The Monsoon Low-Level Jet: Climatology and Impact on Monsoon Rainfall over the West Coast and Central Peninsular India

Low-level jets can be seen in many parts of the world. They are important for the horizontal and vertical transport of heat and moisture (Stensrud, 1996). The Somali low-level jet originates in the Southern Hemisphere and strikes the western Ghats on the west coast of India after traveling thousands of kilometers. It is often known as the monsoon low-level jet (MLLJ) when being seen over the Indian Ocean and South Asia during the Indian summer monsoon season (Joseph and Raman, 1966; Findlater, 1969), with a core at about 1.5 km above mean sea level and core wind speed of the order of 40–60 knots (kt). It starts intensifying from May and remains over India till October, but it attains maximum intensity between June and September with a peak in July (Raman et al., 2011; Viswanadhapalli et al., 2020).

Many studies have been carried out in the past to understand the structure of the MLLJ (Bonner, 1968; Kala-pureddy et al., 2007) and its role in modulating the In-dian monsoon rainfall (Kumar et al., 2007; Chakraborty et al., 2009; Subrahmanyam and Pushpanjali, 2018). Using an atmospheric general circulation model, Chakraborty et al. (2009) suggested that the strength of the meridional jet near the Somali coast has no bearing on Indian monsoon rainfall whereas the magnitude of the low-level westerly jet over the Arabian Sea was directly linked to the strength of the Indian monsoon rainfall. Joseph and Sijikumar (2004) studied the vertical structure and intraseasonal variability of the MLLJ, as well as its relation with the convective heating of the atmosphere over the Bay of Bengal, and showed that the convective heating of the atmosphere over the Bay of Bengal has a significant linear correlation with the zonal component (u) of the wind at 850 hPa over Peninsular India. In this study, Joseph and Sijikumar (2004) used NCEP/NCAR reanalysis wind data. Subrahmanyam and Pushpanjali (2018) also used the NCEP reanalysis data to conclude that a positive correlation exists between intensity of the MLLJ and rainfall during and after the monsoon onset over Ke-rala during the El Niño years. Raman et al. (2011) used high-resolution GPS radiosonde observations of one station in Peninsular India, and NCEP/NCAR and ECMWF interim reanalysis data to conclude that the MLLJ core does not lie at a fixed height or pressure level homogeneously throughout southern Peninsular India but varies with location depending on the topography. Kalapureddy et al. (2007) studied the structure, diurnal, and seasonal features of the MLLJ over a tropical station, Gadanki, using high-resolution wind observations from the lower atmospheric wind profiler.

All these previous studies used either the model reanalysis data or less than five-year upper-air wind data from a single wind profiler or radiosonde. In the present study, 34-yr daily actual radiosonde/radio wind (RS/RW), rainfall, relative humidity (RH), cloud cover, and geopotential height in meters (GPM) data from 9 stations in the island and Peninsular India are used to explain the structure of the MLLJ over the Indian landmass and its impact on monsoon rainfall. According to the criteria of the Indian Meteorogical Department (IMD) (Attri and Tyagi, 2010), June–September is considered as the monsoon period in India in this study.

The MLLJ has been found to be associated with the development and evolution of deep convection. Grossman and Durran (1984) showed that the mountains in western Ghats, though not very high, play an important role in overall monsoon convection in India. They showed that the western Ghats are capable of contributing to the production of deep convection well offshore by the gentle lifting of potentially unstable air as it approaches the coast. This lifting is due to the deceleration of the low-level air as it approaches a zone of high pressure caused by an upstream blocking effect of the mountains. The western Ghats has a stimulating role in summer monsoon characteristics (Sarker, 1966; Sijikumar et al., 2013). In Fig. 1, the 34-yr (June–September) mean monthly reanalysis wind at 850 hPa is averaged to depict the climatological mean position of the MLLJ core. Similarly, the 34-yr (June–September) mean monthly reanalysis vertical wind is averaged to show the region of intense air rise and subsidence at 850 and 500 hPa in Fig. 1. The closely packed contours of cooler colors indicate the region of intense air rise whereas the warmer colors denote the region of air subsidence. The stations used for this study are marked as black triangles in Fig. 1. The region of maximum convection is offshore of the western coast over 10°–18°N, 70°–75°E and the adjacent land area. The region of maximum air subsidence is over Tamilnadu and South Andra Pradesh. Since deep convection produces a significant amount of cloudiness and rainfall, the study of the relationship between MLLJ and rainfall is important for apprehending the Indian summer monsoon rainfall variability.

Figure 1.  (a) 34-yr (1971–2004) averaged monthly mean wind (m s⁻¹) at 850 hPa during the Indian summer monsoon season and mean monthly vertical wind (omega in hPa s⁻¹) at (b) 850 hPa and (c) 500 hPa based on 34-yr (1971–2004) NCEP/NCAR reanalysis data. The black triangles show the location of 9 stations.

In the present study, balloon-borne radiosonde/radio wind (RS/RW) data of 9 stations on the west coast and central Peninsular India are used to unveil some new characteristics of MLLJ and its impact on monsoon rainfall. As the MLLJ is vertically extended to a few kilometers, the present study focuses on the effects of various parameters at different standard pressure levels on the Indian summer monsoon rainfall variability. Here, vertical cross-sections of the MLLJ and its effects on the Indian summer monsoon rainfall variability are thoroughly studied based on actual surface and upper-air meteorological observations as well as reanalysis data.

The India Meteorological Department’s daily 0000 UTC upper-air radiosonde/radio (RS/RW) wind and standard synoptic surface observations containing cloud cover and 24-h total accumulated rainfall data of 34 yr from 1971 to 2004 along the west coast and central Peninsular India are used for this study. The early morning upper-air observations ensure minimal convective and sea-breeze influences. The RS/RW and synoptic observational data of the following 9 stations are used: Amini (11°07′N,72°44′E), Bengaluru (12°58′N,77°35′E), Goa (15°29′N, 73°49′E), Hyderabad (17°27′N, 78°28′E), Mangalore (12°30′N, 74°30′E), Minicoy (8°18′N, 73°09′E), Mumbai (19°07′N, 72°51′E), Nagpur (21°06′N, 79°03′E), and Thiruvananthapuram (8°29'N, 76°57′E). The RS/RW data are available at every 50 hPa interval from 1000 to 200 hPa, every 25 hPa interval from 200 to 100 hPa, and every 10 hPa interval from 80 to 30 hPa. However, the 925-hPa RS/RW data are not available from 1971 to 1991. The spatial coverage of the RS/RW data is large, i.e., around 2.5° of longitude at the equator (Jones et al., 2010). Hence, the upper-air data from the network of these stations can be considered of good density for the study of the MLLJ over Peninsular India.

Since the MLLJ is a planetary scale system and it mostly travels over the Indian Ocean and at very few places comes inland, so there is a lack of in-situ observations. Hence, to study the position of the core of the MLLJ and region of maximum convection holistically, the NCEP/NCAR reanalysis data of mean monthly horizontal wind field and vertical wind at standard pressure levels on a 2.5° latitude–longitude grid for the same period (1971–2004) are used. The ERA5 data of ECMWF or Indian Monsoon Data Assimilation and Analysis (IMDAA) data of National Centre for Medium Range Weather Forecasting (NCMRWF) are high resolution datasets but their horizontal wind data are available from 1979 onwards and the IMDAA’s vertical wind data are available from 1996 onwards, so the NCEP/NCAR data are used. Being a planetary-scale system, the MLLJ has large horizontal extent, so 2.5° gridded data are sufficient to capture its features over sea. Depending on the relative influence of the observational data and the model used on the gridded variable, the NCEP/NCAR data have been classified into four categories. Wind data are under category A, which is the most reliable class.

To find the relationships between different parameters, data of 136 months, 622 weeks are obtained from the daily observational data of 4148 days of monsoon period (June–September) from 1971 to 2004. Therefore, the time series with 136, 622, and 4148 datasets are used to find monthly, weekly, and daily correlations, respectively. The number of degrees of freedom may be less than the number of data points if the autocorrelation is significant; therefore, the effective degrees of freedom are computed for the daily and weekly time series. The effective degrees of freedom (DoF) N* for significance tests of the correlation between two times series X_i and Y_i with different lag-one autocorrelations r₁ and r₂ are computed by using the following formula (Bretherton et al., 1999):

It is to mention that to delineate the climatological features of the MLLJ over inland and Peninsular India, the in-situ observations from the IMD’s observatories have been used whereas the NCEP/NCAR wind data have been used to show the MLLJ’s features only over the sea.

By using the actual RS/RW data from 9 stations, the climatology, structure, characteristics, and influence of the MLLJ on intensity and distribution of rainfall over Peninsular India are discussed in the following sections.

The NCEP/NCAR reanalysis data (Fig. 1a) show that during the monsoon period, the region of highest wind speed of the MLLJ lies at 850 hPa, with a wind speed of 40–60 kt over the Arabian Sea (7°–14°N, 50°–65°E). Then, it turns clockwise and enters into India through the western Ghats; consequently, the wind speed slows down due to mountain blocking and frictional forces but strong horizontal and vertical wind shear exists. Climatologically, over India, the intensity of MLLJ is maximum around 8°N. The mean zonal wind of 9 stations over the Arabian Sea, west coast, and central Peninsular India also shows wind maxima at 850 hPa (Fig. 2).

Figure 2.  34-yr (1971–2004) mean vertical profile of zonal wind during the monsoon season at the 9 stations under consideration.

The wind rose plot (Fig. 3) of 850-hPa daily winds during the monsoon period of 1971–2004 also shows that the axis of MLLJ in the eastern Arabian Sea and Peninsular India is north–west and south–east oriented. The prevalent wind direction is westnorthwesterly in Amini, Hyderabad, Mangalore, Minicoy, Nagpur, and Thiruvananthapuram; westerly in Bengaluru and Goa; and southwesterly in Mumbai. Wind speed frequency distribution shows that the MLLJ wind speed is up to 11 m s⁻¹ on 50% of the days during monsoon season over these peninsular and island stations. The MLLJ speed is more than or equal to 20 m s⁻¹ in Minicoy, Thiruvananthapuram, Hyderabad, and Amini stations on more than 5% of monsoon days during the 34 years with the highest frequency of 8.46% at around 8°N at Minicoy station (Fig. 4).

Figure 3.  Wind rose plot of daily winds at 850 hPa based on 0000 UTC RS/RW observations for the monsoon period of 1971–2004.

Figure 4.  Frequency of the number of monsoon days (%) in which the wind speed is more than or equal to 20 m s⁻¹ at 850 hPa.

The MLLJ is a single narrow band of high wind speed with the maximum wind speed at the core of the jet, which lies at 850 hPa with strong vertical and horizontal wind shear. Using monthly mean winds, Findlater (1971) first suggested the splitting of MLLJ over the Arabian Sea into two branches: one branch passes through Peninsular India while the other moves southeastwards crossing Sri Lanka. Krishnamurti et al. (1976) showed the splitting of the Somali jet through numerical simulation and explained it in terms of barotropic instability. However, using the daily NCEP/NCAR wind data, Joseph and Sijikumar (2004) found that the MLLJ does not split into two branches over the Arabian Sea, rather it changes its position by passing through the Indian landmass or close to Sri Lanka during active and break monsoon spells.

However, in the current study, the analysis of 0000 UTC actual wind data at 850 hPa for individual days showed two cores of high wind speed, one near 8°N and the other around 17°N. This bimodal wind speed characteristic of the MLLJ confirms the presence of two branches of high speed wind over Peninsular India, which can be seen on some days during the monsoon season. The result is consistent with the initial findings of MLLJ splitting. The branch at 8°N is found on most of the days and has the highest average wind. The other branch is found at around 17°N and sometimes at 13°N on a few days. The horizontal extension of the MLLJ across its axis is less. However, on a few occasions, the core of maximum wind speed can span up to 12° in width over Peninsular India. The expansion in the horizontal width of the core of MLLJ is seen one or two times during most of the years and was observed three times in 1986.

Figure 5 shows the images of MLLJ wind speed that are generated by plotting the 5-day moving average wind speed at 850 hPa at individual stations from 1971 to 2004. These stations are located between 72°44′ and 79°03′E. Thus, 75°53′E as the average of the longitudinal range is considered for the analysis. The region of data gap between the stations is calculated by the inverse distance weighting interpolation. The temporal evolution of wind speed at 850 hPa at daily timescale suggests the simulation occurrences of two high wind speed cores on a few occasions. The green and purple color regions show the core of MLLJ with wind speed more than 15 m s⁻¹. Out of 4148 monsoon days in 34 yr, the splitting of MLLJ was observed on 984 days when the MLLJ’s average core wind speed was more than or equal to 15 m s⁻¹ over India. From Fig. 5, it can be seen that the speed of MLLJ over India starts increasing from the first week of June, attaining a peak during the seventh week (12–18 July) of the monsoon season. Thereafter, the speed of MLLJ starts decreasing till the end of the season.

Figure 5.  Wind speed (m s⁻¹) at 850 hPa for monsoon months. Vertical-axis shows the latitude north of the equator and the x-axis shows days from years (a) 1971–1975, (b) 1976–1980, (c) 1981–1985, (d) 1986–1990, (e) 1991–1995, (f) 1996–2000, and (g) 2001–2004.

Figure 5.  (Continued).

The splitting of MLLJ over the Indian region was further investigated by using the ERA5 reanalysis daily wind data available at 0.25° × 0.25° horizontal resolution (Hersbach et al., 2018, 2020). The 850-hPa daily wind data at 0000 UTC from 1979 to 2004 have been analyzed. Two branches of high wind speed are seen on a few days over Peninsular India and Bay of Bengal (BoB) during monsoon season. The mean daily wind of 26 yr (1979–2004) at 850 hPa is shown in Fig. 6a. While entering India, the MLLJ decelerates due to mountain blocking by the western Ghats and after crossing the western Ghats, the MLLJ accelerates. Over Peninsular India and southwestern BoB, the two branches of MLLJ with wind speed > 10 m s⁻¹ are evident in Fig. 6a. The daily analysis of ERA5 wind data confirms the bimodal character of MLLJ over Peninsular India and BoB on a few days of monsoon season. Figure 6b shows some of the days out of a few cases when the bimodal feature of MLLJ was evident. Figure 6c shows splitting of the MLLJ into two branches in the presence of upper-air cyclonic circulation, which may be associated with a low pressure area or a monsoon depression formed in northern BoB and moved northwestwards.

Figure 6.  (a) The 26-yr mean of ERA5 daily 850-hPa wind during monsoon season of 1979–2004, (b) 850-hPa wind of some of the cases when two branches are evident, and (c) 850-hPa wind on days when a cyclonic circulation is present to the north of 18°N. The color shading shows wind speed ≥ 14 m s⁻¹.

The relationship between MLLJ and different meteorological parameters is examined by using Pearson’s linear correlation coefficient. An attempt has been made to identify the strength and prominence of this relationship at various pressure levels, so that relevant weightage can be given to the parameter at that pressure level for monsoon rainfall forecasting. The linear correlation is calculated for different time series, i.e., daily, weekly, and monthly series.

The strong cross-equatorial low-level jet stream is the main conduit for transporting moisture from the southern Indian Ocean and Arabian Sea into the Indian landmass; and it is the most important contributing factor for the summer monsoon rainfall (Findlater, 1969; Kumar et al., 1999). Dew point depression is the difference between temperature and dew point temperature at a given pressure level and it is directly proportional to the relative humidity at that level. On examining the relationship between MLLJ strength and dew point depression, it is found that maximum moisture is transported between 800 and 900 hPa due to MLLJ. The linear correlation coefficients between the magnitude of zonal wind and dew point depression at different pressure levels are shown in Fig. 7. The correlation is higher for the weekly time series and it is highest between zonal wind speed and dew point depression at 850 hPa. This correlation with DoF = 267 calculated by using Eq. (1) is significant at the 99.9% level. For daily time series, the highest [r(1170) = −0.37, p < 0.001] correlation is also found at 850 hPa. The correlation coefficients for 800–900 hPa are between −0.32 and −0.49 for all the time series.

Figure 7.  Linear correlation coefficients between zonal wind speed and dew point depression for daily, weekly, and monthly time series.

Strong vertical wind shear is noticed below the mean height of the jet core, at 850 hPa. The wind shear is determined from the wind speed gradients between two pressure levels. The monsoon season from June to September is divided into 18 weeks, and then the week-wise average of zonal (u) and meridional (v) wind speeds at different pressure levels of 9 stations are calculated. Thereafter, the average vertical wind shears of zonal and meridional winds are calculated by subtracting average wind speed at the lower height from the average wind speed at upper height. The magnitude of highest wind shear of u is found between 950 and 900 hPa followed by 900 and 850 hPa and the lowest is found just above 850 hPa, i.e., for 850–800 hPa (Fig. 8). The wind shear below the MLLJ’s core starts increasing after the first week of June and attains its maximum value during the third week; thereafter, it decreases gradually till the 14th weeks of monsoon and rapidly afterward till the end of September where it approaches zero or reversal of wind shear is seen. From the 3rd to 14th week, the magnitude of wind shear of u ranges from 0.001 to 0.002 s⁻¹ between 900 and 850 hPa and from 0.003 to 0.045 s⁻¹ between 950 and 900 hPa (all in negative sign). Similarly, the magnitude of wind shear of v ranges from 0.0009 to 0.0015 s⁻¹ between 950 and 900 hPa and from 0.001 to 0.0016 s⁻¹ between 900 and 850 hPa. Also, above 800 hPa, the wind shear of u lies between 0.0015 and 0.0025 s⁻¹. However, the magnitude of wind shear of v above 850 hPa is very small, i.e., around 0.0005 s⁻¹ (in negative).

Figure 8.  Week-wise average vertical wind shear of (a) zonal (u) and (b) meridional (v) wind speed from 950 to 500 hPa.

Moisture incursion due to the MLLJ, orographic effect of the western Ghats, and land surface warming causes deep convection inland, and thus cloud formation takes place. The cloudiness over Peninsular India is associated with the arrival and progress of the monsoon over India. Therefore, the relationship between cloud formation and the strength of zonal component of MLLJ is important. Figure 9a shows the linear correlation coefficient values between the average zonal component of MLLJ at different standard pressure levels and the average amount of low cloud cover (okta) for different timescales. It is found that the magnitude of average zonal winds below 800 hPa has the highest correlation with the formation of low clouds over Peninsular India with the correlation coefficient of 0.45, 0.63, and 0.72 for daily, weekly, and monthly time series with DoF = 1436, 299, and 134 respectively at the 99.9% confidence level.

Figure 9.  Linear correlation coefficients between the average zonal component of MLLJ and (a) amount of cloud cover, (b) geopotential height of standard pressure levels of 9 stations for different timescales and different pressure levels.

The relationship between zonal wind speed and geopotential height of different vertical pressure levels is found to be strongest between 800 and 900 hPa. The correlation coefficient is between −0.5 and −0.7 for all the time series at the 99.9% confidence level (Fig. 9b). The temperature advection due to the MLLJ coming from the Arabian Sea has a moderating effect on upper-air temperatures of the Indian landmass, which causes fluctuation in geopotential heights. Also, the change in upper-air pressure over the Indian landmass can cause a change in wind speed of the MLLJ. To further examine this relation, the correlation is calculated first for the lag of geopotential height by 1 and 2 days and then for u. It is found that the correlation decreases for a lag of geopotential height by 1 day and further decrease is noted for a lag of 2 days; whereas, the correlation increases when it is calculated for the lag of speed of u by 2 days and it increases further for a lag of 1 day (Table 1). This suggests that changes in geopotential height between 800 and 900 hPa and thus the increase in the pressure gradient force have a strong bearing on changes in wind speed of the MLLJ than the other way around.

The MLLJ causes large moisture incursion and cloudiness over Peninsular India, and therefore its fluctuation in position and intensity affects monsoon rainfall over Peninsular India. Many earlier studies have shown positive influences of intensity of the MLLJ on the Indian summer monsoon rainfall (Kumar et al., 2007; Aneesh and Sijikumar, 2016; Roxy et al., 2017; Subrahmanyam and Pushpanjali, 2018; Xavier et al., 2018; Viswanadhapalli et al., 2020). Roxy et al. (2017) reported that the strengthening westerlies transport moisture from the Arabian Sea and thus cause intensified precipitation. Some studies have also correlated the weakening trend of the monsoon rainfall over Peninsular India and western Ghats to the weakening of the strength of westerly monsoon flow (Joseph and Simon, 2005; Krishnan et al., 2013). The relationship between rainfall and strength of the MLLJ over Peninsular India is examined by using Pearson’s linear correlation coefficient. Figure 10 shows the linear correlation coefficients between monthly mean, weekly mean, and daily mean of 24-h accumulated rainfall and mean zonal component of the MLLJ for different pressure levels. Of all levels, the correlation at 850 hPa is found to be the highest for all the timescales. The correlation coefficient values are r(134) = 0.65, r(387) = 0.49, and r(1435) = 0.34 for monthly, weekly, and daily series respectively at the 99.9% confidence level. The year-wise correlations between the weekly average of 24-h rainfall and zonal wind at 850 hPa for all the stations are calculated. The correlation coefficient of r > 0.5 at the confidence level of > 98% is found for 18 yr out of 34 yr, i.e., 1972, 1976, 1977, 1979, 1982, 1984, 1985, 1986, 1987, 1989, 1990, 1994, 1995, 1997, 1998, 1999, 2000, and 2003 (Fig. 11).

Figure 10.  Linear correlation coefficients between average rainfall and zonal wind velocity at 9 stations for different timescales and pressure levels.

Figure 11.  Year-wise linear correlation between weekly average rainfall and strength of the zonal wind at 850 hPa.

The MLLJ has strong horizontal and vertical wind shears (Joseph and Raman, 1966); the highest vertical wind shear is found below the core of MLLJ. The MLLJ and the land roughness induced shears in the boundary layer impact the local turbulence parameters, resulting in cloudiness and rainfall. A strong correlation is found between daily, weekly, and monthly average rainfall and magnitude of average vertical wind shear below the core of MLLJ. The correlation coefficient values between the magnitude of monthly, weekly, and daily average vertical wind shear and average rainfall are highest between 950 and 900 hPa and they are 0.62, 0.39, and 0.21, respectively, significant at the 99.9% level (Fig. 12). However, negative correlations are found between the average rainfall and average vertical wind shear just above the MLLJ core (850–800 hPa). This suggests that higher wind speed at 800 hPa, i.e., an elevated core of the MLLJ is more favorable for rainfall in Peninsular India than the MLLJ core at 850 hPa.

Figure 12.  Linear correlation coefficients (CC) between average rainfall and vertical wind shear for different timescales and different standard pressure levels.

Many studies have shown a positive correlation between the strength of low-level monsoon currents and heavy rainfall over Peninsular India (Kumar et al., 2007; Subrahmanyam and Pushpanjali, 2018; Xavier et al., 2018). According to Asnani (1993), the MLLJ becomes strong and the 850–500 hPa layer becomes moist when a synoptic-scale situation [like the formation of a depression over the head of BoB or position of the intertropical convergence zone (ITCZ) over central India, south of its normal position] is favorable for large-scale lifting up of air over the Arabian Sea. Kumar et al. (2007) studied the impact of MLLJ on heavy rainfall events over Mumbai. According to their study, the wind speed of 30 kt or more at 850 hPa and dew point depression ≤ 5°C up to 400 hPa observed at any station along the west coast of Peninsular India from Thiruvananthapuram to Mumbai including Minicoy/Aminidivi islands are the precursors for the occurrence of severe weather conditions over Mumbai station.

The IMD’s rainfall intensity criterion [ADGM(R); IMD, 2015] is used to categorize the 24-h accumulated rainfall into 6 classes, namely, very light, light, moderate, heavy, very heavy, and extremely heavy rainfall. All the above rainfall events at the 9 stations are recorded and plotted against the standardized value of the strength of average zonal and meridional wind at 850 hPa. The standardized values of the zonal and meridional components of MLLJ are calculated by

where $ {u}_{\rm s} \;{\rm{and}}\; {v}_{\rm s} $ are the standardized value, $ { u}_{\rm a} \; {\rm{and}} \; {v}_{\rm a} $ are the actual value of u and v, and $ \sigma $ is the standard deviation of u and v. From Fig. 13, it can be seen that when the strength of the zonal component of MLLJ at 850 hPa increases and the meridional component of MLLJ at 850 hPa decreases (since it is negative), the intensity of rainfall increases. This direct relationship can be explained for very heavy rainfall events; but beyond this, i.e., for extremely heavy rainfall events, the zonal wind velocity decreases whereas there is an increase in meridional wind (which was mainly a northerly wind) velocity from the mean value. This indicates that some other weather systems bringing in the cyclonic winds like westward-moving low-pressure system originating in the BoB have greater influences on the extremely heavy rainfall events in this region.

Figure 13.  Different rainfall events, from light rain to extremely heavy rainfall, with the strength of meridional and zonal wind velocity at 850 hPa.

The MLLJ travels thousands of kilometers before entering India from the west coast. The region of maxi-mum wind strength (RMWS) lies in the western Arabian Sea at 850 hPa, centered at 10°N and around 55°E to 60°E as shown in Fig. 1a. A good correlation coefficient is found between the monthly mean position of the RMWS of MLLJ and rainfall at the 9 stations. The correlation coefficients between the monthly rainfall and latitude and longitude of the MLLJ’s RMWS are 0.47 and −0.396, respectively. The significance of the correlation is tested by using t-test and the correlation is found significant at the 99.9% confidence level. The mean position of MLLJ’s RMWS to the northwest is found to be associated with increased rainfall in Peninsular India and the RMWS’s southeastward position with decreased rainfall.

To further analyze the effect of RMWS’s position on the intensity of rainfall, the mean monthly rainfall at 9 stations is divided into three categories; namely, above normal rainfall (> µ + 1σ), normal rainfall (µ ± 1σ), and below normal rainfall (< µ − 1σ). It is found (Table 2) that when the RMWS shifts north and west from its mean position, these 9 stations on most occasions have above-average rainfall. The southeastward shift of the RMWS indicates below normal rainfall, which is due to its shift near the Andaman Sea from the western Arabian Sea in September. Therefore, this eastward shift indicates the influence of September rainfall on the corresponding position of the RMWS.

It is found that not only the position of the RMWS at 850 hPa influences the intensity of rainfall but the wind speed magnitude of RMWS at 850 hPa also influences the rainfall. The change in the magnitude of the winds at RMWS directly affects the amount of rainfall received at the 9 stations. The yearly rainfalls at the 9 stations are standardized by the following formula:

where $ {R}_{\rm s} $ is the standardized rainfall value and $ {R}_{\rm a} $ is the actual rainfall value. Based on this, the surplus and deficit monsoon rainfall years were identified. Deficit rainfall years are the years when the rainfall is less than 1 standard deviation from the long-term mean value, e.g., 1972, 1973, 1986, and 2002. Similarly, the year in which the rainfall is more than 1 standard deviation from long term mean is identified as surplus rainfall year, e.g., 1975, 1978, 1983, and 2003. As the RMWS lies around 10°N, 55°–60°E, 57.5°E was taken as representative longitude for RMWS. Figure 14 shows the vertical structure of zonal velocity of MLLJ at 57.5°E for surplus and deficit monsoon rainfall years. It is seen that during surplus monsoon rainfall years, the core of zonal component of the RMWS is wider than the deficit monsoon rainfall years, though the rest of the structure of MLLJ is almost identical. The wider core causes quantitatively and spatially increased moisture incursion and convection due to the orography over the Indian landmass.

Figure 14.  (a) Vertical structure of zonal velocity of MLLJ at 57.5°E for surplus rainfall years (1975, 1978, 1983, and 2003) and (b) vertical structure of zonal velocity of MLLJ at 57.5°E based on NCEP/NCAR reanalysis data for deficit rainfall years (1972, 1973, 1986, and 2002).

Previous studies on the structure and characteristics of the MLLJ employed either the model reanalysis data or less than 5 yr of upper-air wind data from a single wind profiler or radiosonde. However, in the present study, 34-yr daily actual radiosonde/radio (RS/RW) wind, rainfall, and other meteorological parameters from 9 stations in the island and Peninsular India are utilized to explain the structure and characteristics of the MLLJ over the Indian landmass and the MLLJ’s impact on monsoon rainfall. The findings of this study are summarized below.

(1) Of all the zonal winds at different standard pressure levels, a significant high correlation was found between rainfall at the 9 stations and the strength of zonal wind at 850 hPa. This is because the MLLJ brings abundant moisture and its core lies at 850 hPa, which is consistent with the high correlation at 850 hPa between the zonal wind speed and dew point depression. It is also observed that the strength of the zonal wind at 850 hPa has a direct relation with the intensity of rainfall. This direct relationship can be explained for very heavy rainfall events, but for extremely heavy rainfall events, the zonal wind velocity decreases whereas the meridional wind velocity increases. This indicates the possible influence of some other low-level cyclonic wind systems like westward moving low-pressure systems originating in the Bay of Bengal on extremely heavy rainfall events.

(2) The bimodal wind speed characteristic of MLLJ on some days during the monsoon season over Peninsular India and BoB is confirmed by the observations and ERA5 wind data. The RS/RW observations suggest the position of one branch at 8°N on most of the days and the position of the other branch at around 17°N and sometimes 13°N on a few days as opposed to the findings of Joseph and Sijikumar (2004). The horizontal width of the core of maximum wind speed of MLLJ expands up to 12° on one or two occasions during most years.

(3) The position of the RMWS has a great influence on rainfall over the west coast and central Peninsular India. The mean position of RMWS is around 10°N, 55°–60°E, which lies in the western Arabian Sea. The high linear correlation between rainfall and latitude (0.47) and longitude (−0.39) of the RMWS suggests that its northwestward shift from the mean position is responsible for more rainfall. The magnitude of the wind at RMWS affects the total seasonal rainfall. During surplus rainfall years, the core of RMWS was wider as compared to deficit rainfall years. A wider core of RMWS causes more moisture influx and deep convection in a larger area.

(4) Wind shear is observed to be highest below the core of MLLJ and the correlation between wind shear below the MLLJ with the average rainfall is highest. The negative correlations between the average rainfall and average vertical wind shear just above the MLLJ core (850–800 hPa) suggest that the higher wind speed at 800 hPa, i.e., elevated core of MLLJ, than that at 850 hPa is more favorable for rainfall in Peninsular India.

(5) Analysis of the relationship between MLLJ’s strength and GPM height showed that the changes in geopotential height between 800 and 900 hPa have a strong bearing on the strength of MLLJ. The change in upper-air pressure over the Indian landmass can cause changes in the wind speed of MLLJ. The rainfall and 900-hPa height have a maximum negative correlation and the 900-hPa height is inversely related to the intensity of rainfall. The strength of MLLJ also influences the process of cloud formation. The 800-hPa zonal wind has the highest correlation with the amount of cloud cover.

The author is thankful to the Director-General of India Meteorological Department and Deputy Director-General of Regional Meteorological Centre, New Delhi for their kind support and encouragement. We thank the Nation Data Centre, Pune for providing the necessary data and the NOAA/OAR/ESRL and ECMWF for providing the reanalysis and associated data.