Examining pathways for modulation of Indian Summer Monsoon Rainfall by extratropical tropospheric temperature pattern

Authors


ABSTRACT

In this article, role of extratropical circulation features in modulating Indian Summer Monsoon Rainfall (ISMR) is examined. Principal component (PC) analysis of extratropical tropospheric temperature data of the monsoon season for 1951–1980 and 1981–2010 years was performed separately. Analysis revealed that a PC (PC 3) of the first period and two PCs (PC 2 and PC 6) of the second period had significant correlation with the ISMR. Composite geopotential height anomaly of middle and upper troposphere, for the dates when extratropical tropospheric temperature anomaly pattern was similar to that of Eigen Vector of the PC 3 of the first period and PC 6 of the second period, highlighted presence of standing waves in the extratropics. These waves had trough over northwestern/northern parts of the Indian subcontinent and adjoining extratropics and ridge eastwards. Such circulation anomaly pattern of the extratropics apparently reduced the north-south tropospheric temperature gradient required for good monsoon flow. This led to weak monsoon currents, resulting subdued ISMR activity during the period. This pathway of interaction of the extratropical circulation anomalies and the ISMR was reported by earlier studies also. Composite geopotential height anomaly of the middle and upper troposphere, for the dates when extratropical tropospheric temperature anomaly pattern was similar to that of Eigen Vector of the PC 2 (explaining larger variance of the temperature data) of the second period also depicted presence of standing waves. These waves were of relatively large amplitude and wavelength and were located at more northern latitudes. The configurations had warm high/ridge with centre over higher latitudes to the north of India and cold low/trough eastwards. Further analysis highlighted a new pathway of modulation of the ISMR by the extratropical circulation anomalies with above features.

1 Introduction

Summer monsoon rainfall over the Indian subcontinent, accounting 75–90% of the annual rainfall, plays a decisive role in influencing quality of life over the region. Quantum of agricultural output in India in a particular year is directly linked to performance of monsoon season in terms of Indian Summer Monsoon Rainfall (ISMR) of the year Parthasarathy et al. (1988). ISMR is also crucial for meeting out drinking water requirements, hydroelectric power generations, etc. It is needless to say that deficient ISMR in a year, even today significantly lowers the Gross Domestic Product (GDP) (a widely accepted measure of economic growth) of India, in spite of rapid industrialization over recent decades and its positive consequences. Owing to its importance, understanding of forcings, apparently responsible for ISMR, and prediction of the ISMR have been a leading area of research. This has certainly led us an enhanced understanding of the factors associated with summer monsoon circulation features. Still, ISMR could not be accurately predicted even in most recent years like 2009 and 2012 with sufficient lead time. Therefore, search and understanding of forcings linked with the ISMR remain ever relevant.

Summer monsoon circulation over the Indian subcontinent is primarily driven by heating contrasts of land and ocean. The differential heating of land and ocean during summer months shifts the Inter Tropical Convergence Zone (ITCZ) northwards over northern and adjoining central parts of Indian subcontinent and North Indian Ocean by the end of May. Subsequently, moisture laden southwesterly currents emanating from the Southern Hemisphere are drawn towards the ITCZ across the Arabian Sea, peninsular parts of India and take opposite turn as monsoon easterlies around the monsoon trough over the Bay of Bengal. Strength of anticyclones in the Southern Hemisphere, sea surface temperatures (SSTs) of equatorial east Pacific Ocean (Keshavamurty, 1982; Barnett, 1983; Rasmusson and Carpenter, 1983; Mooley and Parthasarathy, 1984; Webster and Yang, 1992; Krishna Kumar et al., 1999; Krishnamurthy and Goswami, 2000; Rajeevan and McPhaden, 2004; Kane, 2005), oscillation of different periodicity (Yasunari, 1979; Sikka and Gadgil, 1980), radiation balance over the Tibetan Plateau (Rangarajan, 1963; Kuo and Qian, 1981), etc. play crucial roles in modulating monsoon circulation features and consequently the ISMR. These all have been extensively researched by meteorologists/scientists across the globe. Similarly, Eurasian snow cover in preceding winter and its association with the ISMR has been reported by many studies (Blanford, 1884; Dey and Bhanu Kumar, 1982; Dickson, 1984; Bamzai and Shukla, 1999; Kripalani and Kulkarni, 1999; Robock et al., 2003). Recently, the role of North Atlantic Oscillation (NAO) and mid-latitudinal circulation anomalies in influencing the ISMR has also been examined by some studies (Joseph and Srinivasan, 1999; Chang et al., 2001; Srivastava et al., 2002, 2007; Ding and Wang, 2005, 2007; Goswami et al., 2006; Li et al., 2008; Krishnan et al., 2009).

Eurasian snow cover and heating of the Tibetan Plateau directly affect the subtropical and adjoining mid-latitudinal tropospheric temperature. Apart from this, transient waves associated with the mid-latitudinal circulation pattern during the monsoon season intermittently affect the tropospheric temperature pattern. These waves either strengthen or slacken north-south tropospheric temperature gradient, which modulate the intensity of the monsoon flow over the Indian subcontinent and thus significantly affect the ISMR. However, there are not many studies which examined mid-latitudinal tropospheric temperature pattern and its relation with the ISMR exclusively.

It may be mentioned that the whole troposphere over northern parts of India and adjoining extratropics becomes warm by the end of May and remains so during ensuing monsoon season. This is due to the movement of the Sun in the Northern Hemisphere. Surface heating, during the course of movement of the Sun, establishes monsoon heat low (thermal) in lower levels over the northwestern parts of the Indian subcontinent by the end of May. The heat low is capped by the subtropical high aloft and further there is excessive radiation heating over the Tibetan Plateau. This makes whole troposphere over the region substantially warmer than the surroundings. This warm troposphere over the subtropics and adjoining extratropics causes a north-south tropospheric temperature gradient and apparently acts as a pulling force for moisture laden monsoon currents from the South Indian Ocean to the Indian subcontinent. However, with passage of cold trough of slowly moving extratropical transient waves to the north of Indian subcontinent and warm ridge further eastwards, troposphere over the northern/north western parts becomes cooler. This leads to weakening of monsoon flow due to reduced north-south temperature gradient and consequently lull in the ISMR is observed. This was reported by many studies like Raman and Rao (1981). It may be mentioned here that configuration of such waves is best demonstrated in the middle and upper troposphere and over relatively more northern latitudes than over the subtropics. Analysis of past data also reveals the existence of opposite circulation pattern at times, when warm ridge of extratropical transient waves to the north of Indian subcontinent enhances tropospheric temperature over the northern/northwestern parts. This results in more north-south temperature gradient, stronger monsoon flow and active phase of the ISMR.

However, in recent years, there were many occasions when troposphere to the north of the Indian subcontinent (with core warm region at slightly more northwards) was warm mainly due to movement of extratropical transient waves, the ISMR was above normal, still monsoon currents were relatively weak. This has prompted us to take up this study and examine probable new route of interactions between the tropospheric temperature of extratropics and the ISMR.

In this study, we focus on variability of the mid-latitudinal tropospheric temperature pattern adjacent to the Indian subcontinent and its concurrent association with the ISMR. We find that in recent years, tropospheric temperature pattern with positive anomaly over the extratropics to the north of the Indian subcontinent and negative anomaly eastward (probably linked with extratropical transient waves) is positively correlated with the simultaneous ISMR. Corresponding lower level circulation pattern indicates a new pathway of their association ship.

2 Data and methodology

For this study, we used daily and monthly data of temperature, geopotential height, zonal and meridional wind data at different pressure levels for the period 1951–2010 downloaded from the National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) global reanalysis data website (Kalnay et al., 1996). The gridded reanalysis data are available at a horizontal resolution of 2.5° latitude × 2.5° longitude. ISMR data from the IMD archives and daily gridded rainfall data (Rajeevan et al., 2006) were used in the study. We divided the data into two parts 1951–1980 and 1981–2010 and did analysis separately for both the periods. This was performed mainly due to the reason that many studies report climate shift around the year 1979 Meehl et al. (2009). In addition, noticeable shift in evolution of El Niño events, a major forcing influencing the ISMR was reported after mid-seventies (Wang 1995) and significant rise in SSTs was highlighted (Nitta and Yamada, 1989). It is to be mentioned that the ISMR did not depict any systematic trend or discernible deviation during the periods of the study. However, a study by Joseph and Simon (2005) reported weakening of lower level monsoon circulation (westerly zonal flow at 850 hPa level) over the peninsular India in the later period.

In this study, we computed vertically averaged temperature data (averaged for 700, 600, 500, 400 and 300 hPa levels over each grid) for four individual monsoon months from June to September and then averaged it over the each grid for the 4 months again to get the data for the monsoon season. As the height of the mountains in the lower Himalayas, in general, varies between 1800 and 4600 m, we considered 700 hPa and above levels for averaging. Further, troughs and ridges of transient waves of the extratropics, lying between longitudes, 60°–120°E, influence tropospheric temperature to the north of the Indian subcontinent. This affects the north-south temperature gradient and consequently the ISMR. Therefore, we did principal component analysis (PCA) of vertically averaged temperature of the monsoon season over region bound between latitudes 30°–55°N, longitudes 60°–120°E for two periods separately and identified principal components (PCs) correlated with the ISMR for further analysis.

PCA, already widely used in meteorology, is a useful tool by which variables originally correlated are made mutually independent through a new set of variables which are linear combinations of the original variables. The new variables are called as PCs and are obtained by solving the equations under following two conditions:

  1. They are mutually independent.
  2. The variance explained by the PCs is in decreasing order. Thus, the first component explains maximum variance; and the second PC will explain second highest variance and so far so on.

Statistically, for p variables X = [x1, x2,…, xp], we find bi (1 ≤ ip) such that the variance of b′ix = bi1x1 + bi2x2 + ⋯ + bi pxp = b′iCbi is maximum (bi is the transpose of the matrix [bi j], 1 ≤ jp), provided

  1. bi bi = 1 (This condition is necessary for unique solution to exit)
  2. biC bi = 0 for 1 ≤ ijp

where C is a covariance or correlation matrix.

We investigated spatial pattern of Eigen Vectors of the PCs linked with the ISMR (showing significant simultaneous correlation with the ISMR). We further, prepared and analysed composite tropospheric temperature pattern in tune with the spatial pattern of Eigen Vectors of the PCs (averaged for the dates when extratropical tropospheric temperature patterns were similar to spatial patterns of Eigen Vectors of the correlated PCs) having significant correlation with the ISMR of each period. Corresponding composite vector wind pattern at 850 hPa, geopotential height anomaly pattern of the middle and upper troposphere and spatial rainfall patterns over India (using daily gridded data) were also examined to investigate the probable pathway of their association ship for the two periods separately.

3 Results

3.1 Tropospheric temperature pattern over subtropics and adjoining extratropics during the monsoon season and its modulation by extratropical transient waves

In this section, we present normal tropospheric temperature pattern during the monsoon season and a particular case when movement of transient wave modulated tropospheric temperature to the north of Indian subcontinent and consequently the ISMR. Figure 1 shows climatology of vertically averaged (averaged for 700, 600, 500, 400 and 300 hPa levels) tropospheric temperature pattern during the monsoon season based on period 1951–2000. We may observe that whole troposphere lying between latitudes 20°–30°N is warmer than the surroundings and temperature falls southwards as well as northwards. In Figure 2(a), locking of cold trough (associated with extratropical transient waves) to the north of Indian subcontinent and warm ridge further eastwards during 12–25 July 1970 is shown by presenting longitude versus geopotential height anomaly pattern (averaged for the latitudes 40°–60°N) for 500 and 300 hPa levels. Similarly, longitude versus height temperature anomaly pattern (averaged for the latitudes 30°–55°N) for 700 and 300 hPa levels averaged for the above dates is shown in Figure 2(b). We may observe that anomaly in geopotential height is substantially negative over the longitudes corresponding to Indian subcontinent and is positive further eastwards (Figure 2(a)). Longitude versus height temperature anomaly pattern (Figure 2(b)) reveals cooler troposphere (negative anomaly) over longitudes corresponding to the longitudes of negative geopotential height anomaly and warmer (positive anomaly) troposphere over longitudes corresponding to the longitudes of positive geopotential height anomaly. This indicates the presence of a wave in extratropics with low/trough to the north of Indian subcontinent and high/ridge further eastwards. Vector wind anomaly at the 850 hPa level (with easterly anomaly over southern and peninsular parts) and spatial pattern of rainfall anomaly over India (negative over major parts) averaged for the same period (Figure 2(c) and (d) respectively) highlighted weaker monsoon flow and subdued rainfall activity over India. This clearly shows the modulation of ISMR by the extratropical transient waves through its impact on low-level monsoon circulation features. The presence of extratropical circulation features with opposite polarity, at times, was also found while analysing the data. In which slow passage of warm ridge of extratropical transient waves to the north of Indian subcontinent enhanced tropospheric temperature over the northern/northwestern parts resulting more north-south temperature gradient, consequently stronger monsoon flow and active phase of the ISMR (not shown here). The above mechanism in which extratropical circulation anomalies modulated the ISMR through affecting the north-south temperature gradient and the monsoon flow is logical and clear. But, in recent years, there were occasions, when, ISMR was above normal, tropospheric north-south temperature gradient was strong, still monsoon currents were relatively weak. Analysis presented in subsequent sections examines this aspect.

Figure 1.

Climatology of vertically averaged (averaged for 700, 600, 500, 400 and 300 hPa levels) tropospheric temperature (°C) pattern during the monsoon season based on 1951–2000 period.

Figure 2.

Composite diagrams of (a) longitude versus geopotential height anomaly (gpm), averaged for latitudes 40°–60°N, (b) longitude versus tropospheric temperature anomaly (°C), averaged for latitudes 30°–55°N, (c) 850 hPa vector wind anomaly (m s−1) and (d) rainfall anomaly (mm) over India, during 12–25 July 1970.

3.2 PCA and correlation matrix

Results of PCA of vertically averaged extratropical tropospheric temperature data of the monsoon season (June–September, averaged for 700, 600, 500, 400 and 300 hPa levels for each grid) between latitudes 30°–55°N, longitudes 60°–120°E for the first six PCs for the two periods, viz. 1951–1980 and 1981–2010 are shown in Table 1. These PCs explain the variability of the data in decreasing order for both the periods, separately. This implies that the first PC explains the maximum variance of the data set and the second PC explains the second highest variance and so on. The first six PCs explain 96.34 and 93.72% variances (respectively) of the temperature data set of two periods. We did not consider other PCs as variance (of the vertically averaged extratropical tropospheric temperature data, under domain of our study) explained by PCs beyond PC 6 was quite less. To examine the relation between the extratropical tropospheric temperatures and the ISMR, we calculated correlation coefficients of the six PCs with the ISMR. In addition, SSTs of NINO 3 region (June–September) and the Monsoon Index (MI) defined by Wang and Fan (1999) were considered for the correlation. Correlation of the PCs with the MI and SSTs was calculated to examine and understand probable pathway through which PCs of extratropical temperature and the ISMR may interact with each other. While SSTs of NINO 3 region is linked to the ISMR through monsoon El Niño Southern Oscillation (ENSO) teleconnection, MI is a representative of strength of monsoon circulation (usually defined as difference of area averaged zonal/meridional winds) and has a very significant correlation with the ISMR. Wang and Fan (1999) defined MI as difference of area averaged zonal wind at the 850 hPa level [viz. U850 (5°–15°N, 40°–80°E) – U850 (20°–30°N, 70°–90°E)]. There are other Monsoon Indices defined by Webster and Yang (1992) and Goswami et al. (1999) also, but the chosen one for the study has the highest and consistent correlation with the ISMR. Correlation matrices for the two periods are shown in Tables 2 and 3. It may be seen from the Table 2 that during the period 1951–1980, only PC 3 explaining 9.9% variance of the temperature data for the 1951–1980 period has significant negative correlation (−0.76) with the ISMR, as well as with the MI (−0.71). It has positive correlation (0.58) with the SSTs of NINO 3 region. Thus PC 3, SSTs of the NINO 3 region and the MI are significantly correlated and linearly dependent also, as correlation coefficient is a measure of linear association ship. We may further note that almost 34% (correlation coefficient (CC) between these two being +0.58) variability of PC 3 is explained by the SSTs of the NINO 3 region and vice versa. Therefore, we may infer from the correlation pattern that the probable forcing linked with the PC 3 and ENSO phenomenon may not be completely independent (linearly). Further, the forcing associated with the PC 3 may modulate the ISMR by weakening the monsoon flow at the 850 hPa level (this has been more elaborated in subsequent sections).

Table 1. General characteristics of the first six Principal Components (PCs) for the two periods, viz. 1951–1980 and 1981–2010
 1951–19801981–2010
Eigen value% Total varianceCumulative % varianceEigen value% Total variance explainedCumulative % variance explained
PC 1177.246264.3310464.331058.5314747.1609547.1609
PC 241.457715.0469779.378023.5703418.9914866.1524
PC 327.36189.9308989.308917.8969914.4202680.5727
PC 49.98933.6256092.93458.885847.1596487.7323
PC 55.17431.8780194.81254.278823.4476091.1799
PC 64.21511.5298696.34243.163042.5485893.7285
Table 2. Correlation Matrix for the period 1951–1980
  Average temperature 700_300 hPa, 30°–55°N, 60°–120°E    
Rainfall percentage departure June–SeptemberPCA 1PCA 2PCA 3PCA 4PCA 5PCA 6U850 5°–15°N, 40°–80°EU850 20°–30°N, 70°–90°EMonsoon IndexNINO 3 June–September
Rainfall percentage departure June–September1.00          
Average temperature 700_300 hPa, 30°–55°N, 60°–120°EPCA1−0.231.00         
PCA2−0.150.01.00        
PCA 3−0.760.00.01.00       
PCA 4−0.060.00.00.01.00      
PCA 5−0.060.00.00.00.01.00     
PCA 60.080.00.00.00.00.01.00    
U850 5°–15°N, 40°–80°E0.840.210.000.740.06−0.180.301.00   
U850 20°–30°N, 70°–90°E−0.570.200.580.390.160.050.02−0.431.00  
Monsoon Index0.870.24−0.240.710.01−0.160.220.930.731.00 
NINO 3 Index June–September−0.640.220.170.580.09−0.040.39−0.700.32−0.661.00
Table 3. Correlation Matrix for the period 1981–2010
  Average temperature 700_300 hPa, 30°–55°N, 60°–120°E    
Rainfall percentage departure June–SeptemberPCA 1PCA 2PCA 3PCA 4PCA 5PCA 6U850 5°–15°N, 40°–80°EU850 20°–30°N, 70°–90°EMonsoon IndexNINO 3 Index June–September
Rainfall percentage departure June–September1.00          
Average temperature 700_300 hPa, 30°–55°N, 60°–120°EPCA 1−0.211.00         
PCA 20.530.01.00        
PCA 30.090.00.01.00       
PCA 4−0.130.00.00.01.00      
PCA 50.170.00.00.00.01.00     
PCA 6−0.480.00.00.00.00.01.00    
U850 5°–15°N, 40°–80°E0.59−0.050.29−0.24−0.09−0.09−0.521.00   
U850 20°–30°N, 70°–90°E−0.490.150.16−0.190.200.010.28−0.321.00  
Monsoon Index0.67−0.110.15−0.09−0.17−0.08−0.520.89−0.711.00 
NINO 3 Index June–September−0.390.36−0.07−0.070.130.250.25−0.350.290.401.00

Similarly, from the Table 3, we may see that during the 1981–2010 period, PC 2 (explaining 18.99% variability of the temperature data set) and PC 6 (accounting for 2.55% variability of the temperature data set), respectively, have significant positive (0.53) and negative (−0.48) correlation with the ISMR. We find that PC 2 does not have significant correlation with the MI and NINO 3 SST anomalies. This suggests that the probable forcing linked with the PC 2 (explaining 18.99% variability of the temperature data set for the period) and ENSO phenomenon appears to be independent (linearly) as correlation coefficient measuring linear relationship between the two is not significant. Further, since the PC 2 does not have significant correlation with the MI, but has a significant correlation with the ISMR, it may not be incorrect to infer that the PC 2 modulates the ISMR but not through the monsoon zonal flow at the 850 hPa level. However, PC 6 (like PC 3 of the first period) has significant negative correlation (−0.52) with the MI and hence it may be inferred that forcing linked to the PC 6 may affect the ISMR by modulating monsoon flow at 850 hPa level.

3.3 Spatial patterns of Eigen Vectors

Spatial patterns of Eigen Vectors of PC 3 of 1951–1980 and PC 2 and PC 6 of 1981–2010 periods are shown in Figures 3, 4 and 5, respectively. Figure 3 showing spatial pattern of Eigen Vector of PC 3 of the 1951–1980 period indicates a southwest (negative values) and northeast (positive values) dipole-like structure in temperature field over major parts and negative values (of less magnitude) further eastwards.

Figure 3.

Spatial patterns of Eigen vectors of PC 3 of averaged tropospheric temperature (June–September) data for 1951–1980 period.

Figure 4.

Spatial patterns of Eigen vectors of PC 2 of averaged tropospheric temperature (June–September) data for 1981–2010 period.

Figure 5.

Spatial patterns of Eigen vectors of PC 6 of averaged tropospheric temperature (June–September) for 1981–2010 period.

Figure 5, representing spatial pattern of Eigen Vectors of PC 6 of the 1981 –2010 period, broadly resembles pattern of Figure 3 (showing spatial pattern of PC 3 of the 1951–1980 period). Thus, there is similarity in spatial pattern of Eigen Vectors of the PC 3 of the first period and that of PC 6 of the second period. Moreover, both these PCs have significant correlation (and sign of the correlation is also identical) with the ISMR and MI. However, while PC 3 of the first period had significant correlation with the SSTs of the NINO 3, PC 6 of the second period did not show significant correlation with the SSTs of the NINO 3 region. This indirectly suggests that the forcing apparently responsible for PC 3 explaining 9.9% variance of the extratropical averaged tropospheric temperature data set of the monsoon season for the region under study in the period 1951–1980 and that of PC 6, explaining 2.55% variance of the extratropical averaged tropospheric temperature data set of the monsoon season for the period 1981–2010, may not be exactly same. But both appear to modulate the ISMR by influencing monsoon flow at the lower levels.

Spatial patterns of Eigen Vectors of PC 2 of 1981–2010 period (shown in Figure 4) indicate a strong dipole-like structure in temperature field, with positive anomalies over all latitudes and longitudes of the extratropics, corresponding to the Indian longitudes and westwards and negative anomalies further eastwards. It may be observed that the peak positive and negative values are located at more northern latitudes.

3.4 Composite temperature, rainfall and vector wind pattern

In earlier section, spatial pattern of Eigen Vectors of the PCs (PC 3 for the 1951–1980 period and PC 2 and PC 6 for the 1981–2010 period) of extratropical averaged tropospheric temperature data for the monsoon season having significant correlation with the ISMR was discussed. To diagnose the probable mechanism of mutual interaction (physical linkage) between the PCs (showing significant correlation with the ISMR) and ISMR, we examined daily data of the two periods separately for dates when extratropical tropospheric temperature pattern over region under study was similar to that of spatial pattern of Eigen Vectors of the PC 3 of the 1951–1980 period and that of the PC 2 and PC 6 of 1981–2010 period. For this, we marked the positive and negative areas in spatial pattern of the Eigen Vectors of PC 3 (Figure 3), PC 2 and PC 6 (Figures 4 and 5) and then counted the number of grid points in each area of the spatial pattern of these Eigen Vectors respectively. The dates when tropospheric temperature anomaly had similar sign (positive or negative) over more than 50% of the grid points in each area (in spatial pattern of each of the three Eigen Vectors separately), we flagged the dates. Further, we prepared composite diagrams for these dates in respect of averaged tropospheric temperature anomaly, geopotential height anomaly pattern of the middle and upper troposphere, 850 hPa vector wind anomaly and spatial rainfall anomaly patterns over India, for each period separately. These are shown in Figures 6, 7 and 8. It is to mention here that for calculating daily anomaly with respect to meteorological parameters used in the study, we calculated long-term mean (based on 1971–2000 period) for each day and long-term mean value of a particular day was subtracted from the observed value of the day to get the anomaly value. We would like to mention that there were total 268 days (around 9 days per year) in the first period 1951–1980 and 347 days (around 11 days per year) in the second period 1981–2010, when spatial pattern of tropospheric temperature anomaly resembled spatial pattern of the Eigen Vector corresponding to PC 3 of the first period and PC 6 of the second period, respectively. Similarly, there were total 1032 days (around 34 days per year) in the period 1981–2010, when spatial pattern of tropospheric temperature anomaly matched spatial pattern of the Eigen Vector corresponding to PC 2 of the second period. Composite tropospheric temperature anomaly pattern (of the dates when tropospheric temperature anomaly pattern matched the spatial pattern of the Eigen vectors of PC 3 of the 1951–1980 period and PC 6 of the 1981–2010 period) in Figures 6(a) and 8(a) show a tripole-like structure in tropospheric temperature field of the extratropics. Corresponding composite geopotential height anomaly patterns of the middle and upper troposphere (Figures 6(b) and 8(b)) also reveal a tripole of negative and positive and further negative geopotential height anomalies (nearly identical to the observed pattern of tropospheric temperature anomalies). This suggests the presence of a standing wave (of relatively small amplitude and wavelength) having cold low/trough (over lower latitudes with centre near southwestern parts), warm high/ridge (over higher latitudes with centre near northeastward) and again cold low/trough over further eastward in the middle and upper troposphere over the region of extratropics under study. In such configuration anomalous cold low/trough over the northern/northwestern parts of India in the middle and upper troposphere appears to have been blocked by the anomalous high positioned eastwards. This type of circulation anomalies in the extratropics and adjoining northern latitudes of India have been found a conduit for intrusion of dry air from the extratropics into the Indian subcontinent. Many studies (viz. Raman and Rao, 1981; Krishnamurti et al., 1989) found that this type of extratropical circulation anomalies weakened monsoon flow and were responsible for monsoon break-like situations in the ISMR. During such a circulation anomaly structure of the extratropics, simultaneous monsoon flow over India is weak and anomalous easterly flow appeared over the peninsula in the lower levels and monsoon rainfall activity is subdued over major parts of India. This is precisely what we observed in the anomalous corresponding vector wind at the 850 hPa level (Figures 6(c) and 8(c)) and rainfall patterns (Figures 6(d) and 8(d)).

Figure 6.

Composite diagrams of (a) averaged tropospheric temperature anomaly (°C), (b) geopotential height anomaly (gpm) of the middle and upper troposphere, (c) 850 hPa vector wind anomaly (m s−1) and (d) rainfall anomaly (mm) over India, for the dates, when spatial pattern of averaged tropospheric temperature resembled spatial pattern of Eigen vector of the PC 3 of tropospheric temperature for 1951–1980 period.

Figure 7.

Composite diagrams of (a) averaged tropospheric temperature anomaly (°C), (b) geopotential height anomaly (gpm) of the middle and upper troposphere, (c) 850 hPa vector wind anomaly (m s−1) and (d) rainfall anomaly (mm) over India, for the dates, when spatial pattern of averaged tropospheric temperature resembled spatial pattern of Eigen vector of the PC 2 of tropospheric temperature for 1981–2010 period and magnitude of the anomaly was more (less) than 0.5°C in the positive (negative) regions.

Figure 8.

Composite diagrams of (a) averaged tropospheric temperature anomaly (°C), (b) geopotential height anomaly (gpm) of the middle and upper troposphere, (c) 850 hPa vector wind anomaly (m s−1) and (d) rainfall anomaly (mm) over India, for the dates, when spatial pattern of averaged tropospheric temperature resembled spatial pattern of Eigen vector of the PC 6 of tropospheric temperature for 1981–2010 period.

Figure 7 showing composite tropospheric temperature pattern of the extratropics, composite geopotential height anomaly pattern of the middle and upper troposphere, 850 hPa vector wind anomaly and spatial rainfall anomaly pattern presents very interesting facts. Composite tropospheric temperature patterns (of the dates when tropospheric temperature anomaly pattern matched the spatial pattern of the Eigen vectors of PC 2 of the 1981 – 2010 period and magnitude of the anomaly was more[less] than 0.5°C) in Fig 7(a) shows a very prominent di-pole structure in tropospheric temperature field of the extra tropics. There were total 584 days (around 20 days per year) during the 1981–2010 period, when anomaly in tropospheric temperature was more (less) than 0.5°C over more than 50% grids of the positive (negative) regions. Corresponding composite geopotential height anomaly patterns of the middle and upper troposphere (Figure 7(b)) also reveal a strong dipole of positive and geopotential height anomalies of the middle and upper air and which match the observed pattern of tropospheric temperature anomalies. This anomalous configuration indicates the existence of a standing wave in the extratropics with relatively large amplitude and wavelength, having warm high/ridge over longitudes corresponding to the Indian continent and a cold low/trough further eastward in the middle and upper troposphere over the region of extratropics under study. We may also observe that centres of the high and low are at relatively higher latitudes (more northwards). Corresponding composite rainfall anomaly pattern in Figure 7(d) reveals good rainfall activity with positive rainfall anomalies over the core monsoon regions (Rajeevan et al., 2006) and other areas indicating active monsoon conditions. However, composite vector wind anomaly at 850 hPa level (Figure 7(c)) with anomalous easterly flow over the peninsula for the corresponding dates does not support active monsoon conditions. Then a natural question arises that if the monsoon flow is weak, how rainfall anomaly is positive. We may closely observe the composite vector wind pattern at the 850 hPa level (Figure 7(c)) to find that an anomalous anticyclonic circulation prevails over the head bay and its outflow is pumping moist air through anomalous easterly flow over the central parts of India. These easterlies meet northerly currents coming from the extratropics over the region. These interactions of moist tropical easterly flow and relatively dry winds from the extratropics caused good amount of rainfall. We also calculated anomalous moisture flux at 850 hPa level for the above dates and prepared composite analysis (shown in Figure 9). We may again observe interaction of moist easterlies of the tropics and dry northeasterlies from the extratropics over the north central and northern parts of India. This sort of wind configuration and meeting of two different kinds of air masses may lead to enhance in situ instability and may lead to intense convective precipitation.

Figure 9.

Composite diagrams of anomalous moisture flux (kg m/kg s) at the 850 hPa level for the dates, when spatial pattern of averaged tropospheric temperature resembled spatial pattern of Eigen vector of the PC 2 of tropospheric temperature for 1981–2010 period and magnitude of the anomaly was more (less) than 0.5°C in the positive (negative) regions.

4 Discussions and conclusions

In this study, we investigated the role of extratropical tropospheric temperature in influencing the ISMR. PCA of extratropical tropospheric temperature data of the monsoon season (June–September) for the region bound between latitudes 30°–55° N, longitudes 60°–120°E for the two periods, viz. 1951–1980 and 1981–2010 was performed separately. We found that in the first period, third PC (PC 3) explaining 9.55% variability of the extratropical tropospheric temperature data was very significantly correlated (with CC = −0.76, significant at the 0.1% level) with the ISMR. Similarly, in the second period, second PC (PC 2) and sixth PC (PC 6) explaining 18.55 and 2.55% variability of the extratropical tropospheric temperature data, respectively, were significantly (positively and negatively respectively) correlated with the ISMR.

Spatial patterns of Eigen Vector of PC 3 of the first period 1951–1980 and that of PC 6 of the second period 1981–2010 were broadly similar. Composite analysis of tropospheric temperature anomaly for the dates when extratropical tropospheric temperature anomaly pattern was similar to that of respective Eigen Vectors of the two periods and corresponding composite analysis of geopotential height anomaly suggests the presence of standing wave-like structure with trough/low over northern parts of India and adjoining extratropical latitudes and ridge/high eastwards. These waves have known pathways (by reducing tropospheric temperature gradient and consequently pressure gradient to support good monsoon flow) to weaken monsoon flow and this is reflected in vector wind anomaly pattern of 850 hPa level and the rainfall anomaly pattern over India. This is the reason that PC 3 (1951–1980 period) and PC 6 (1981–2010 period) have shown significant correlation not only with the ISMR but also with the monsoon index.

Composite analysis of tropospheric temperature anomaly for the dates when extratropical tropospheric temperature anomaly pattern was similar to spatial pattern of Eigen Vectors of the PC 2 (explaining much more variability, viz. 18.55%) of the later period 1981–2010 and corresponding composite analysis of geopotential height anomaly also suggests the presence of standing wave (over more northern latitudes) of comparatively higher wave length with ridge/high over extratropical latitudes and adjoining northern parts of India and trough/low eastwards. However, simultaneous circulation anomalies at the 850 hPa level suggested the interaction of moist easterlies (pumped through anticyclone over the Northern Bay of Bengal) and dry northwesterly flow from the extratropics as the probable cause of increased rainfall over central and adjoining parts of the country. This sort of mechanism has been reported for the first time. This may also explain occurrence of observed (of late) enhanced convective rainfall pattern during the monsoon season.

To examine robustness of results, we carried all the analysis using ECMWF ERA 40 data (1958–2002) and got the similar results. Further, we examined composite temperature, geopotential height anomaly, vector wind and ISMR for the dates corresponding to spatial pattern of Eigen vector associated with the correlated PCs in which polarity was reversed for each period separately. Composite spatial patterns in respect of all parameters were also opposite.

This study could not examine and pinpoint the forcings apparently responsible for the correlated PCs (correlated with the ISMR) of the extratropical tropospheric temperature data for both the periods, none the less results of the study will motivate more dynamic studies to investigate the same.

Acknowledgements

The authors are grateful to the DGM, IMD for providing necessary facilities for the study. They are thankful to Shri S. Krishnaiah, ADGM(R) and Shri B. Mukhopadhyay, DDGM(C) for providing constant encouragement. Thanks are also due to Smt. U. J. D'souza for typing the manuscript.

Ancillary