The Indian Summer Monsoon Rainfall (ISMR), which begins on 1 June and lasts for four months through 30 September, contains large spatial and temporal variabilities. Parthasarathy et al. (1994) used data from a network of 306 climatic rain gauge stations well distributed in India recorded from 1871 to the present to determine prominent inter-annual variability in the quantum of ISMR; this database is available on the Web site of the Indian Institute of Tropical Meteorology (IITM), Pune (www.tropmet.res.in). The long-term mean of ISMR is close to 85 cm, and its standard deviation is approximately 10%. Approximately 70% of the annual rainfall in India is attributed to the ISMR, the most prominent interannual variability of which occurs in tropospheric biennial oscillation (TBO; Meehl, 1997). Drought monsoons, which have one standard deviation less than the long-term mean of the ISMR, are followed by monsoons with normal or excessive rainfall. ISMR has a prominent decadal scale variability in the form of alternating 30-year epochs of dry and wet monsoons. Frequent drought monsoon years occur during dry epochs, at an average of once in every three years. In wet epochs, drought monsoon years are infrequent and generally occur once in every 10-15 years. This epochal nature of monsoon was reported by Joseph (1976, 1978), who found that in dry (wet) epochs, tropical cyclones of the Bay of Bengal had preferred northward (westward) movement, suggesting that equatorward intrusion of the mid-latitude westerly winds of the upper troposphere over south Asia had a role in the epochal nature of the monsoon. Figure 1a shows the standardized anomalies of ISMR recorded from 1871 to 2010 including both inter-annual and inter-decadal variabilities.

Figure 1

Figure 1 (a) Standardised anomaly of ISMR of the years 1871 to 2010 (data taken from the IITM website www.tropmet.res.in); (b) Detrended global scale SST gradient anomaly of SST difference of northern box (30° N-60° N, 120° E-360° E) minus equatorial box (20° S-10° N, 50° E-360° E). The red line is its 11-year moving average; (c) 200 hPa vector wind (JJAS) as the mean of the WET decade 1950 to 1959; (d) Mean wind (JJAS) at 200 hPa of the decade 1970 to 1979 minus the mean wind (JJAS) of the decade 1950 to 1959 (wind anomaly). Magnitudes of the wind in m s-1 are shown by the shaded contours in (c) and (d).

A main component of the Indian summer monsoon is the cross-equatorial low level jet stream (LLJ), with an axis of strongest wind occurring at the 850 hPa level (Joseph and Raman, 1966; Findlater, 1969). The mean LLJ of June to September for 1950-2010 is shown in Fig. 2a. The LLJ acts as a conduit to carry moisture generated by the trade winds over the southern Indian Ocean and the evaporative flux from the Arabian Sea to the areas of monsoon rainfall generation over south Asia, which includes India. In addition, the area of cyclonic vorticity in the atmospheric boundary layer just north of the LLJ axis is a dynamic forcing for the generation of upward motion of the moist monsoon air for the production of monsoon rainfall.

Figure 2

Figure 2 (a) Mean low level jetstream of June to September (1950-2010) at 850 hPa level with isotachs of the wind magnitude in m s-1 shaded and the vector wind in m s-1 marked by the arrows; (b) Mean HadISST anomaly of September to November as composite of five severe drought monsoons of 1965, 1972, 1979, 1982, and 1987; (c) Correlation between ISMR and mean OLR of June to September using data of 1979 to 2010 showing the two poles of convection (monsoon heat source) anomaly of Indian (A) and West Pacific (B) oceans.

The ISMR exhibits large interannual variability. Generally, a deficient monsoon year is followed by a normal or excessive monsoon year to produce biennial oscillation. The summer monsoons of India and Australia and the SSTs of the tropical Indian and western Pacific oceans have roles in TBO, as demonstrated in observational and modeling studies conducted by Meehl (1997), Chang and Li (2000). In TBO, the SST anomalies averaged from September to November in five severe Indian monsoon drought years of 1965, 1972, 1979, 1982, and 1987 in the 30-year epoch of 1961-90 are shown in Fig. 2b. Following one year or two consecutive years of monsoon drought, the ocean around India exhibits a warm SST anomaly that persists until the following monsoon, which is accompanied by normal or excessive rainfall. A comparison of Figs. 2a and 2b reveals that the warm anomaly occurs mainly near the LLJ. This anomaly is forced by a weaker than normal LLJ of the drought monsoon, which produces less upwelling and evaporation from the ocean, both of which are conducive to less cooling of the SST during monsoons, and leads to the warm SST anomaly occurring in post-monsoon months. The linear correlation coefficient between the magnitude of the 850 hPa LLJ winds from June to September and the ISMR for 1950 to 2010 varied from 0.6 to 0.8 in different portions of the LLJ (figure not shown). These high values are statistically significant. Less cooling results in a positive anomaly of SST after the monsoon season. During the same post-monsoon months, the western Pacific ocean exhibits a cold SST anomaly, which persists for one year. The months following monsoons giving excess rains show weaker SST anomalies of opposite signs. Conditions in India differ from those in Africa, where the Sahelian drought has occurred during 10-15 successive years.

Figures 3a and 3b show the June-September 200 hPa wind anomalies of the five severe drought years and five excessive monsoon rainfall years, respectively, which occurred during the 30-year epoch. Figure 3c shows the linear correlation between the June-September averaged meridional component of the 200 hPa wind and the ISMR of the epoch period. The large-scale wave shown in Figs. 3a-c, a Rossby wave train known as the Asia-Pacific Wave (APW; Joseph and Srinivasan, 1999), has a wavelength of approximately 60° longitude (wave No. 6). In the drought monsoon 200 hPa composite, a wave trough in the westerlies intrudes into northwestern India.

Figure 3

Figure 3 (a) Vector wind anomaly in m s-1 at 200 hPa level of DRY monsoon (JJAS) composite of 1965, 1972, 1979, 1982, and 1987; (b) Vector wind anomaly in m s-1 at 200 hPa level of WET monsoon (JJAS) composite of 1961, 1970, 1975, 1983, and 1988; (c) Linear correlation coefficient (LCC) between ISMR and 200 hPa meridional wind component of June to September of 1961 to 1990 showing the Asia-Pacific wave (wave number 6). In drought monsoons this wave has a trough over Northwest India and another trough close to Japanese Islands. DRY monsoons are associated with such a wave number-6 trough from the month of May and through the monsoon season. The LCC is large and statistically significant.

The next wave trough to the east near the Japanese Islands steers the tropical cyclones of the western Pacific Ocean northward. In the excess monsoon rain composite, a half-wavelength shift in the APW causes the cyclones of the western Pacific Ocean to move westward and generate monsoon depressions in the Bay of Bengal. This northern motion preference of tropical cyclones of the western Pacific Ocean during drought ISMR seasons has been discussed by Kumar and Krishnan (2005). The large annual variations in 200 hPa meridional wind over northwestern India in May associated with this wave, formed the basis for a long-range (seasonal) method for forecasting ISMR designed by Joseph et al. (1981). In addition, that with data of a longer period was developed by Parthasarathy et al. (1991).

ISMR showed only small long-term trends during the period of recorded rainfall measurements; however, a prominent decadal change was noted (Fig. 1a). During the three-decade long dry epochs of 1901-30 and 1961-90, India experienced drought monsoons, approximately once every three years. In contrast, during the three-decade long wet epochs of 1871-1900 and 1931-60, the frequency of droughts was approximately once every 10-15 years. Thus, during the 120 years of 1871 to 1990, regular 30-year alternating dry and wet epochs occurred. Sontake and Singh (1996) used the available network of a smaller number of rain gauge stations in India to determine ISMR which showed that 1841-70 was a dry epoch.

During the decades of the dry epoch in 1961-90, wind reanalysis data of the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) showed that subtropical westerlies of the upper troposphere moved to lower latitudes over southern Asia as a wave No. 3 trough (Kalnay et al., 1996). The equatorward intrusion of the 200 hPa westerlies during the dry decade of 1970-79 is shown in Fig. 1d as an anomaly from the wet decade of 1950-59. The mean wind and sub-tropical westerly jet-stream at 200 hPa of June-September in the wet decade is shown in Fig. 1c. During the other two dry decades also, a similar wave No. 3 appeared as equatorward westerly intrusion. During the dry decades, with upper tropospheric westerlies at lower latitudes, the monsoon convective heat source likely acted on the middle latitude upper tropospheric westerly winds to produce the APW. Convective heat source anomalies showed large east-west oscillations inter-annually between the Indian and western Pacific oceans, as shown in Fig. 2c. The positive monsoon heat source anomaly occurred at point B in the figure in drought monsoon years in the western Pacific ocean and at point A in excessive monsoon rainfall years in the western Indian ocean. These heat sources, with divergence areas in the upper troposphere, likely generated large-amplitude wave No. 6 Rossby waves, or APWs, in the mid-latitude upper tropospheric westerlies; opposite spatial phases occurred in years of excessive and deficient ISMR (Figs. 3a-c). When the convective heat source shows a positive anomaly in the western Pacific ocean, an APW with a wave trough over northwestern India is induced to force a deficient (drought) monsoon over India. This deficient monsoon with a weak LLJ generates a positive SST anomaly associated with a positive convection anomaly in the post-monsoon months over the western Indian Ocean, which induces a half-wavelength phase shifting of the APW and results in a monsoon with excessive rainfall in the following year. This negative feed-back process is repeated to cause frequent monsoon droughts, as observed in the dry epochs. For this hypothesis to be valid, the mid-latitude westerly belt of the upper troposphere must move on the decadal scale to lower latitudes over southern Asia, as indicated in the data of 1961-90. This hypothesis requires modeling support.

A mechanism relating the upper tropospheric mid-latitude westerlies with the interannual variation in ISMR has been reported by Meehl (1997). The schematic diagram shown in Fig. 11 of his paper is relevant in this context. In his model, the upper tropospheric wave in the westerlies is not wave of No. 6 but has longer wavelength similar to the wave No. 3 on the decadal scale discussed in this section. Wave No. 6, which is an APW, is important in relation to the inter-annual variations of ISMR (Krishnan and Majumdar, 1999; Ding and Wang, 2005; Krishnan and Kumar, 2009).

Analysis of SST time series showed that during dry (wet) epochs, SST anomalies were negative (positive) over areas in the Pacific and Atlantic oceans between latitudes 30° N and 60° N and that SST gradients over these oceans were large (small) in the tropics to high latitudes. The SST variation in the Pacific Ocean is known as Pacific Decadal Oscillation (PDO); that in the Atlantic Ocean is known as Atlantic Multi-decadal Oscillation (AMO). These basin-wide oscillations have temporal periods of 60 to 70 years and are known to have global teleconnections (Kerr, 2000; Delworth and Mann, 2000; Mantua and Hare, 2002). Teleconnections of the AMO with Indian monsoon rainfall research includes that by Goswami et al. (2006), Lu et al. (2006), Zhang and Delworth (2006), Li et al. (2008), Wang et al. (2009), and Luo et al. (2011); the relationship between PDO and ISMR is less explored. The cold phase of the AMO is generally associated with negative ISMR anomalies on the decadal scale. Analysis of data of the most recent 150 years indicates that frequent monsoon drought years occur during the cold phase of the AMO, which is associated with large inter-annual variability of the monsoon and lower than normal decadal mean monsoon rainfall. This aspect has not been adequately addressed in the previous studies. Goswami et al. (2006) defined a troposphere temperature (TT) index in which temperature is averaged between 600 and 200 hPa. The TT for the July-September was computed for a warm AMO phase of 1950-60 and a cold phase of 1970-80. A warm (cold) AMO phase was associated with a positive (negative) TT anomaly over Eurasia and a positive (negative) anomaly of Indian monsoon rainfall. In an observational-cum-modeling study, Wang et al. (2009) determined that the AMO influences Indian monsoon rainfall through the warming and cooling of the Eurasian middle and upper troposphere. Our analysis shows that during the decades of the dry epoch of ISMR (1961-90) the upper troposphere at 500 hPa and 300 hPa showed negative temperature anomalies over Eurasia in the longitude belt in which the equatorward westerly intrusions occurred on the decadal scale. Thermal wind considerations indicate that the southward movement of the westerly belt at 200 hPa is consistent with the cold anomalies at 500 hPa and 300 hPa over Eurasia.

Goswami et al. (2006) used SST anomalies of the area of the Atlantic ocean equator to latitude 70° N and longitudes 270° E to 20° E to define the index for AMO. The 11-year moving average of the mean June-September SST of this area showed a statistically significant linear correlation coefficient (LCC) of only 0.29 with the 11-year moving average of ISMR for 1871 to 2008. As indices of AMO and PDO, we used the June-September SSTs averaged over the 30° N-60° N, 300° E-360° E and 30° N-60° N, 150° E-220° E, respectively. The 11-year moving averages of these indices (de-trended) and ISMR showed statistically significant LCCs of 0.42 for AMO and 0.46 for PDO for the same period. These indices were derived by using extended reconstructed SSTs (ERSSTs; Smith and Reynolds, 2003). Lag correlations for the same period between the indices of AMO and PDO show maximum and significant LCCs at a lag of one to two years, which indicates that the two oscillations have very little phase differences. Thus, both AMO and PDO have tele-connections with ISMR.

Meridional SST gradients showed significantly higher LCCs with ISMR. The 11-year moving averages of June-September SST gradients over the Atlantic Ocean, which is the difference in SSTs between 20° S-10° N, 300° E-360° E and 30° N-60° N, 300° E-360° E, and ISMR showed an LCC of 0.42; a similar SST gradient over the Pacific Ocean, which is the difference in SSTs between 20° S-10° N, 150° E-220° E and 30° N-60° N, 150° E-220° E, showed an LCC of 0.66 for 1871 to 2008. The 11-year moving average of the global meridional SST gradient between 20° S-10° N, 50° E-360° E and 30° N-60° N, 120° E-360° E and ISMR showed a very high LCC of 0.73 for 1871 to 2008 (0.74 with linear trend removed). To determine whether large SST gradient phases of the global oceans induce upper tropospheric Eurasian cooling and westerly wind intrusions equatorward over southern Asia and create decadal and inter-annual variations in ISMR, as given in our hypothesis, GCM modeling studies are necessary.

This paper has presented recent research on inter-annual and inter-decadal variabilities of ISMR. A hypothesis has been presented for the relationship of frequent droughts in monsoons during 30-year dry epochs occurring during the cold phase of the PDO/AMO and large meridional SST gradients over global oceans. This mechanism involves interaction of the monsoon convective heat source, the southward displaced subtropical westerlies and the subtropical jet stream of the upper troposphere, and the generation of a large amplitude Rossby Wave train, or an APW.