Relationship between surface atmospheric convergence over Indian Ocean and Indian rainfall



[1] Atmospheric convergence regions over the tropical Indian Ocean have been mapped for the first time using 31 years of vector wind data. The convergence fields reveal that the energetic summer monsoon winds enhance convergence in the central Arabian Sea and in the eastern Bay of Bengal when compared with weak winter monsoon winds. The summer monsoon is also effective in spatial migration of convergence regions in both basins. Area-integrated convergence time series reveals an annual cycle with high amplitude during summer monsoon, which occurs in phase with Indian rainfall. The study explores the prospects of using the convergence as one of the predictors of Indian rainfall.

1. Introduction

[2] The Indian climate is characterized by “wet” and “dry” phases corresponding to warm, moist and energetic winds which blow from the southwest during June to September (summer monsoon) and cold, dry and weak continental winds which blow from northeast during November to March (winter monsoon). The southwesterly winds pick up moisture from the western Arabian Sea and deposit rainfall over 80% of the Indian subcontinent, which meets the socio-economic needs of about one billion people. These winds exhibit converging and diverging patterns over the ocean. The converging winds enhance convective activity, increase moisture in the atmosphere through evaporation, which eventually increase regional precipitation. Such regions are referred to as atmospheric convergence zones; the Intertropical Convergence Zones which develop between 5 ∼ 12° to the north and south of the equator are the strongest zones which stem from the convergence of the southeast and northeast trades. The position of these zones has a major impact on rainfall in certain areas and hence, their composite picture reflects the scenario of the tropical atmosphere [Philander, 1990]. In addition to these large-scale convergences, there are also regional atmospheric convergence (hereinafter convergence) regions over the Indian Ocean promoted by the monsoon winds. In this work, for the first time, the convergence regions in the Arabian Sea and Bay of Bengal are mapped and the relationship between convergence and Indian rainfall is highlighted.

2. Data and Method

[3] In this work, ocean surface wind data, obtained from the Florida State University, mapped on 1° latitude by 1° longitude resolution and monthly rainfall data over India have been used. The vector wind was derived from merchant ships, buoys, and other marine observing stations. The quality checks involved objective filtering to eliminate spurious wind data. A variational, direct-minimization objective analysis was performed on the quality-controlled data using the 1° gridded data as a first guess [Legler et al., 1989]. This data set covers the period 1970 – 2002, which is the longest time span when compared to scatterometer-based winds. Using vector wind we computed convergence over the ocean surface as follows:

display math

where u and v are zonal (positive eastward) and meridional (positive northward) wind components in m s−1, respectively. C takes large positive values in conditions of strong convergence in the Arabian Sea during May–September. To relate the convergence to Indian rainfall, we employed the monthly rainfall data derived for 29 meteorological subdivisions of India during 1970 – 1991 [Parthasarathy et al., 1995]. The data set has been further extended up to 2001 by using preliminary estimates based on the sub-divisional means supplied by the India Meteorological Department. We relate Arabian Sea (Bay of Bengal) convergence to rainfall over western coast of India (Gangetic West Bengal) and further explore the possibility of using convergence as one of the prediction parameters for Indian rainfall.

3. Annual Cycle of Surface Atmospheric Convergence

[4] Figure 1 depicts bi-monthly maps of convergence computed using equation (1). The convergence is weak (2 ∼ 4 × 10−6 s−1) during January–February; it dominates along northwestern Arabian Sea and at the tip of India (Figure 1a). Higher intensity near the Indian tip results due to the convergence of strong northeasterly winds which are channeled through the sea-level gap between elevated topography of Sri Lanka and south India; these winds have been referred to as gap winds [Luis and Kawamura, 2000]. During spring (March–April) the convergence is enhanced by 50% east of Arabia and in the northern Bay of Bengal (Figure 1b). During May – June the convergence region moves into the central Arabian Sea in response to the formation of the strong southwesterly wind jet (Findlater jet, see the dashed arrow in Figure 1d) which extends from Somalia to India. This jet induces deep convection offshore of the west coast of India which increases latent heat from the ocean [Grossman and Garcia, 1983] which in turn leads to cloud formation. Further enhancement in the offshore convection results due to upstream blocking and gentle ascent of the unstable air mass on approaching the western Ghats mountains [Grossman and Durran, 1984]. On the other hand, a large region in the eastern Bay of Bengal extending from the bay down to south of Sumatra experiences convergence (Figure 1c). At the peak of the summer monsoon, the convergence is greatly enhanced in intensity (>5 × 10−6 s−1) and in the spatial extent in the Arabian Sea (Figure 1d). With the weakening of the summer monsoon during late September – October, a shift in the convergence region toward the western Arabian Sea is evident; the convergence region shrinks and its intensity weakens (<2 × 10−6 s−1) in the eastern Bay of Bengal (Figure 1e). With the onset of the winter monsoon, the gap winds enhance the convergence intensity south of the Indian tip. However the convergence intensity is greatly subdued during November–December over the north Indian Ocean due to weaker northeast trades.

Figure 1.

Climatology of bi-monthly fields of surface wind convergence (10−6 s−1).

[5] The seasonal mean pattern of convergence is portrayed in Figure 2. It is pertinent to note that the convergence is strongest in the Arabian Sea during the regime of the strong southwesterly winds (Figure 2a). These strong winds enhance turbulent mixing and transfer heat to the ocean mixed layer owing to overturning and render deep mixed layer in the Arabian Sea [Shenoi et al., 2002]. The Bay of Bengal exhibits an elevated convergence in the bay during winter monsoon, which migrates southeastward and occupies the south Andaman Sea during summer monsoon. Nevertheless, the convergence in the bay is weaker by 20% compared to that in the Arabian Sea during summer monsoon.

Figure 2.

Seasonal-mean surface wind convergence (10−6 s−1) fields for (a) Arabian Sea during summer monsoon and Bay of Bengal during summer (b) and winter monsoon (c). The area which is enveloped by dotted contours indicates that convergence is greater than zero; integration over this region yields total convergence which is shown in Figure 3.

4. Temporal Variability of Total Convergence and its Relation to Indian Rainfall

[6] The total convergence was computed by integrating equation (1) over the region of interest. It takes the form:

display math

In equation (2), A represents the surface area of the convergence region. According to Gauss theorem, TC is the horizontal flux of velocity field across the border line encapsulating the convergence region, which equals to the velocity flux into the atmosphere, assuming that the vertical velocity at the ocean surface to be zero. Based on this criterion, we generated a time series of total convergence by integrating equation (1) such that C > 0. As an example, the convergence area enveloped by dotted line which satisfies (C > 0) in Figures 2a and 2c is indicated for Arabian Sea and Bay of Bengal, respectively.

[7] The anomalous time series (32-year mean removed) Arabian Sea convergence and rainfall over the west coast of India, which is one of the regions which experiences highest rainfall during the southwest monsoon, are depicted in Figures 3a and 3b, respectively. Similarly, the relationship between the anomalous convergence over the Bay of Bengal and rainfall over Gangetic West Bengal is depicted in Figures 3c and 3d, respectively. The following features can be inferred from Figure 3. The time series for the two cases exhibits an annual cycle with maximum summer-monsoon convergence coinciding with high (low) rainfall during July (January). The contemporaneous variability of the two parameters suggests that they are in phase. Anomalous time series of convergence over the tropical Indian Ocean was also found to be in phase with all-India rainfall (figure not shown). We note that the convergence anomalies over the Bay of Bengal display interannual variability, since the bay experiences more than 80% low pressure systems than the Arabian Sea [see, e.g., Mooley and Shukla, 1989]. Although the years 1972, 1979 and 1987 experienced below-average rainfall, there is no clear reflection of this on the convergence intensity. For example, the summer monsoon convergence exhibits highly positive anomalies (∼25 × 106 km2 s−1) during July 1979, while slightly negative anomalies occur during July 1972. Similarly, the excess rainfall years 1988 and 1998 are marked by weak convergence anomalies, i.e., 15 × 106 km2 s−1 and −1 × 106 km2 s−1, respectively. In brief, Figure 3 suggests that march of monsoon-induced convergence over the ocean occurs in tandem with the regional rainfall.

Figure 3.

Temporal march of (a) and (c) Arabian Sea and Bay of Bengal convergence anomalies (106 km2 s−1), respectively. (b) and (d) Rainfall anomalies (1000 mm) over west India coast and Gangetic plains, respectively. Note that both the parameters exhibit an annual cycle in tandem.

[8] In order to gain further insights into this relationship we have constructed a scatter plot of maximum convergence and all-India summer rainfall (Figure 4). An interesting feature is the clustering of the points which suggests that the summer rainfall is predictable when the convergence exceeds a threshold value of 0.5 × 106 km2 s−1. This statement should be interpreted with caution since the analysis under consideration spans only 31 years of data; further analysis with a longer time series is necessary.

Figure 4.

Relationship between maximum convergence (108 km2 s−1) and all-India summer rainfall (mm).

[9] One of the physical mechanisms responsible for this relationship can be briefly explained as follows. The summer monsoon circulation induces a deep convection offshore from the west India coast and enhances latent heat supply from the ocean to the atmosphere [Grossman and Garcia, 1983], which consequently leads to an increase in moisture in the lower troposphere. The low-level atmospheric circulation, significantly the Findlater jet, transports the unstable air mass towards the west coast, where the elevated Western Ghats mountains further enhance the offshore convection by gently lifting the unstable air mass to a zone of higher pressure caused by an upstream blocking of the mountains [Grossman and Durran, 1984]. Large vertical/horizontal wind shears near the mountains decelerate the low level winds and modulate the precipitation.

5. Summary

[10] Surface atmospheric convergence over the Indian Ocean has been mapped for the first time using 31 years of wind data. The spatial distributions reveal that the energetic summer monsoon winds enhance convergence by a factor of two in the central Arabian Sea and eastern Bay of Bengal when compared with weak winter monsoon winds. The summer monsoon circulation is also effective in spatial migration of convergence regions in both basins. Convergence anomaly time series suggests that the variability is higher during July, which occurs in phase with the Indian rainfall. It would be fruitful to continue researching on this topic in order to gain more insights of the monsoon-induced convergence and its role on modulating the rainfall over one of the world's highest populated country. Further analysis could be undertaken by mapping the convergence on higher spatial scales by employing scatterometer winds observed by NASA's SeaWinds on board QuikSCAT with a sufficiently longer time series.


[11] We thank Dr. James J. O'Brien and coworkers for providing surface wind data. Figures were drawn using GrADS. Anonymous reviewer's comments were very useful in revising the earlier draft version.