Intraseasonal oscillation associated with the Indian winter monsoon


  • A. P. Dimri

    Corresponding author
    1. School of Environmental Sciences, Jawaharlal Nehru University, New Delhi, India
    • Corresponding author: A. P. Dimri, School of Environmental Sciences, Jawaharlal Nehru University, New Delhi, 110067, India. (

    Search for more papers by this author


[1] The Indian winter (December–February) monsoon (IWM) contributes almost one third of the annual precipitation over the western Himalayas (WH). In winter, eastward moving synoptic weather systems, or “western disturbances” (WDs), yield precipitation in either liquid or solid form. Although previous studies have examined the interannual variation (IAV) of the IWM, little is known of the intraseasonal oscillation (ISO) associated with the IWM. The present study examines the ISO and its plausible effects on the IWM using 28 years (1980–2007) of precipitation, height, wind, and outgoing longwave radiation (OLR) fields. The dominant ISO mode is found during the active IWM phase with well-defined cyclonic circulation in the midtroposphere over the WH. The relationship between OLR and circulation indicates that this ISO mode is driven by moisture convergence. During the peak phase, a strong moisture influx from the Arabian Sea dominates. This moisture incursion adds to the precipitation over the WH. Successive growth and decay of anomalous cyclonic and anticyclonic circulation takes place within ISO periodicity. Strong convection always seems to precede anomalous cyclonic circulation. In addition, in-phase wind and convection (over the WH region) are associated with the ISO phase.

1 Introduction

[2] During winter (December–February [DJF]), the western Himalayas (WH) receive almost one third of their annual precipitation, brought primarily by eastward moving synoptic weather systems called “western disturbances” (WDs) [Dimri and Mohanty, 2009]. The interplay of these WDs with the heterogeneous topography and variable land use of the WH defines precipitation thresholds [Dimri, 2009, 2012a]. Winter precipitation provides crucial inputs to rivers, glaciers, ecosystems, and habitats in northern India and is therefore important to understand [Lang and Barros, 2004; Ueno, 2005; Ueno and Aryal, 2008]. Compared to the Indian summer monsoon (ISM) [Krishnakumar et al., 1999, 2006; Kuchraski et al., 2007; Saha et al., 2012], the Indian winter monsoon (IWM) [Yadav et al., 2010; Dimri, 2012b] has not been well studied, possibly because the ISM has wider impacts. However, the IWM is linked to the Himalayan climate, cryosphere, hydrology, and socioeconomic conditions and is important to understand in the context of anticipated global climate changes. However, despite its importance, the IWM and characteristics of its subcontinental-scale circulation over the Himalayas are not fully understood [Ueno, 2005]. In addition, most in situ observations in the WH region have been made in either valley bottoms or at windward or leeward sides of ridges, posing limitations for evaluations or assumptions based on in situ observations (Figure 1a).

Figure 1.

(a) Topography (m: shade) and ratio of 0.05° grids with stations (%: contour) over the western Himalayas. The area of 30°N, 72°E to 37°N, 82°E is considered the “western Himalayas” in the present study. (b) Based on 28 years (December 1979, January 1980, and February 1980 to December 2006, January 2007, and February 2007) of data, time series of WHDP climatology (bar; left axis), pentad precipitation climatology (black line with open circles; left axis), and 7–25 d filtered precipitation anomaly (red line with open circles; right axis). The line of 95% significance is also shown, and the period above this corresponds to climatological active phases.

[3] The dominant intraseasonal oscillation (ISO) modes discussed in the literature occur at two timescales: Madden-Julian Oscillation of 30–60 d [Madden and Julian, 1972, 1994; Wheeler and Hendon, 2004] and submonthly-scale oscillation of approximately 6–25 d [Vincent et al., 1998; Hoyos and Webster, 2007; Fukutomi and Yasunari, 2012]. The 10–20 d or quasi-biweekly timescale mode discussed in various studies is contained within the submonthly timescale [Chatterjee and Goswami, 2004; Kikuchi and Wang, 2009] and not discussed here. Many studies have described the role of dominant submonthly-scale ISOs during the ISM [Hartmann and Michelsen, 1989; Goswami and Mohan, 2001; Hoyos and Webster, 2007]. Whereas ISOs of precipitation/convection and associated atmospheric circulations predominate in the ISM, the IWM is associated with the north-south fluctuation of a 200 hPa subtropical westerly jet (SWJ). Variations in precipitation depend largely on the positioning of the SWJ during the winter. The interannual variations (IAVs) in the IWM have been found to be in phase with El Niño-Southern Oscillation forcings [Yadav et al., 2010; Dimri, 2012b]. However, the ISOs associated with the IWM have not been investigated previously.

[4] This study has two main objectives. The first is to illustrate robust features of ISOs associated with the IWM over the WH. The second is to reveal the atmospheric circulation patterns associated with this ISO structure. Section 2 describes the data and analysis methods used in this study. Details regarding ISO characteristics and associated atmospheric circulation and large-scale convection are discussed in section 3. Finally, we present conclusions in section 4.

2 Data and Methodology

[5] We examined data from 28 winters (December 1979, January 1980, and February 1980 to December 2006, January 2007, and February 2007) to study the ISOs associated with the IWM over the WH. Height, wind, stream function, and moisture flux data were obtained from the U.S. National Centers for Environmental Prediction/National Center for Atmospheric Research reanalysis datasets [Kanamitsu et al., 2002]. Long-term satellite observations of outgoing longwave radiation (OLR) provide important information to diagnose the intraseasonal variability in climate in remote regions. We used interpolated OLR data from the U.S. National Oceanic and Atmospheric Administration (NOAA) [Liebmann and Smith, 1996] to identify seasonal and surface dependencies. For precipitation, we used data from the APHRODITE project to assess the IWM over the WH [Yatagai et al., 2009]. These data capture the varying spatial distribution of precipitation over the WH. The WH is composed of deep valleys and high ridges, and therefore, high-quality and representative observations are key initial inputs for any kind of analysis in the region. Information on the number of stations per grid cell is available for APHRODITE data (Figure 1a). This information can be very useful in identifying the extent to which the gridded precipitation values were determined from station data or instead derived using interpolations between the stations. In APHRODITE, daily observation records are collected by meteorological and hydrological organizations in countries throughout the Asian monsoon regions. Because there are very few observation stations in the WH, APHRODITE also uses topographic information (vertical and horizontal) to derive values, which is particularly important for the mountainous Himalayan region. The methods used to construct APHRODITE precipitation reanalysis data have been described by Yatagai et al. [2009]. Many of the data used to prepare precipitation reanalysis data for the WH have been produced “synthetically” by various algorithms, rather than being measured directly. To balance or remove the effect of inadequate representation by in situ observations, we employed a field mask using a 10 mm/month precipitation threshold.

[6] To homogenize and deseasonalize the data, we compiled time series of daily anomalies of precipitation, OLR, and reanalysis data. These anomaly time series were prepared by subtracting the first three harmonics of the annual cycle (about 120 d) for each year. Then submonthly (7–25 d) perturbations were computed by applying a Lanczos filter [Duchon, 1979] to the detrended anomaly time series. Figure 1b shows a time series of WH daily precipitation (hereafter WHDP) for 28 years (bar), the pentad climatology (black curve with open circles), and 7–25 d filtered precipitation (red curve with open circles). The WHDP time series shows clear intraseasonal variation on submonthly timescales. The active and break peaks of the 7–25 d filtered WHDP time series correspond well with those of the unfiltered WHDP time series. For further analysis, an active phase was defined as a period in which the 7–25 d WHDP value exceeded zero within one cycle of the ISO, while its peak value in that cycle exceeded more than 0.5 standard deviation from the climatological mean (Figure 1b). Composites and daily time lag composites were used to investigate the temporal phase relationships of the WHDP fluctuation, circulation fields, and OLR. With reference to Lang and Barros [2004], in the present work, commutation or processes of calculating 140 active and 119 break peak phases are considered, which exceeded 0.5 standard deviation of 28 winter climatological winter. This aspect is very important but is not looked into [Lang and Barros, 2004]. Lang and Barros [2004] have provided 30 years climatology on notch depression but without filtered anomaly. Lang and Barros [2004] have provided a simple composite based on snow minus no-snow days during a season which may not have a well-defined periodicity of IWM (please see Figure 1b, which represents climatological periodicity of IWM).

3 Results and Discussion

3.1 Intraseasonal Variability in WHDP

[7] To show the dominant ISO features in WHDP for the 28 winters of interest, spectrum analysis was performed using a fast Fourier transform (FFT) technique. The FFT analyses used the detrended and masked WHDP (the first three harmonics of the annual cycle were removed) from 15 November to 15 March (120 d). Cosine tapering was applied to 10% of the time series at either end. A three-point running mean in the frequency domain was applied to the raw spectra to estimate errors. Figure 2 shows the FFT spectrum for WHDP for the 28 winters. A pronounced peak appears at approximately 16 d periods, with statistical significance that exceeds the 95% confidence level (Figure 2a). Another peak with a 30–40 d periodicity also appears but is not statistically significant. Figure 2b presents the IAV of the WHDP power spectrum for the 28 years. A 7–25 d periodicity was found in most of the years. A large IAV at a 30–40 d timescale also appeared. Dominant statistically significant peaks often appeared at an approximately 16 d scale in each year (Figure 2b), corresponding closely with the ensemble spectrum (Figure 2a). Therefore, the submonthly-scale ISO is the predominant mode of WHDP. An in-depth analysis is carried out while proposing on ~7–25 d periodicity. It is important to arrive at a decision whether quasi-biweekly (10–20 d) or submonthly (7–25 d) periodicity dominates. Based on detailed analysis, particularly with focus on Figure 2b, sometimes, this periodicity has an upper “significant” extreme until ~25 d. Moreover, it is found that submonthly periodicity dominates more than does the quasi-biweekly periodicity. Keeping this fact in mind, it is proposed to have submonthly periodicity here.

Figure 2.

(a) The 28-winter (DJF) ensemble spectrum of the WHDP time series from 15 November to 15 March (120 d). A red noise spectrum (dashed curve) and its 95% level of significance (solid curve) are also shown. (b) Interannual variation in the WHDP spectrum from 1980 to 2007. The thick black solid line shows the 95% level of significance.

3.2 Corresponding Atmospheric Circulation and Convection Fields

[8] This section focuses on the space-time structure of atmospheric circulation and large-scale convection associated with submonthly-scale precipitation variation over the WH, revealed by composite analyses. Phases that exceeded (above/below) the 28-winter climatological 0.5 standard deviation in 7–25 d WHDP anomalies were considered (active/break) peak phases for composites. This process identified 140 active and 119 break peak phases, which is a sufficient number of samples to reveal robust features in the space-time structure associated with ISO over the WH region.

[9] Figure 3, using the above composites, shows the total atmospheric circulation fields and large-scale convective activity for the peak active and break phases. The statistical significance at each grid point was estimated using Student's t-test. Shaded areas represent OLR and moisture flux divergence fields meeting the 95% confidence level. Associated with the submonthly ISO over the WH, the 200 hPa nondivergent component of flow, represented by stream function, shows lower values in active (Figure 3a) than in break (Figure 3b) peak phases. This lower value corresponds to a lower geopotential height, which indicates suitable conditions for storm or trough formation/intensification. Associated convection anomalies have a negative sign (i.e., enhanced convection) over the entire Himalayan region, with a slightly positive sign (i.e., suppressed convection) over the northwest WH and Indian subcontinent during active peak phases (Figure 3c). Conversely, during break peak phases, suppressed convection is shown all over the Himalayas (Figure 3d). Orographic masking is employed while computing vertical integrated moisture fluxes. Active precipitation/convection was associated with a deepening trough and WDs embedded in large-scale westerlies. In the active peak phases, low-level westerly to southwesterly moisture flows toward high mountain regions in the WH. These westerly to southwesterly flows stagnate on the windward side of the Himalayas, forming a remarkable convergence zone (Figure 3c). The center of active convection was found over the Himalayan region, corresponding to the center of large moisture flux convergence. In contrast, during the break period (Figure 3d), low-level westerlies were slightly stronger and became more southwesterly when approaching the mountainous Himalayan region. In such a situation, incoming moisture fluxes easily flowed out of the WH. This circulation change induced a significant decrease in moisture flux convergence over the entire WH region.

Figure 3.

(a) Composites of total OLR (W/m2; shade) and 200 hPa stream function (*1e–6 m2/s; contour) in a peak active phase for 7–25 d WHDP. (b) Same as Figure 3a, but for a peak break phase. (c) Composites of vertical integrated moisture flux (kg/m/s; vector) and divergence (*1e5/s; shade) in a peak active phase for 7–25 d WHDP. (d) Same as Figure 3c, but for a peak break phase. Values above 95% statistically significant are plotted with shade.

[10] To further illustrate the time evolution of the system, Figure 4 shows the daily-lag composite of vertically integrated moisture flux and divergence from day –8 to day +8 based on 7–25 d precipitation variation over the WH. Active precipitation peaks over the WH are referred to as day 0 phases, whereas day –8 and day +8 correspond to break peaks, indicating a cycle of composite variation of 16 d. These days are used because, as seen in Figure 2, statistically significant peaks often appeared within approximately 16 d periods in each year, and therefore, a large number of composite samples have an approximately16 d period. In each phase, 140 samples were used to construct the daily-lag composites. As illustrated in Figure 4, during the peak break phase, divergence occurs over the WH, with moisture fluxes flowing out of the region. By day –6, convergence starts building up over the Arabian Sea near the Gujarat coast, with fluxes still moving out of the WH. On day –4, a zone of convergence shifts toward the Himalayas with subtle changes in flux direction. By this time, two breakaway flux flows occur, one flowing into the WH region and the other out of the eastern Himalayan region. As the precipitation peak approaches in the active phase, a convergence zone strengthens over the Himalayas with well-defined moisture flux flowing into the region. This situation persists on the peak day with an intense convergence zone only over the Himalayas with moisture influx. By day +2, this situation weakens. Outward flow and divergence start from and over the Himalayas. In the following days, stronger outflow dominates with no moisture presence over the Himalayas.

Figure 4.

Composite of vertical integrated moisture flux (vector, 40 kg/m/s) and divergence (shaded, *1e5/s) based on 7–25 d filtered values of specific humidity and wind anomalous fields from day –8 to day +8 based on WHDP. Day 0 corresponds to the active peak of 7–25 d WHDP variation. Values above 95% statistically significant are plotted with shade.

[11] To explicitly assess the dynamics of the atmospheric structure, daily-lag composites of 500 hPa stream function and wind from day –8 to day +8 based on 7–25 d precipitation variation over the WH are shown in Figure 5. Shaded areas meet the 95% confidence level, as do red-shaded wind vectors. These shaded areas correspond to the strength along with the movement of the systems. On day –8, a well-defined anomalous higher stream function, corresponding to anticyclonic circulation, was seen over the western part of the WH, which, on subsequent days, moved eastward. On day –4, when the anomalous anticyclonic circulation moved eastward and weakened over and around the WH, a large lower stream function, corresponding to anomalous cyclonic circulation, was seen developing over Saudi Arabia. Significant winds preceded (succeeded) the anomalous cyclonic (anticyclonic) circulation. By day –2, this anomalous cyclonic circulation intensified and moved eastward over the Afghanistan and Pakistan region. Associated winds intensified during this phase. On day 0, this system peaked, became most intense, and was located over the WH and adjoining Pakistan region. Interestingly, the center of anomalous cyclonic circulation over and near the WH remained from day –2 to day +2. This outstanding feature of anomalous cyclonic circulation suggests that southwesterly flows play a dominate role in bringing moisture flux from the Arabian Sea during active peak phases. The role of Himalayan topography was also shown in the weakening, stagnation, or change in direction of the southwesterly flow and in significant winds. Figure 6 presents the corresponding upper level, 200 hPa, stream function and wind from day –8 to day +8 based on 7–25 d precipitation variation over the WH. The basic atmospheric structure prevailing in the upper atmosphere is similar to that in the midtroposphere. The major difference is that a larger, elongated, and significant area corresponds to the peak active phase. In the upper troposphere, significant southwesterly winds precede the system. Figures 5 and 6 show that the anomalous cyclonic circulation depth extends from almost the midtroposphere to the upper troposphere. Such significant vertical depth during the peak active phase is associated with stronger southwesterly flow over the Himalayas. In the upper troposphere, a kind of wave train of alternating anomalous anticyclones and anomalous cyclones persists. On subsequent days, the anomalous cyclonic circulation decays, followed by another anomalous anticyclonic circulation, and so on. The upper tropospheric anomalous cyclonic circulation persists longer than that in the midtroposphere.

Figure 5.

Composite of 7–25 d filtered 500 hPa stream function (*1e–6 m2/s) and wind (m/s) anomalies from day –8 to day +8 based on WHDP. Day 0 corresponds to the active peak of 7–25 d WHDP variation. The contour interval for stream function is 5*1e–6 m2/s, and the 95% statistically significant stream function is shaded. The strength of the wind vector is 15 m/s, and the 95% statistically significant wind is plotted with red color.

Figure 6.

Composite of 7–25 d filtered 200 hPa stream function (*1e–6 m2/s) and wind (m/s) anomalies from day –8 to day +8 based on WHDP. Day 0 corresponds to the active peak of 7–25 d WHDP variation. The contour interval for stream function is 5*1e–6 m2/s, and the 95% statistically significant stream function is shaded. The strength of the wind vector is 20 m/s, and the 95% statistically significant wind is plotted with red color.

[12] Figure 7 presents corresponding OLR distribution from day –8 to day +8 based on 7–25 d precipitation variation over the WH (a similar distribution, Figure 5, of 500 hPa stream function and wind is also shown for better assessment). During the break phase, suppressed convection dominates over the Himalayan region, which strengthens with the advancement of the anomalous cyclonic circulation field system. As the anomalous cyclonic circulation deepens by day –4, strong convection builds up over the Himalayas, which continues to intensify and become more localized over the WH. Strong convection is seen in advance, and suppressed in the rear, of the anomalous cyclonic circulation. On day 0, the active peak displays its strongest convection as anomalous cyclonic circulation in the rear persists. Figure 7 suggests that convection always precedes this anomalous cyclonic circulation. This typical characteristic is associated with cloud formation during frontal interface; however, this dimension of frontal formation is not discussed here, as it is beyond the scope of the present study. These convection patterns correspond well with the chain of anomalous cyclonic circulation, followed by another anomalous anticyclonic circulation, and so on.

Figure 7.

Composite of 7–25 d filtered OLR (W/m2; shade), 500 hPa stream function (*1e–6 m2/s), and wind (m/s) anomalies from day –8 to day +8 based on WHDP. Day 0 corresponds to the active peak of 7–25 d WHDP variation. The contour interval for stream function is 4*1e–6 m2/s. The strength of the wind vector is 5 m/s.

[13] To illustrate the periodic occurrences of alternating anomalous anticyclonic and cyclonic circulations forming a kind of wave structure, Figure 8a presents time-lag composites based on 7–25 d active peak phases, where the clear evolution of cyclonic systems is seen. The systems move as a “family,” with one following the other in a symmetric wave that peaks at day 0, followed by decay. Associated convection started building up by day –6 and peaked on day 0. The corresponding 500 hPa height indicated that storm intensification started by day –6 and/or day –4, and an intense storm formed by day 0. Interestingly, the associated 500 hPa wind showed symmetrical phase movement associated with each cyclonic storm. Every storm peaked in a similar phase of wind. Total precipitation was highest on day 0, and corresponding 7–25 d filtered precipitation also showed a similar anomalous distribution. This baroclinic structure became a dominant mechanism in storm intensification and had an ISO character while storms evolved. To further illustrate convection and associated clouding, Figure 5b represents 7–25 d filtered OLR during the winter. Eastward propagating convective activity is shown to have a submonthly scale with a phase speed of ~2°/d. This corresponded to a strong ISO associated with a winter storm contributing to WHDP during the IWM.

Figure 8.

(a) Time-longitude cross-sectional distribution at 35°N of composites for 7–25 d filtered OLR (W/m2; shade), 500 hPa height (hPa; contour), 500 hPa wind (m/s; arrow), and precipitation (mm/d; red curve; upper axis) anomalies from day –8 to day +8 based on WHDP. Day 0 corresponds to the active peak of 7–25 d WHDP variation. The yellow curve corresponds to anomalous precipitation (mm/d; yellow curve; upper axis) from day –8 to day +8 based on WHDP. (b) Time-longitude cross-sectional distribution of 7–25 d filtered OLR (W/m2) anomaly at 35°N based on 28 winters (December 1979, January 1980, and February 1980 to December 2006, January 2007, and February 2007).

4 Conclusions

[14] This study examined ISO characteristics associated with the IWM. Winter storms associated with WHDP have a certain pattern of evolution corresponding to their intensification and decay. We found that these storms have an average life cycle of approximately 10–12 d. The precipitation spectrum shows the dominance of approximately 16 d periodicity associated with WHDP. Correspondingly, a cyclonic storm evolves with peaks and decays. During the active peak phase, similar convective activities in association with westerly to southwesterly enhanced moisture flux, providing higher precipitation. Convection seems to always precede anomalous cyclonic circulation. A succession of cyclonic-anticyclonic-cyclonic, and so on, circulation dominates in the SWJ. A detailed analysis of ISO characteristics of the IWM in association with the 200 hPa SWJ is planned for future research.


[15] Author acknowledges the guidance of Prof. T. Yasunari and discussions with Dr. H. Fujinami, Nagoya University, Nagoya, Japan, during the preparation of the manuscript.