Xie, 1998) region and eastern Pacific, respectively. Easterlies (westerlies) 850-hPa wind anomalies over tropical Indian Ocean (central Pacific).

Figure 4.

Same as Figure 3 but for Niño-3.4 index. The thin white contours interval is 0.03

Figure 5.

Regression of 200-hPa velocity potential and wind, and OLR onto Niño-3.4 index for (a) Period 1 (1949–1978), and (b) Period 2 (1979–2008). Area of 95% significance level for OLR are shaded and thin white contours (Wm−2), 200-hPa velocity potential are thick black contours (m2s−1), and winds are arrows (ms−1). Negative (positive) anomaly is indicated by dotted (continuous) contours. The thin white contours interval is 0.03

Figure 6.

Regression of omega (vertical velocity) and zonal wind at pressure levels 1000- to 100-hPa onto Niño-3.4 index to show the anomalous Walker circulation represented by zonal wind and omega (multiplied by − 1.5) averaged between 5°S–5°N for (a) Period 1 and (b) Period 2. The omega (multiplied by − 1.5) is shown as shaded and thin white contours, and the anomalous Walker circulation as arrows. The thin white contours interval is 0.04

In Figure 4, the significant differences between Period 2 and Period 1 are as follows: a larger extent of positive MSLP anomalies over BoB and Indian subcontinent, the area of positive SST anomaly over the northern Indian Ocean has reduced to north of the equator, the easterly anomalies over the tropical Indian Ocean have increased, the tropical and north subtropical Pacific cold SST anomalies are wider and the area of significant negative MSLP anomaly over eastern Pacific have increased in Period 2 compared to Period 1. Also, the zonal wind anomalies are stronger over the equatorial Indo-Pacific Oceans, suggesting stronger Walker circulation. The wind shows stronger anticyclonic circulation anomaly over BoB in Period 1 compared to Period 2. Annamalai et al. (2005) have noted similar basinwide SST warming over Indian Ocean during the developing phase of most El Niños prior to the 1976–1977 climate shift, whereas after the shift they had an east–west asymmetry—a consequence of El Niño being associated with the IOD/Zonal mode.

Similarly, Figure 5(a) and (b) show the regression of 200-hPa velocity potential and wind for Period 1 and Period 2 onto the Niño-3.4 index, respectively. The velocity potential and wind patterns are shown as thick black contours and arrows, respectively. The shaded and thin white contour regions in Figure 5(b) is the regression of OLR onto Niño-3.4 index, which is not shown in Figure 5(a) due to nonavailability of data during that period. Both the cases shows the similar features for 200-hPa velocity potential and wind: the upper-level divergence (velocity potential minimum) anomaly in the eastern Pacific and upper-level convergence (velocity potential maximum) anomaly in the warm-pool region, the twin anticyclonic circulation anomalies along the equatorial central Pacific, the easterly wind anomaly over equatorial central Pacific, and westerly wind anomalies over 20°N and 20°S between longitude 40° to 100°E. The twin anticyclonic circulation anomalies along the equatorial central Pacific are due to warming of the central and eastern equatorial Pacific (Figure 4), which moves atmospheric convection eastward, during the positive phase of ENSO (El Niño) phenomenon. The westerly and easterly anomalies over equatorial eastern Indian Ocean and western Pacific, respectively, suggest the convergence anomaly over the western Pacific.

In Figure 5, the significant differences between Period 2 and Period 1 are as follows: a significant negative velocity potential anomaly over western equatorial Indian Ocean during Period 2, which is not present in Period 1. The positive and negative velocity potential anomalies over the warm-pool and eastern Pacific, respectively, are much stronger and widespread, and the westerly and easterly anomalies over equatorial eastern Indian Ocean and western Pacific, respectively, are much stronger in Period 2 compared to Period 1. This suggests that the anomalous Walker circulation in ENSO events were much stronger during Period 2 than in Period 1. Also, there was an ascending motion anomaly over the NEMR region and equatorial western Indian Ocean during Period 2 El Niño years, which did not exist during Period 1 El Niño years. The OLR anomalies in Figure 5(b), shows increase in convection over equatorial central Pacific and western Indian Ocean, and decrease in convection over the warm-pool region. The increased convection over the equatorial central Pacific and western Indian Ocean are consistent with the warm SST anomalies over there (Figure 4(b)).

Further, to show the anomalous Walker circulation, the regression of omega (vertical velocity) and zonal wind at pressure levels 1000- to 100-hPa for Period 1 and Period 2 onto the Niño-3.4 index have been calculated and the height–longitude cross-section averaged over the latitudes 5°S–5°N are shown in Figure 6(a) and (b), respectively. The omega (multiplied by − 1.5) is shown as shaded and thin white contours, and the anomalous Walker circulation represented by zonal wind and omega (multiplied by − 1.5) are shown as arrows. Both the cases show similar features: the anomalous Walker circulation with upward motion anomalies between longitudes 180°W–120°W and 40°E–70°E and downward motion anomaly between longitudes 100°E–150°E. The significant differences between Period 2 and Period 1 are as follows: The upward and downward motion anomalies over central Pacific and the warm-pool region, respectively, are much stronger and widespread in Period 2 compared to Period 1.

The widespread subsidence anomaly over the warm-pool region is due to cold SST anomaly over the western Pacific (Figure 4(b)) and the compensatory subsidence due to eastward shifted strong convection over central Pacific during the positive phase of ENSO. The lower level easterly anomaly over equatorial Indian Ocean, westerly anomaly over Pacific and widespread MSLP anomaly over the warm-pool region are due to the subsidence anomaly over the warm-pool region (Figure 4).

3.5. Composite anomalies during pre- and post-1978

To understand the circulation features associated with the correlation and regression patterns, composite plots have been constructed. The distinctive features in the composite plots take into account the physical realism and represent configurations of the variable that are comparable to observations. Therefore, as an indication of the relationship between observational patterns and the coupled circulation features, composites are constructed directly from a number of years with the largest departure of rainfall and Niño-3.4 index from the mean values (i.e. flood and drought for NEMR, and El Niño and La Niña for ENSO) because they indicate situations in which the corresponding circulation features are dominant. The composite anomalies are averages of the anomalies of flood and drought years for NEMR and El Niño and La Niña years for ENSO. The anomalies are derived by compositing fields that exceed 1 standard deviation (SD) departures from a seasonal mean climatology for the Period 1 (1949–1978) and 2 (1979–2008). The years selected as floods (above + 1 SD) and droughts (below − 1 SD) for NEMR and as El Niño and La Niña years for ENSO are shown in Table I.

Table I. List of flood and drought years of the northeast monsoon and El Niño and La Niña years
Northeast Monsoon (OND)ENSO (OND)
Flood (above + 1 SD)Drought (below − 1 SD)El NiñoLa Niña
YearRainfall anomaly (mm)YearRainfall anomaly (mm)YearTemp. anomaly ( °C)YearTemp. anomaly ( °C)
1956128.421949− 115.9819571.051949− 1.13
1966143.921950− 86.4819630.921955− 1.84
1969123.821951− 105.8819651.441973− 1.7
197288.321965− 90.6819721.811975− 1.51
1977171.221974− 125.4819821.911984− 0.75
1987132.121984− 94.2819871.221988− 1.6
1993162.221988− 197.1819911.231998− 0.99
1994106.821989− 138.5819941.131999− 1.14
1997115.621995− 81.2819972.242007− 0.91
2005189.222000− 106.1820021.26  

Figure 7(a) and (b) show the composite anomalies of SST and 850-hPa wind patterns for the flood and drought years, respectively, for Period 1 (upper two panels) and, Figure 7(c) and (d) for Period 2 (lower two panels). The SSTs are shown by shaded and thin white contours, and winds as arrows. The thick black contours in Figure 7(c) and (d) are the composite anomalies of OLR during Period 2 flood and drought years, respectively, which are not shown in Figure 7(a) and (b) due to nonavailability of data during that period. The basic features for the flood and drought years composite anomalies are: during flood years, winds show easterly anomalies over central and eastern equatorial Indian Ocean and BoB. There is a divergence anomaly over the western Pacific. The SSTs are warmer over central and eastern equatorial Pacific. For drought years, the SSTs show negative anomalies over central and eastern equatorial Pacific. The difference between Period 1 and Period 2 are: for the flood years, divergence anomaly over western Pacific, equatorial westerlies over the central Pacific, easterlies over south BoB, and positive SST anomalies over tropical Indo-Pacific Oceans are much stronger during Period 2 compared to Period 1. For the drought years, negative SST anomalies over central and eastern equatorial Pacific, positive SST anomalies over tropical Indian Ocean and western Pacific, and convergence anomaly over western Pacific are much stronger during Period 2 compared to Period 1. The winds show westerly anomalies over south BoB during Period 1 drought years. This suggests that the stronger easterlies over BoB are favourable for flood years of NEMR. The OLR anomaly during Period 2 flood years shows reduced convection over the warm-pool, and increased convection over peninsular India and central Pacific and vice versa during drought years.

Figure 7.

Composite anomalies of SST ( °C) and 850-hPa wind (m/s) during Period 1 (a) flood and (b) drought years, and Period 2 (c) flood and (d) drought years. The shaded and thin white contours are SSTs and arrows are winds. The thick black contours in (c) and (d) are the composite anomalies of OLR (Wm−2) during Period 2 flood and drought years, respectively. Wind greater than 1 m/s are plotted. The thin white contours interval is 0.2

Figure 8(a) and (b) shows similar composite anomalies for El Niño and La Niña years, respectively, for Period 1 (upper two panels) and, Figure 8(c) and (d) for Period 2 (lower two panels). The basic features for the El Niño and La Niña years composite remains the same: warm SST anomalies over central and eastern equatorial Pacific, divergence anomaly over western Pacific, and easterly and westerly anomalies over equatorial Indian Ocean and western Pacific, respectively, for El Niño years, and vice versa for La Niña years. The differences are: during El Niño years, warm SST anomalies over equatorial Indian Ocean and central and eastern Pacific are much stronger during Period 2 compared to Period 1. The wind anomalies are easterly over equatorial eastern Pacific during Period 1 and westerly during Period 2. The easterly wind anomaly over the Indian Ocean is shifted more northward towards south BoB during Period 2 compared to Period 1, which is favourable for NEMR (Figure 7(a) and (c)). For La Niña years, cold SST anomalies over central and eastern equatorial Pacific, warm SST anomalies over tropical India Ocean and the warm-pool region are much stronger during Period 2 compared to Period 1. Figure 8(c) and (d) demonstrate one of the most remarkable features is the difference in spatial structure. The location of the maximum SST anomaly during La Niña (Figure 8(d)) is shifted to the west by about 30° in the central east Pacific relative to that of El Niño (Figure 8(c)). Similar patterns have been observed by Ohba and Ueda 2009. This may also be the reason for no significant positive correlation over equatorial eastern Pacific in Figure 3(b). The OLR anomaly shows increased convection over equatorial central Pacific and tropical western Indian Oceans, and decreased convection over the warm-pool region during El Niño years, and vice versa during La Niña years. Both the composites of El Niño and La Niña years shows increase in SST all over tropical Indo-Pacific Oceans during Period 2 compared to Period 1. It is interesting to note that during Period 2 El Niño years, the SSTs over tropical Indian Ocean were much warmer than all other cases, and during Period 2 La Niña years over the warm-pool region. During El Niño years (Figure 8(c)) the regions of deep convection were elongated over equatorial Indian Ocean and central Pacific (OLR less than 220 Wm−2 splits and centred over tropical eastern Indian Ocean, west and central Pacific) while they were contracted over Indonesian region (Figure 8(d)) during La Niña years. The warm SST forcing favours deep convection, which modifies wind. The latent heat release due to deep convection is the tropospheric heat source of the tropics.

Figure 8.

Composite anomalies of SST ( °C) and 850-hPa wind (m/s) during Period 1 (a) El Niño and (b) La Niña years, and Period 2 (c) El Niño and (d) La Niña years. The shaded and thin white contours are SSTs and arrows are winds. The thick black contours in (c) and (d) are the composite anomalies of OLR (Wm−2) during Period 2 El Niño and La Niña years, respectively. Wind greater than 1 m/s are plotted. The thin white contours interval is 0.2

To show the anomalous Walker circulation, the composite anomalies of omega (vertical velocity) and zonal wind at pressure levels 1000- to 100-hPa for El Niño and La Niña years for Period 1, and the composite differences between Period 2 and 1 for El Niño and La Niña years are calculated and height–longitude cross-section averaged over the latitudes 5°S–5°N are shown in Figure 9(a)–(d), respectively. The omega (scaled by − 0.003) and the Walker circulation anomaly represented by zonal wind and omega (scaled by − 0.003) are shown as shaded and thin white contours, and arrows, respectively, in Figure 9(a) and (b). Similarly, in Figure 9(c) and (d) the omega (scaled by − 0.001), and the Walker circulation difference represented by zonal wind and omega (scaled by − 0.001) are shown as shaded and thin white contours, and arrows, respectively. The El Niño year composite anomaly for Period 1 shows (Figure 9(a)) reversal of Walker circulation: upward motion anomaly over central Pacific (120°W–180°W) and downward motion anomaly over the warm-pool region (120°E–160°E). Upward motion anomaly is also observed over 50°E. While, La Niña year composite anomaly for the Period 1 shows (Figure 9(b)) strong Walker circulation: strong upward motion over the warm-pool region (120°E–150°E) and strong downward motion over central Pacific (130°W–180°W). Downward motion anomaly is observed over eastern Indian Ocean (50°E–105°E). The difference of composites between Period 2 and 1 El Niño years shows (Figure 9(c)) downward motion anomalies over central Pacific (120°W–180°W) and the warm-pool region (120°E–150°E) and upward motion anomaly over central Indian Ocean (60°E–90°E). Thus, during Period 2, the downward motion anomaly over the warm-pool has increased and the upward motion anomaly over central Pacific has decreased. The upward motion anomaly over the Indian Ocean has increased and shifted to the central Indian Ocean, which is consistent with the warm SST anomaly over there (Figure 8(c)). While for La Niña years (Figure 9(d)), downward motion anomalies are seen over the western Pacific (120°W–150°W) and central Pacific (140°W–180°W) and upward motion anomaly over Indian Ocean (60°E–110°E). This suggests stronger Walker circulation during Period 2 compared to Period 1: stronger downward motion over central pacific and widespread upward motion over the western pacific and eastern Indian Oceans. The strong upward motion is consistent with the positive SST anomaly over there (Figure 8(d)).

Figure 9.

Composite anomalies of omega (vertical velocity) and zonal wind at pressure levels 1000- to 100-hPa to show the anomalous Walker circulation represented by zonal wind and omega (scaled by − 0.003) averaged between 5°S and 5°N for Period 1: (a) El Niño, and (b) La Niña years. Difference of Period 2 and Period 1 composites of (c) El Niño, and (d) La Niña years (omega scaled by − 0.001). The omega is shown as shaded and thin white contours, and the Walker circulation as arrows. The thin white contours interval for (a) and (b) is 2, and for (c) and (d) is 4

Gill (1980) in an elegant analytical study on heat-induced equatorial circulations showed that rising motion occurs directly above the heat source with easterlies to the east of it and a smaller region of westerlies to the west. He interpreted this pattern in terms of equatorially trapped Kelvin and Rossby waves. When the heating is switched on at an initial time, Kelvin waves travel eastward creating easterlies to the east of the heating. Rossby waves that have a phase speed approximately one third of that of Kelvin wave speed travel westward. Because of the lower Rossby wave speed, the region of westerlies to the west is more limited. Thus, easterlies to the east and westerlies to the west in the lower levels strengthened the easterlies and westerlies over equatorial western Pacific and eastern Indian Ocean, respectively, during strong La Niña years as the heat source concentrates over the warm-pool region. The strengthening of the westerlies over equatorial Indian Ocean impeded the easterlies over south BoB. During strong El Niño years, as the heat sources are split and centre over tropical Indian Ocean and central Pacific, the tropical easterlies and westerlies relax over tropical central Pacific and eastern Indian Oceans, respectively. The convergence to the west of the heat source and warmer SST over north Indian Ocean are favourable for the intensification of easterlies over south BoB.

3.6. Tropical cyclones (TCs) tracks

The maximum number of tropical cyclones (TCs) and storms are formed in the north Indian Ocean during the season (OND). These tropical storms and TCs strike the NEM region and considerably impact the IAV of NEMR. So, it is important to study the track of TCs. In Figure 10, the TCs tracks have been plotted over north Indian Ocean for the El Niño (Figure 10(a) and (c)), and La Niña (Figure 10(b) and (d)) years during Period 1 (upper panels) and Period 2 (lower panels). In all the cases, the TCs are formed near the equator and move in a west-northwest direction. After crossing 15°N most of the TCs recurve towards the northeast direction, under the influence of mid-latitude westerlies. The basic differences in these cases are as follows: During Period 1 El Niño years (Figure 10(a)), almost half the TCs formed over BoB had struck southern peninsular India (NEM region), and half of them recurved towards the head of BoB. It is interesting to note that, among all the cases, the largest numbers of TCs were formed in the Arabian Sea. During Period 1 La Niña years (Figure 10(b)), there was a maximum number of TCs formed over BoB. They had struck all along the eastern coast of the Indian peninsula and Bangladesh. This means that, during Period 1, the TCs did not have preferred track, and had struck all along the eastern coast of peninsular India. Therefore, the influence of tropical storms and TCs on the IAV of NEMR was weak during the ENSO years. The anticyclonic circulation and high MSLP anomalies over BoB (Figure 4(a)), were responsible for less number of cyclone formations during El Niño years and vice versa for La Niña years. Also, warm SST anomalies and no significant MSLP anomaly over Arabian Sea during Period 1 El Niño years were favourable for more cyclone formations over there.

Figure 10.

Tropical Cyclones Tracks over north Indian Ocean for Period 1 (a) El Niño, and (b) La Niña years, and for Period 2 (c) El Niño, and (d) La Niña years. The shaded areas are the NEM region

During Period 2 El Niño years (Figure 10(c)), the maximum numbers of TCs formed over BoB had struck southern peninsular India (i.e. NEM region). While during Period 2 La Niña years (Figure 10(d)), the maximum TCs formed over BoB had struck the northern part of peninsular India and towards the head of BoB, missing the NEM region. Also, the maximum numbers of TCs were originated more deep towards southeastern part of BoB, among all the cases. The reason for this is the centre of deep convection situated in the Indonesian region (Figure 8(d)). The deep convective and stratiform heating drive lower troposphere westerlies, and the trailing Rossby-gyres formed along the equator, migrates pole-ward. The intensification of trailing Rossby-gyres in the tropical baroclinic atmosphere is the fundamental in the formation of TC away from the equator in the warm SST. Since the deep convection is centred over the Indonesian region, the trailing Rossby-gyres to the west of the deep convection intensified as TCs over warm SST southeast of BoB (Figure 8(d)). These TCs had to travel a longer distance towards a west-northwestward direction over BoB and missed the southern peninsular (NEM region) and strike northern peninsular India. Some of them recurved and moved towards the head of BoB. While during Period 2 El Niño years, the tropical Indian Ocean was much warmer (Figure 8(c)), the deep convections were centred along equatorial Indian Ocean. The Rossby-gyres intensified as TCs had to travel a shorter distance towards west-northwestwards and strike the southern peninsular India (NEM region). This suggest out-of-phase relationship between southern peninsular and northeast (NE) India rainfall pattern during the season (OND). Therefore, the correlation between Niño-3.4 and rainfall shows positive correlation over peninsular India and negative correlation over NE India (Figure 1). The strong easterly and warm SST anomalies over the northern equatorial Indian Ocean during El Niño years (Figure 4(b)), were also responsible for keeping the TCs track westward. This is the reason for the origin of more TCs over the Arabian Sea during El Niño years (Figure 10(c)) than La Niña years (Figure 10(d)).

During Period 1, the SSTs were warmer than 29 °C east of 130°E. Therefore, the deep convections were centred over the central and western pacific during El Niño and La Niña years, respectively, and the Rossby-gyres had to travel a longer distance to intensify as TCs over BoB. Moreover, during Period 2 El Niño years, the warm SST anomaly along the north of the equator had shifted the inter-tropical convergence zone (ITCZ) over there. The vortices formed near the ITCZ, intensified as tropical storms and TCs and moved along the ITCZ. This is the reason why the TCs moved in a more westerly direction and were recurving less towards the northeast direction during El Niño years than La Niña years, as observed by De and Mukhopadhyay (1999). During Period 1, the warmer SST anomaly over the north Indian Ocean during El Niño years and cyclonic circulation anomaly during La Niña years (Figure 4(a)) were responsible for formation of more TCs over north Indian Ocean compared to Period 2. Also, during Period 2, the high MSLP anomaly over BoB and Indian region during El Niño years and weak cyclonic circulation anomaly over BoB during La Niña years (Figure 4(b)), reduces the formation of TCs over the north Indian Ocean, as observed by Kumar et al., 2007.

3.7. Longitude–time cross-section of OLR

In the earlier subsection, we have seen that in recent decades the TCs were having preferred track and were affecting the IAV of the NEMR. The positions of the heat source affect the tracks of the TCs.

Figure 11 shows the longitude–time cross-section of the OLR averaged over the latitudes 7.5°N–12.5°N, depicts well organized westward movement of the deep convection away from the equator that formed due to intensification of the trailing Rossby-gyres. Six years from Period 2 have been selected, 3 flood years 1987, 1994, and 1997 with El Niño years (upper panel), and 3 drought years 1984, 1988, and 1995 (lower panel). The drought years, 1984 and 1988 were La Niña years. The year 1995 was not an ENSO year. OLR minima are surrogates for the persistent deep convection and latent and radiative heat sources within the troposphere. The shaded areas show OLR < 220 Wm−2, which is considered to indicate convection and precipitation in the tropics. In all the years, the maximum rainfall belts between latitudes 7.5°N–12.5°N were moving westward.

Figure 11.

Time–longitude section of the OLR (Wm−2) along 40°E–180°E averaged for the latitudes from 7.5°N–12.5°N, and for 1 October through 31 December. The years, (a) 1997, (b) 1994, and (c) 1987 are the flood years (upper panel), and (d) 1988, (e) 1984, and (f) 1995 are the drought years (lower panel). The vertical thin lines are along 80°E, 120°E, and 160°E longitudes. The 80°E longitude is the coast line of the eastern peninsular India

The NEM region of India lies in southern peninsular India between longitudes 75°E–80°E. So, the west-northwestward travelling tropical storms and TCs must reach up to 80°E to produce heavy rainfall over NEM region of India. In the upper panel, which were El Niño years and the tropospheric heat sources were centred along the equatorial Indian Ocean and central Pacific. In most cases, the deep convections have originated at longitudes 90°E and 170°E. The convections originated at 90°E have moved up to 70°E i.e. NEM region. In the lower panel year 1984, the deep convections have originated at 160°E and moved up to 110°E. During 1988, the deep convections have originated over 140°E and 100°E. The convections originated at 140°E have moved up to 100°E, and the convections originated at 100°E have moved up to 85°E. Both, 1984 and 1988 were La Niña years, and the tropospheric heat sources were centred over the warm-pool region. The convections moving west-northwestwards away from the equator (between latitudes 7.5°N–12.5°N) have missed NEM region. During 1995, which was not an ENSO year, the convections have originated at 130°E and moved up to 100°E, missing NEM region. In the early part of the season, the deep convections have originated at 170°E and moved up to 80°E have produced heavy rainfall over NEM region. This is the reason for the highest rainfall recorded among all the drought years (Table I). This suggest that the rainfall was confined east of 80°E during drought and La Niña years, but it can extend westward to around 70°E during flood and El Niño years. That is consistent with the westward extension of the high pressure during El Niño, and the eastward withdrawal in La Niña years (shown in Figure 4). Thus, the southerly wind and moisture convergence can (can not) cover NEM region, resulting flood (drought) years.

4. Proposed physical mechanisms

The proposed physical mechanism to explain the effect of ENSO on NEMR during post-1978 is as follows. During post-1978 (Period 2), the SST over equatorial Indo-Pacific Oceans has increased and also other characteristics, such as the position and shape of the ENSO SST anomaly, have changed. Krishna Kumar et al. (2006) have documented that the El Niño events with the warmest SST anomalies in the central equatorial Pacific are more effective in focusing drought-producing subsidence over India than events with the warmest SSTs in the eastern equatorial Pacific during Indian summer monsoon season (JJAS). Ashok et al. (2007) have observed events with increased SSTs in the central Pacific, sandwiched by anomalous cooling in the east and west, since the late 1970s. These are not like the conventional El Niño, with the warming of SSTs around the west coast of Peru rather than the Date Line. The decadal changes in ENSO properties have also been discussed in the scientific literature (e.g. Fedorov and Philander, 2000; Wang and An, 2002). In this study, we show that the recent changes in ENSO, which has caused the warming of SSTs around the warm-pool region (equatorial Indian Ocean) during La Niña (El Niño) years, are more effective in producing droughts (floods) over NEM region.

During pre-1978, the SSTs over the warm-pool region and equatorial Indian Ocean were not much warmer during ENSO years. The SSTs greater than 29 °C were centred over tropical central and western Pacific during El Niño and La Niña years, respectively. Therefore, the deep convections, which are sensitive to SST forcing, were not having stronger amplitude over the warm-pool region and equatorial Indian Ocean. However, the deep convections were having stronger amplitude over the western part of the western Pacific. The trailing Rossby-gyres with cyclonic circulation anomalies formed west of the deep convection had to travel a long distance from there to the northern Indian Ocean. Sometimes, the remnants of tropical disturbances from the western Pacific gets intensified into TCs over BoB and Arabian Sea. Therefore, the TCs did not have a preferred track over the north Indian Ocean and had no significant impact on IAV of NEMR. The NEMR was influenced by the intensification of local circulation i.e. intensification of the northeasterlies over BoB which converged moisture over southern peninsular India during flood years and vice versa for the drought years of NEMR. The intensification of the northeasterlies over BoB was associated with an increase of surface high pressure over Thailand.

During post-1978, the NEMR were associated with ‘remote forcing’ such as ENSO. During El Niño years, the SST greater than 28.5 °C were zonally elongated over the tropical Indian Ocean to the central Pacific. The deep convections were centred along the equatorial Indian Ocean. The northwestward propagating trailing Rossby-gyres intensified as tropical storms and TCs had to travel a much shorter distance over BoB, strike southern peninsular India, and thus, result in NEMR. Also, the warm SSTs north of the equator over the Indian Ocean pull the ITCZ over there, and vortices formed along the ITCZ get intensified as tropical storms and TCs strike southern peninsular India. While, during La Niña years, the SST greater than 28.5 °C were contracted to the western Pacific. The centre of deep convection was concentrated over the Indonesian region. The trailing Rossby-gyre intensified as tropical storms and TCs had travelled longer distance towards northwestward direction over BoB and missed southern peninsular India, and thus result in a sparse NEMR. Also, the westerlies over equatorial Indian Ocean, associated with deep convection over Indonesian region, does not allow the ITCZ to penetrate to lower latitudes. The vortices formed along the ITCZ intensified as tropical storms and TCs missed the NEM region, and hence, a poor NEMR.

5. Summary and discussion

It has been found that the IAV of Indian northeast monsoon rainfall (NEMR) has increased in the recent decades. After analyses of SST, OLR, and reanalysis data it has been confirmed that, in the earlier decades (1949–1978; Period 1), the NEMR was associated with the rise of surface pressure over Thailand which strengthens the northeasterlies (trade winds) over BoB. While, in the recent decades (1979–2008; Period 2), the NEMR was associated by remote forcing such as ENSO. The change of this relationship is likely due to the change in the tropical atmospheric response to ENSO. The regression of SST, MSLP, 200-hPa velocity potential, 200- and 850-hPa wind, and omega (vertical velocity) and zonal wind at 1000- to 100-hPa levels onto Niño-3.4 index for the Periods 1 and 2 shows strengthening of surface pressure anomaly over the south Asian region, lower level tropical easterlies over Indian Ocean, negative SST anomaly over the western Pacific, upper level subsidence anomaly over the warm-pool region, and stronger anomalous Walker circulation and Indian Ocean dipole mode (IODM) in Period 2 than in Period 1.

The composite anomalies of El Niño and La Niña years during Period 2 shows that during El Niño years the SSTs over the northern part of the equatorial Indian Ocean has increased while, during La Niña years the SSTs has increased over the equatorial eastern Indian Ocean and the warm-pool region. The composite of both gives rise to IODM. The association of IODM and NEMR has been proposed by Kripalani and Kumar (2004). Also, the composite anomalies of La Niña years shows westerlies over equatorial Indian Ocean and easterlies during El Niño years, resulting in the strengthening of equatorial easterlies during positive phase of ENSO. As proposed by Kumar et al. (2007), the strengthening of easterlies converges moisture over the southern peninsular, thus resulting in a good NEMR. However, this study shows that during recent decades of ENSO, the SSTs over the Indian Ocean and the warm-pool region have changed the position of centre of deep convection over equatorial Indo-Pacific Oceans, which have changed the track of tropical storms and TCs. The tracks of tropical storms and TCs have affected the IAV of NEMR.

In recent decades with El Niño years, the SST north of the equatorial Indian Ocean has increased. The increased SST has increased the deep convection over there. The trailing Rossby-gyres formed to the west of the deep convection along the equator got intensified into TCs and storms over the warm north Indian Ocean and strike southern peninsular India, thus resulting in a good NEMR. Also, the increased SST north of the equatorial Indian Ocean had shifted the ITCZ over there. The vortices formed along the ITCZ got intensified into tropical storms and TCs, and strike NEM region. While during La Niña years, the SST over the warm-pool region has increased. The increased SST has amplified the deep convection over the Indonesian region, rather than the Indian Ocean, as in the case of El Niño years. The trailing Rossby-gyres had to travel a longer distance in the northwestward direction over BoB and missed the southern peninsular India, thus resulting in a poor NEMR.

In the earlier studies, it is well noted that El Niño causes SST to increase over the tropical Indian Ocean (TIO) during boreal fall. Prior to the 1976–1977 climate shift (PRE76), SST warming in the TIO consist of basinwide, whereas after the shift (POST76) they had an east–west asymmetry—a consequence of El Niño being associated with the Indian Ocean Dipole/Zonal mode. An important difference between the epochs, as noted by Annamalai et al. (2005), is the formation (absence) of the South China Sea anticyclone during PRE76 (POST76). Using the 10-member AGCM ensemble simulations, separately for PRE76 and POST76 El Niño events, they concluded that tropical west Pacific convection and the subsequent development of the South China Sea anticyclone are sensitive to Indian Ocean SST anomalies. In the PRE76 case, the basinwide SST anomalies over the Indian Ocean force easterly wind anomalies over the near-equatorial western central Pacific, whereas in POST76 the east–west contrast in the SST anomalies weaken the easterly wind anomalies over the equatorial Pacific.

The El Niño-induced TIO warming also acts like a capacitor, anchoring atmospheric anomalies over the Indo-western Pacific Oceans that persists through summer following El Niño. It causes tropospheric temperature to increase by a moist-adiabatic adjustment in deep convection, emanating a baroclinic Kelvin wave into the Pacific. In the northwest Pacific, this equatorial Kelvin wave induces northeasterly surface wind anomalies, and the resultant divergence in the subtropics triggers suppressed convection and the anomalous anticyclone (Xie et al., 2009). The internal air–sea interaction within the TIO is the key to sustaining the TIO warming through summer. During El Niño, anticyclonic wind curl anomalies force a downwelling Rossby wave in the south TIO through Walker circulation adjustments, causing a sustained SST warming in the tropical southwest Indian Ocean (SWIO) where the mean thermocline is sallow. During the spring and early summer following El Niño, this SWIO warming sustains an antisymmetric pattern of atmospheric anomalies with northeasterly (northwesterly) wind anomalies north (south) of the equator (Du et al., 2009). The SWIO SST variations impact local precipitation anomalies. These variations also affect the Pacific-North American (PNA) circulation during boreal winter of El Niño years. Analysis of the circulation response shows that over the PNA region, the 500-hPa height anomalies, forced by Indian Ocean SST anomalies, oppose and destructively interfere with those forced by tropical Pacific SST anomalies (Annamalai et al., 2007). The results presented herein suggest that the TIO plays an active role in climate variability and accurate observation of SST there is of urgent need.

Acknowledgements

The author is grateful to the ‘Development of a Seasonal Prediction System for Indian Monsoon Program’ Group, Indian Institute of Tropical meteorology (IITM), for their constant encouragement and kind help. The data have been taken from Web sites. Computational and graphical analyses required for this study have been completed with the free software, like Red Hat Linux, Intel Fortran, GrADS, and xmGrace. The author sincerely thanks both the anonymous reviewers, whose comments and critiques improved the quality of the manuscript.

Ancillary