Climatic Teleconnections Associated with a Western Himalayan River: !"Melt Period Inflow into the Bhakra Dam in India #"

$" %"

Indrani Pal1*, Upmanu Lall2, Andrew Robertson1, Mark A Cane3, Rajeev Bansal4 &" '" ("

1 International Research Institute for Climate and Society, The Earth Institute at Columbia University, )"Palisades, NY 10964 *"

2 Department of Earth and Environmental Engineering, Columbia University, New York, NY 10027 !+"3 Lamont-Doherty Earth Observatory, Columbia University, Palisades, NY 10964 !!"

4 Bhakra Beas Management Board, Punjab, India !#" !$" !%" !&" !'" !(" !)" !*" #+" #!" ##" #$" #%" #&" #'" #(" #)" #*"

""""""""""""""""""""""""""""""""""""""""""""""""""""""""*Corresponding author information: Indrani Pal, International Research Institute for Climate and Society, The Earth Institute at Columbia University, 143 Monell, 61 Route 9W, Palisades, NY 10964. E-mail – indrani@iri.columbia.edu. Phone: +1 845 680 4508. Fax: +1 845 680 4866.

Abstract $+" $!"Understanding the relationship between low frequency variations in large-scale $#"atmospheric circulation driven by the oceans and high latitude hydro-climatic variability $$"at regional scales is important to assess the projections of climate change in these regions, $%"and for designing measures for climate change adaptation. Here we analyze the recent $&"(1978-2004) variability in the melt period river flow into the Bhakra dam in India through $'"correlation and composite analyses of associated large-scale climatic patterns in winter. $("Bhakra total inflow is a sum of the Satluj River flow and the volume of flow diverted $)"from the Beas River through the Beas Satluj Link (BSL) channel. These flow components $*"are highly correlated with each other. The spring (March-June), melt period inflow of %+"Bhakra dam is very well correlated with the winter precipitation and temperature of the %!"Satluj basin. Spring seasonal inflow anomalies are primarily connected to the winter %#"atmospheric circulation patterns over the Western Himalayas and adjoining north and %$"central Indian plains, which are in turn linked to the fluctuation of equatorial winter Sea %%"Surface Temperature over the western Indian Ocean and the central Pacific Ocean. They %&"also appear to be modulated by the changes in the winter pressure fields over Indonesian %'"Throughflow, the connecting point of Indian and Pacific Oceans. Low spring inflows are %("associated with negative temperature anomalies over Siberian High and central to eastern %)"Pacific and a negative pressure system over the South-east Asia, Indonesian Seas, eastern %*"Indian Ocean and western Pacific, and a positive pressure system over Iceland and Arctic &+"during winter corresponding to weaker westerlies. On the other hand, a high inflow in &!"spring is linked with a deep low-pressure system and convergence of wind from two &#"opposite directions over the North Atlantic Ocean, and stronger westerlies. &$"Teleconnections of spring Bhakra inflow with the IOD, SOI and Nino climate indices in &%"winter are also noted, such that positive winter IOD and ENSO events (positive pressure &&"field over east Indian Ocean and west Pacific) are related to higher winter precipitation &'"over the Western Himalayas leading to higher spring inflow of Bhakra. These &("teleconnections provide some promise for improving the long lead prediction capability &)"for inflows into this important multi-purpose reservoir that is operated at seasonal time &*"scales for irrigation, flood control and hydropower production. They also provide '+"

potential directions for the statistical downscaling of precipitation, temperature and/or '!"streamflow directly from large-scale climate model simulations of ocean temperature and '#"atmospheric pressure fields. '$" '%"Keywords – Spring Inflow, Bhakra dam, Satluj River, Winter Precipitation, Western '&"Himalaya, Teleconnection ''" '("1. Introduction ')" '*"The major river systems of the Indian subcontinent that originate in the Himalayas are (+"expected to be very sensitive to climate change because of substantial contributions from (!"the snow and glacier melt (Singh and Jain, 2002). Winter precipitation in the form of (#"snow over the Western Himalayas feed the glaciers, which serve as a vast storehouse of ($"freshwater for the Indus river basin, which provides the primary water supply for the (%"breadbaskets of India and Punjab. The supply of water through these rivers is important (&"for ecological habitats, power generation and irrigation in the dry pre monsoon (March-('"June) season. Winter precipitation and temperature are key determinants of spring flow of (("Satluj and Beas Rivers, both of which originate from the Western Himalayas at different ()"elevations. Through a diversion (Beas Satluj Link/BSL), the Beas River is one of the (*"major tributaries of Satluj River. Both are major tributaries of the Indus River; together )+"with another tributary Ravi carrying about 1/5th of annual Indus River flow (~ 112 km3) )!"(Sen Gupta and Desa, 2001). )#" )$"The Bhakra dam across Satluj River in north India is the major point of water supply )%"(~22000-35000 cfs and irrigation to 10 million acres of land) and electricity generation )&"(1325 MW) for the neighboring states of Punjab, Rajasthan and Haryana, including the )'"national capital territory of Delhi (different web sources). The irrigation canal systems )("connected with Satluj River and Bhakra dam in India turned the Punjab into the ))"breadbasket of the country, providing the agrarian economic foundation for the arid )*"provinces and feeding the majority of the populations approximately since early 1960s. *+"Bhakra inflow is a joint contribution of the flow from Satluj River and Beas Satluj Link *!"

channel, which came into effect in 1977. Thus, winter precipitation in the Western *#"Himalayas in the form of snow has the major contribution to the total volume of melt *$"period (spring season) inflow of Bhakra. Melting snow and ice provides the water supply *%"to much of the north Indian regions under Bhakra command during the dry months before *&"the summer monsoon. *'" *("Winter precipitation in the Western Himalayas is mainly associated with the mid-latitude *)"jet stream and low-pressure synoptic systems known as Western Disturbances (WD) **"(Dimri, 2005, 2006; Yadav et al., 2009). WDs originate over the Mediterranean Sea or !++"North Atlantic Ocean, with secondaries developing over the Persian Gulf and Caspian !+!"Sea either directly or as a result of the arrival of low-pressure systems from southwest !+#"Arabia, and travel eastward over Iran, Afghanistan, Pakistan, and northwest India (Dimri, !+$"2007). In the winter months the midlatitude depressions move to their lowest latitudes !+%"and in their pathway travel across the north and central parts of India in a phased manner !+&"from west to east, disturbing the normal circulation patterns (Yadav et al., 2009, 2010). !+'"Past theoretical and synoptic studies indicate that the development of WDs in the mid-!+("latitudinal synoptic system is associated with baroclinic activity and therefore potential !+)"energy residing in the latitudinal temperature gradient is the main source of energy (Rao !+*"and Rao, 1971; Dimri et al., 2004). !!+" !!!"Several diagnostic studies have focused on wintertime circulation patterns during the !!#"years of extreme precipitation events (deficit and surplus) over Western Himalayas !!$"considering the region of 15oS-45oN and 30oE-120oE (Dimri et al., 2004; Dimri, 2005, !!%"2006; Yadav et al., 2009, 2010). The larger-scale land-ocean interaction and !!&"teleconnections and especially the effects of Atlantic, Pacific and Indian Ocean SST and !!'"pressure fields in developing and modulating the WD events, have not been as well !!("studied. While the intensity of an individual WD event determines the strength of a !!)"particular winter precipitation event (Dimri et al., 2004), the average volume of March-!!*"June (MAMJ) inflow of Bhakra reservoir is a function of average winter season !#+"precipitation amount that reflect the cumulative result of a series of individual WD events !#!"(Pal et al., 2011). Therefore, we want to explore the variability of spring total inflow of !##"

Bhakra dam due to large-scale atmospheric circulation patterns affecting the interannual !#$"variability of volume of winter precipitation over Western Himalayas. !#%" !#&"Oceans are the largest heat storage on a global scale and hence provide the interannual !#'"and longer time scale boundary conditions for short-term evolution of atmospheric !#("circulation. The Indian Ocean near equatorial latitudes is an active convective region in !#)"winter. Therefore, an increase in Sea Surface Temperature (SST) increases the !#*"convection in this region resulting in more latent heat release that enhances the !$+"temperature of the atmosphere up to mid-tropospheric levels (Yadav et al., 2007; Barlow !$!"et al., 2007). This increase in mid-tropospheric atmospheric temperature in the equatorial !$#"Indian Ocean increases the meridional temperature gradient intensifying and shifting the !$$"sub-tropical jet stream towards lower latitudes as a consequence of thermal wind balance !$%"(Yadav et al., 2007). The jet stream increases the upper level divergence equator wards !$&"due to increased horizontal shear favoring strengthening of cyclogenesis over the Indian !$'"region (Yadav et al., 2007). Therefore, a variation in equatorial Indian Ocean SST may !$("have an impact on winter precipitation in the Himalayas and hence on the total spring !$)"inflow of Bhakra. !$*" !%+"The ENSO (El-Nino Southern Oscillation) is a coupled ocean-atmosphere phenomena !%!"that has strong influence on the climate in the Northern Hemisphere, particularly in peak !%#"boreal winter. ENSO is phase locked to the annual cycle with central to eastern Pacific !%$"SST anomalies developing during boreal summer, peaking in winter, and decaying in the !%%"following spring (Schott et al., 2009). ENSO events are also characterized by a large !%&"seesaw of sea level pressure between the eastern and western tropical Pacific, which is !%'"referred to as the Southern Oscillation (SO). During El Nino, atmospheric convection !%("shifts eastward and intensifies over the central to eastern equatorial Pacific, which alters !%)"atmospheric circulation remotely in both tropics and sub-tropics via atmospheric wave !%*"adjustments (Yadav et al., 2010). The tropical Indian Ocean gradually warms during an !&+"ENSO event (Schott et al., 2009). An increasing influence of ENSO on winter !&!"precipitation over northwest India has been noted (Yadav et al., 2009, 2010) in the recent !&#"decades starting in early 1980’s coincident to an increase in the frequency of stronger !&$"

ENSO events (Herbert and Dixon, 2002). !&%" !&&"Based on the review presented so far, if the latitudinal or meridional temperature/pressure !&'"gradient is modified by the large scale climate phenomena such as ENSO or Indian !&("Ocean or North Atlantic mechanisms, which may all be related, it is expected that the !&)"winter precipitation and temperature will vary in the study region and will be modified, !&*"and this will impact the magnitude of the MAMJ flow volume. Hence, in this study we !'+"explore how the interannual variability of the spring (MAMJ) inflow of Bhakra reservoir !'!"over Satluj River relates to the large-scale winter atmospheric circulations, and related !'#"climate precursors. !'$" !'%"2. Study Area and Data !'&" !''"The Satluj River basin upstream of Bhakra dam is the prime area of interest here. Satluj !'("River is a major tributary of the Indus River. It’s source is the Rakas lake in Tibet on the !')"southern slopes of Mount Kailash and it flows generally west and southwest entering !'*"India through the Shipki La pass (altitude 6608m) in Himachal Pradesh. It leaves !(+"Himachal Pradesh to enter the plains of Punjab at Bhakra, where the world’s highest !(!"gravity dam has been constructed on this river in 1963. The Satluj finally drains into the !(#"Indus in Pakistan. The upper territories of the Satluj River catchment area of about !($"50,140 km2 are located above the permanent snowline at an altitude of 4500m. The Beas !(%"River is the second easternmost of the rivers of the Punjab, a tributary of Satluj. The river !(&"rises at an elevation of 4361m at Rohtang Pass of central Himachal Pradesh, India, and !('"flows for some 470 km to the Satluj River in the Punjab state of India. !((" !()"The Satluj River basin experiences two major high flow seasons, melt/pre-monsoon and !(*"monsoon, which are not always distinct. Melt season is primarily MAMJ when the flow !)+"comes from the runoff generated by snowmelt from the Western Himalayas. The !)!"monsoon season is primarily July-August-September-October (JASO) when the monsoon !)#"rainfall in the lower elevations and the glacier melt from the upper elevations !)$"predominantly contributes to the flow. !)%"

Observed hydro-climatic data, including daily total inflow data of Bhakra dam (1963-!)&"2004), daily net Satluj River flow (Bhakra inflow minus BSL flow for 1963-2004), daily !)'"BSL inflow (1978-2004), daily-accumulated snow data at 12 stations (1976-2006), daily !)("1o gridded precipitation (1963-2004 for 30-33oN and 76-79oE) and temperature (1969-!))"2004 for 30-33oN and 76-79oE) were procured from the Bhakra Beas Management Board !)*"(BBMB for flow and station snow data) and Indian Meteorological Department (IMD for !*+"meteorological data, Rajeevan et al., 2005, 2006) respectively. Since the Satluj River !*!"started receiving additional volume of flow through the Beas Satluj Link (BSL) channel !*#"from the late 1977, a step change occurred in the Bhakra inflow data 1978 onwards (Fig !*$"1). Therefore, we considered the total inflow of Bhakra (hereafter inflow) from 1978-!*%"2004. !*&" !*'"For spatial correlation analysis and for examining the teleconnections we extensively !*("utilized the tools and datasets available at the KNMI Climate Explorer (http://climexp. !*)"knmi.nl) and IRI data library (http://iridl.ldeo.columbia.edu/), notably the Met Office !**"Hadley Centre Sea Ice and SST data set version 1 - HadlSST1 (available for 1870-now), #++"GISS 1200km gridded temperature data (1880-now), CRU TS 3.0 i.e University of East #+!"Anglia Climatic Research Unit (CRU) global 0.5° monthly time series for daily #+#"maximum (Tmax), minimum (Tmin), mean surface temperature and diurnal temperature #+$"range (DTR) (available for 1901-2006), and HadSLP2 reconstruction dataset for mean #+%"sea level pressure (MSLP) (1850-now). In addition, we also used gridded atmospheric #+&"circulation variables from the NCEP/NCAR reanalysis V1.0 products from the KNMI #+'"Climate Explorer, and moisture flux from the IRI data library for composite analyses. #+("Finally, various well-known teleconnection indices such as Nino3.4, Nino4, Southern #+)"Oscillation Index (SOI), AMO, DMI, NAO, and AO were also used from the same source #+*"directly. #!+" #!!"Fig 1 shows a contour plot (years in x axis, days in y axis and flow in cfs in z axis) #!#"displaying the seasonality of the daily Bhakra inflow, where the days from 63-187 marks #!$"MAMJ flow pattern. There is a clear spread (i.e. step change in flow due to BSL #!%"diversion) visible from 1978 onwards. Approximately 32% of average annual inflow of #!&"

Bhakra comes in MAMJ season, which is primarily dominated by the snowmelt. Since #!'"monsoon onset usually takes place at the end of June or early July, #!("(http://es.wikipedia.org/wiki/Archivo:India_southwest_summer_monsoon_onset_map_en#!)".svg), June was the last month considered for melt period analysis. #!*" ##+"3. Results and Discussions ##!" ###"On average, the total Bhakra inflow volume comprises of 65-80% of net Satluj River ##$"flow and 20-35% Beas Satluj Link diversion from the Beas River. As a result, an analysis ##%"of Bhakra inflow and net Satluj River flow for 1978-2004 yielded a Pearson product-##&"moment correlation coefficient (hereafter correlation coefficient) of 0.995 for monthly ##'"and 0.993 for seasonal (MAMJ) data; and the same with BSL diverted flow volume was ##("0.84 for monthly and 0.80 for seasonal (MAMJ) data. High correlation was also noticed ##)"for net Satluj River flow and BSL flow, 0.72 for MAMJ and 0.78 for monthly data from ##*"1978-2004. #$+" #$!"3.1. Interannual Variability of Total Bhakra Reservoir Inflow #$#" #$$"Bhakra inflow in the winter months (DJF) is usually the lowest among all the seasons and #$%"is mainly contributed by the low elevation rainfall (corresponding to higher elevation #$&"snow) over the basin during that time, as manifest in the positive correlation of average #$'"DJF inflow and average DJF low elevation rainfall (0.56) and no correlation with the #$("higher elevation DJF snow. Due to the time lag between a rainfall event and the surface #$)"or subsurface water reaching the river, the correlation for the Jan-Feb inflow versus Dec-#$*"Jan (DJ) rainfall in the low elevations (0.72) and DJ snow in the high elevations (0.43) is #%+"higher. Our study on the variations of the first derivatives of daily total inflow found that, #%!"snow deposited in the lower elevations in the months of Dec and Jan, starts melting, on #%#"average, at the end of January (~ 25th Jan) whenever the temperature is > 0oC particularly #%$"in the lower elevation areas. The date (measured as the number of days from 1st Jan every #%%"year to first zero crossing of the first order derivatives of daily total inflow preceding a #%&"continuous positive value of the same to the peak, Fig 2) is significantly negatively #%'"

correlated with the average maximum daily temperature (Tmax) in Jan, which means that #%("the date varies with Tmax and an increase in average Tmax in Jan makes the melt starting #%)"date earlier or vice versa. Total Bhakra inflow comprises of a number of such events #%*"(e.g. three such zero crossings of first derivatives of daily total inflow) in January-June, #&+"as also shown in Fig 2. By contrast, JASO has one big streamflow event due to #&!"monsoons. Therefore, the number (and volume) of springflow events in Jan to June #&#"depends on the fluctuation of local melting temperature or sometimes an occasional #&$"rainfall event or both. Hence, average MAMJ flow has no significant correlation (Pearson #&%"or rank) either with average MAMJ rainfall, or average MAMJ temperatures. #&&" #&'"Surface temperature in the winter months drops due to cooling brought about by #&("excessive precipitation, particularly snow albedo (Bush, 2000), and sweep of cold air #&)"advection in rear end of the storms (Yadav et al., 2007), as evident in negative significant #&*"correlations between average precipitations (rainfall and snow) and temperatures. #'+"Therefore, negative significant correlation was also noticed between both winter (DJF) #'!"and spring (MAMJ) total inflow volume and winter (DJF) temperatures. While snow gets #'#"deposited in DJFM months, average snow available in FMA over the whole Satluj River #'$"basin relates to the interannual variability of average MAMJ total Bhakra inflow, which #'%"is apparent in the positive significant correlation between MAMJ flow and FMA snow #'&"(0.72) that is higher than the correlation with average measured snow in DJFM (0.58). #''"Therefore, average accumulated snow in FMA and MAMJ inflow both relate to #'("atmospheric circulation patterns in DJF (as discussed in later sections). MAM months are #')"usually dry in terms of rainfall in the Satluj basin (with monsoon rainfall starting in late #'*"June or early July); but snow cover is observed until April especially in the higher #(+"elevations. March usually has the maximum snow cover and therefore average snow #(!"available in FMA yields the highest correlation with the average MAMJ flow. In #(#"addition, some of the snow deposited in DJ would already have melted in JF and #($"therefore the correlation with DJFM snow reduced. Therefore, average volume of snow #(%"in FMA is the prime determinant of the average volume of MAMJ flow and any variation #(&"in winter snow will have significant impact on the variation of inflow of Bhakra in the #('"following season. #(("

A trend analysis of the monthly and seasonal total inflow data from 1978-2004, and #()"selected local climate variables based on the non-parametric Mann-Kendall test with #(*"Sen’s method (Sen, 1968) revealed no significant trends in total MAMJ Bhakra inflow or #)+"rainfall or snow or Tmax in the region. There are trends in Tmin and DTR in some #)!"months but these do not seem to impact the seasonal inflows. A statistically significant #)#"decrease in July flow is noted. Its causes need further investigation. #)$" #)%"3.2. Spatial Correlation Patterns between Total Bhakra Reservoir Inflow and #)&"Winter Large-scale Atmospheric Circulation #)'" #)("Regional precipitation and surface temperature fields over the Satluj River basin in the #))"Western Himalayas respond to large-scale circulation patterns. We explore these spatial #)*"patterns through correlation and composite analyses. This will help us to identify the #*+"possibility for downscaling climate change or seasonal climate forecasts, and/or to #*!"explore how trends in circulation patterns relate to trends in surface variables and total #*#"inflow into Bhakra. #*$" #*%"Average MAMJ Bhakra inflow is positively correlated with the SSTs over the equatorial #*&"central/western Indian Ocean and southern Arabian Sea adjacent to the Somalia coast, #*'"and equatorial central Pacific, as shown in Fig 3(a). The strong positive anomalies of the #*("southern Arabian Sea, as noticed in Fig 3(a), are related to Indian Ocean anomalies #*)"(Vinayachandran and Kurian, 2008). These results are supported with the fact that winter #**"rainfall over southwest Asia and particularly the northwest India fluctuates with winter $++"SSTs over the equatorial Indian Ocean and central Pacific, and winter MSLP over eastern $+!"tropical Hemisphere (Barlow et al., 2007; Yadav et al., 2010), which translates into $+#"significant correlation between winter (DJF) SST and MSLP fields versus lower $+$"elevation rainfall and upper elevation snow in the Western Himalayas since the same $+%"climate mechanism causes winter precipitation in these regions. Positive SST anomalies $+&"over the equatorial western Indian Ocean and central to eastern Pacific signify the Indian $+'"Ocean Dipole (Saji et al., 1999) and warm phase of ENSO events respectively, which are $+("also related to each other (Yamagata et al., 2004). Indian Ocean Dipole (IOD) is defined $+)"

as anomalous positive SST over western IO and negative SST over east (Saji et al., $+*"1999). High SST in the western Indian Ocean (IO) increases local convection, altering $!+"Walker Circulation over IO and causing subsidence (the descent of a body of air in a $!!"high-pressure area) over the tropical eastern IO and Indonesian Seas, also called $!#"Indonesian Throughflow that is the connecting point of two oceans (Annamalai et al., $!$"2010). The warm oceanic phase, El-Niño, accompanies high air surface pressure in the $!%"western Pacific, while the cold phase, La-Niña, accompanies low air surface pressure in $!&"the western Pacific. During a positive IOD event the anomalous coastal upwelling due to $!'"strong southeasterly winds along the Java coast causes further SST cooling in the eastern $!("IO, further enhancing the IOD. This anomalous atmospheric subsidence induces drought $!)"over Indonesia (Yamagata et al., 2004). Therefore, a positive anomaly in the pressure $!*"field over the western Pacific and the region of Indonesian Throughflow (ITF) is due to $#+"SST forcing, which is teleconnected with the winter precipitation over Western $#!"Himalayas (not shown), and therefore for the same reason with the spring seasonal $##"Bhakra inflow, as displayed in Fig 3(b). $#$" $#%"As mentioned earlier, Fig 3(a) and (b) show winter IOD and ENSO signature associated $#&"with the average MAMJ Bhakra inflow. A direct correlation analysis with various Nino $#'"indices and SOI, and Indian Ocean Dipole Mode Index (DMI) also support this. Table 1 $#("shows correlations that are statistically significant for selected climate indices. The $#)"correlation with the SOI was the highest (negative) in February and that with the DMI $#*"was the highest in December. Of the “Nino” indices, the Jan Nino-3.4 and the Nino-4 for $$+"Jan/Feb are best correlated with the MAMJ Bhakra flows. The joint influence of the IOD $$!"and ENSO on Western Himalayas winter precipitation is consequently notable. In $$#"addition, a correlation analysis with different climate indices over Atlantic and Arctic $$$"regions for the winter months showed that Atlantic Multidecadal Oscillation (AMO) in $$%"February is negatively correlated with the MAMJ inflow and NAO and AO are weakly $$&"but positively correlated (not shown). Given the long memory associated with these $$'"indices and the short record used here, these associations were not explored further. $$(" $$)" $$*"

3.3. Low/High Total Bhakra Reservoir Inflow Composites Associated with Large-$%+"scale Winter Circulations $%!" $%#"Composite analyses were conducted to identify the large-scale climatic states associated $%$"with the high/low volume of spring seasonal total inflow of Bhakra reservoir. The $%%"average MAMJ inflow volumes below the 25th percentile in the 27 years (1978-2004) $%&"were categorized as low inflow and the same above the 75th percentile were categorized $%'"as high inflow. High/low spring seasonal inflow volume is a result of high/low winter $%("precipitation (mainly snow in FMA) as a result of DJF circulations, as mentioned earlier. $%)"Since the low/high inflow years are small in number, the results discussed here are more $%*"indicative than conclusive. $&+" $&!"Fig 4(a) shows negative low-inflow composites with the MSLP anomalies over the warm $&#"pool region of Indonesian Throughflow, western Pacific Ocean and eastern Indian Ocean $&$"and Fig 5(a) shows negative low-inflow composites with temperature over the central and $&%"eastern pacific indicating a La-Nina behavior. On the other hand, for the high inflow $&&"composites, opposite of these phenomena wasn’t noticed in Fig 4(b) and 5(b); however, $&'"positive inflow composites were found with MSLP over the region of Indonesian $&("Throughflow and temperature over the central to eastern Pacific region when we defined $&)"the ‘high’ inflow as inflow greater than long-term 50th percentile (not shown). These $&*"signify positive (negative) ENSO and IOD effects on the above average/high (low) $'+"inflow of Bhakra. $'!" $'#"The pressure field over the east Indian Ocean, western Pacific and their connecting point $'$"Indonesian Throughflow is related to the winter precipitation over the Western $'%"Himalayas and thus the spring seasonal inflow of Bhakra dam (Fig 3(b)). Previous studies $'&"indicated that this pressure field exerts significant control over the global climate through $''"its effects on the atmospheric pressure in the tropics and Northern Hemisphere via $'("teleconnections (Schneider, 1998; Barlow et al., 2007). The Indonesian Throughflow $')"(ITF) is widely known, on average, to carry warm and fresh Pacific waters through the $'*"Indonesian archipelago into the Indian Ocean in winter (Gordon et al., 2003). Therefore, $(+"

any variation in net heat input to the Indian Ocean through the Indonesian Throughflow $(!"would remotely affect the winter precipitation over the Western Himalayas due to its $(#"potential effect on the Indian Ocean dipole. In addition, during an El-Niño event in $($"winter, western Indian Ocean (and Arabian Sea) tends to be warmer (and low pressure $(%"region) than the eastern part, which may enhance the dipole effect and the flow of the $(&"upper Indian Ocean is usually directed westward from near the Indonesian Archipelago $('"towards the Arabian Sea and western Indian Ocean, which is the most important moisture $(("source for the winter precipitation over Western Himalayas (Sakal et al., 2010). On the $()"contrary, the heat transfer through Indonesian Throughflow to eastern Indian Ocean may $(*"negatively affect this mass transfer from east to west and also jeopardize this ENSO-IOD $)+"tie making it contradictory and non-linear. This non-linear relationship was recently $)!"recognized by a study that also observed an association of La-Niña and negative IOD $)#"events in winter (Sakal et al., 2010). During a negative IOD event in winter, we also $)$"expect lesser amount of winter precipitation (and inflow in spring) to happen in the study $)%"region, as also evident from the positive correlation in Table 1 and low-inflow-surface $)&"temperature composite field in Fig 5(a). This is because the atmospheric convection shifts $)'"due to Indian Ocean SST changes and affecting the jet stream (Yadav et al., 2007). $)(" $))"Fig 6(a) and (b) shows the average winter latent heat (LH) and moisture flux captured for $)*"the low and high inflow years. The figure shows LH in color and arrows show moisture $*+"flux. Latent heat (LH) relates to convection and moisture sources. Purple color means $*!"more evaporation and blue more precipitation. In the high inflow years there is more $*#"evaporation over the Arabian Sea and a stronger southerly jet to the North Western India $*$"including the Western Himalayas relative to the smaller inflow year, and a better-$*%"developed cyclone is also noticed over Southern Arabia feeding moisture in both from $*&"Caspian Sea and the branch from the Arabian Sea in the high inflow year. $*'" $*("Large-scale wind currents determine winter weather over the mid-latitude North Atlantic $*)"Ocean. Near Iceland, when the atmospheric pressure is low, regional air flows in a $**"counterclockwise direction but air flows clockwise around the Azores, which is a high-%++"pressure area at the same time. The winter jet stream intensity and position across the %+!"

North Atlantic and over Western Europe is determined to a great degree by the strength %+#"and sign of these pressure centers that are often used as indicators of the North Atlantic %+$"Oscillation. The frequency, position and intensity of the western cyclonic disturbances %+%"are associated with this large-scale structure. Weaker (stronger) westerlies tend to reduce %+&"(increase) winter precipitation in the Western Himalayas and Northwest India. The zonal %+'"wind stress pattern is closely related to the variability of the surface pressure field and an %+("important decisive factor of atmospheric moisture transport. Positive zonal wind stress %+)"occurs at the vicinity of lower SLP (convergence zone/trough) and vice versa (Nobre and %+*"Shukla, 1996). Two opposite direction wind currents (Fig 7(b)) and their convergence %!+"zone corresponding to a deep pressure system in the North Atlantic (Fig 4(b)) forms %!!"strong westerlies that tend to bring western disturbances. Weaker westerlies are a result %!#"of higher MSLP anomalies (weaker wind stress) near Iceland, which is noted in Fig 4(a) %!$"and Fig 7(a). %!%" %!&"Another winter phenomenon that may also be playing a role in modulating winter %!'"precipitation patterns over the Western Himalayas is the northeast monsoon that is also %!("related to the jet stream. In Southern Asia, the northeastern monsoons take place from %!)"December to early March when the northern hemisphere land surface high-pressure %!*"system is strongest (Fig 4(a)). The subtropical branch of jet stream directs northeasterly %#+"winds that bring cold air from the ‘Siberian High’ to blow across Southern Asia, creating %#!"dry air streams, which produce clear skies over India (also in Fig 6). Meanwhile, when a %##"low-pressure system develops over South-East Asia and Australasia (as mentioned %#$"before), winds are directed towards Australia known as a monsoon trough and cause high %#%"rainfall over those regions. Strong negative temperature anomalies over Siberian High %#&"region (Fig 5(a)), high-pressure anomalies over eastern Arctic north of Siberia, and low-%#'"pressure region over South-East Asia and Australasia (Fig 4(a)) thus explain low-inflow %#("(Fig 6(a)). Therefore, low inflow (low winter precipitation) may happen in a year when %#)"there is northeastern winter monsoon induced high precipitation (and flooding) in %#*"Australia and South-east Asia, which is also linked with the Pacific La-Nina events, as %$+"was in news recently in 2010-11 (http://portal.iri.columbia.edu/Production/Home/assets/ %$!"imgs/LaNinagraphic.jpg) and also evident in Fig 6(a) and (b). On the other hand, a higher %$#"

temperature anomaly over Siberian High region (Fig 5(b)) and lower pressure anomalies %$$"over northern Eurasia (Fig 4(b)) are also associated with the high inflow when %$%"northeastern winter monsoon tends to be weaker. %$&" %$'"Surface flow associated with Siberian High or winter anticyclone is directed along the %$("elevated terrain in northeastern China and extending west adjacent to the Tibetan plateau %$)"(Jones and Cohen, 2010). We found a negative significant correlation (-0.47) between %$*"winter rainfall (and snow) in winter from 1963-2004 with winter ‘Siberian High’ central %%+"intensity, defined as the regional mean SLP averaged over 700E-1200E and 400N-600N, as %%!"suggested by Gong and Ho (2002). Therefore, lower/higher temperature (higher/lower %%#"winter atmospheric pressure) over the Siberian high-pressure zone goes with less/high %%$"winter precipitation over Satluj River basin over Western Himalayas, which contributes %%%"to lesser/higher volume of river flow in spring season and vice versa. This was consistent %%&"with the high/low flow composites with the T2/10m (Fig 5). Based on the 79 years data %%'"(1922-2000), Gong and Ho (2002) showed that cooling occurs over the middle to high %%("latitude Asia and Eurasia when the Siberian High strengthens in winter (JFM). However, %%)"they found warming over north and northwestern plains of India including the vicinity of %%*"the Western Himalayas and in middle Asia. They also found negative correlation of %&+"Siberian High with the winter precipitation over the same region in India; this different %&!"winter precipitation pattern from the temperature is consistent with our studies as well %&#"(Pal et al., 2011) because the precipitation and storm activities in winter reduces %&$"temperature. Thus, change in Siberian temperature and wind pressure fields is also %&%"remotely connected with the intensity of winter precipitation over Western Himalayas. %&&" %&'"Furthermore, during a drier (wetter) winter, the average wind stress in the North Arabian %&("Sea is negative (positive) signifying reduced (increased) moisture transport from there by %&)"the westerlies/WD for Western Himalayan/Northwestern India (Yadav et al., 2010), as %&*"also noted in Fig 6 and7. Strong positive low inflow and zonal wind pressure composites %'+"over eastern Indian Ocean in Fig 7(a) signifies strong eastward wind and that over %'!"tropical Pacific is towards west, which was found to be exactly opposite for the higher %'#"average inflow (not shown). All the high/low inflow composites discussed here agree %'$"

with the low/high snow (FMA) composites (not shown). %'%" %'&"4. Summary %''" %'("Anomalous coupled ocean-atmosphere phenomena generated in the tropical oceans %')"produce global atmospheric and oceanic circulation changes that influence regional %'*"climate conditions even in remote region and eventually regional hydrology. This study %(+"looked at the annual variability and teleconnections of melt dominated Bhakra reservoir %(!"inflow across Satluj River that originates from the Western Himalayas and flows over the %(#"northwestern plains of India. The analyses were based on the correlations and composites %($"using observed and reanalysis fields procured from different sources for 1978-2004. We %(%"have found that the volume of inflow of Bhakra reservoir in spring season is strongly %(&"correlated with the winter precipitation (and temperature) over the Satluj River basin, %('"which is modulated with the tropical winter SST conditions over Indian Ocean and %(("Pacific, signifying teleconnections with the ENSO and Indian Ocean dipole phenomena. %()"Mid-latitude westerlies when interact with tropical convection anomalies in winter, %(*"associated precipitation gets modified. Composite analyses indicated that high (low) %)+"spring reservoir inflow happens in the years when average westerly wind is strong (weak) %)!"or has a shift due to change in tropical pressure fields (convection fields) over eastern %)#"tropical region. The deep pressure system over the North Atlantic that helps to generate %)$"strong westerly wind was found to be the key for high inflow volume and the westerly %)%"wind in winter is weak in the low spring inflow years. Surface pressure field over the %)&"Indonesian Throughflow, eastern Indian Ocean and western Pacific in winter are also %)'"linked with low and/or above average spring inflow volume. In addition, positive %)("(negative) winter temperature anomalies over the ‘Siberian High’ and negative (positive) %))"mean sea level pressure anomalies north of Siberia (eastern Arctic) are associated with %)*"the high (low) inflows. %*+" %*!"Taken together, the study here indicates that high/low spring inflow to Bhakra over Satluj %*#"River is directly connected to modulation of Western Himalayan precipitation and %*$"temperatures in winter that is remotely linked with the large-scale winter atmospheric %*%"

circulations over the major oceans, which are linked with enhanced/subdued westerlies %*&"accompanied with the jet stream. We are at a very early stage of our understanding of %*'"regional climate variability and it’s impact on the river flows in the Himalayas. Since %*("there are many large-scale climate actors in play (IOD, ENSO, Northeast monsoon), %*)"dynamical model experiments are needed to better diagnose the response of Western %**"Himalayan climate and river flows to factors that influence westerly propagation in &++"winter that varies with SST boundary conditions. &+!" &+#" &+$" &+%" &+&" &+'" &+(" &+)" &+*" &!+" &!!" &!#" &!$" &!%" &!&" &!'" &!(" &!)" &!*" &#+" &#!" &##" &#$" &#%" &#&"

References &#'" &#("Annamalai H, Kida S and Hafner J. 2010. Potential impact of the tropical Indian Ocean-&#)"

Indonesian Seas on El-Nino Characteristics. Journal of Climate 23 : 3933-3952. &#*"Barlow M, Hoell A and Colby F. 2007. Examining the wintertime response to tropical &$+"

convection over the Indian Ocean by modifying convective heating in a full &$!"atmospheric model. Geophysical Research Letters 34 : L19702. &$#"

Bush ABG. 2000. A positive climatic feedback mechanism for Himalayan glaciation. &$$"Quaternary International 65-66 : 3-13. &$%"

Dimri AP, Mohanty UC and Mandal M. 2004. Simulation of heavy precipitation &$&"associated with an intense western disturbance over western Himalayas. Natural &$'"Hazards 31 : 499-521. &$("

Dimri AP. 2005. The contrasting features of winter circulation during surplus and &$)"deficient precipitation over western Himalayas. Pure Applied Geophysics 162 : &$*"2215-2237. &%+"

Dimri AP. 2006. Surface and upper air fields during extreme winter precipitation over the &%!"western Himalayas. Pure Applied Geophysics 163 : 1679-1698. &%#"

Dimri AP. 2007. The transport of momentum, sensible heat, potential energy and &%$"moisture over the western Himalayas during the winter season. Theoretical and &%%"Applied Climatology 90 : 49-63. &%&"

Gong DY and Ho CH. 2002. The Siberian high and climate change over middle to high &%'"latitude Asia. Theoretical and Applied Climatology 72 : 1-9. &%("

Gordon AL, Susanto RD and Vranes K. 2003. Cool Indonesian throughflow as a &%)"consequence of restricted surface layer flow. Nature 425 : 824-828. &%*"

Herbert JM and Dixon RW. 2002. Is the ENSO phenomenon changing as a result of &&+"global warming? Physical Geography 23(3) : 196-211 &&!"

Jones JE and Cohen J. 2010. A diagnostic comparison of Alaskan and Siberian strong &&#"anticyclones. Journal of Climate (online). &&$"

Nobre P and Shukla J. 1996. Variations of sea surface temperatures, wind stress, and &&%"rainfall over the tropical Atlantic and South America. Journal of Climate 9 : 2464-&&&"2479. &&'"

Pal I, Lall U, Robertson A, Cane M and Bansal R. 2011. Predictability of Western &&("Himalayan river flow: Melt seasonal inflow into Bhakra reservoir in northern &&)"India. (in preparation). &&*"

Rajeevan M, Bhate J, Kale JD and Lal B. 2005. Development of a high resolution daily &'+"gridded rainfall data for the Indian region, Indian Meteorological Department, &'!"Met. Monograph Climatology No. 22/2005, pp. 26. &'#"

Rajeevan M, Bhate J, Kale JD and Lal B. 2006. A high resolution daily gridded rainfall &'$"for the Indian region: analysis of break and active monsoon spells. Current &'%"Science 91(3):296-306. &'&"

Rao VB and Rao ST. 1971. A theoretical and synoptic study of western disturbances. &''"Pure Applied Geophysics 90 : 193-208. &'("

Saji NH, Goswami BH, Vinayachandran PN and Yamagata T. 1999. A dipole mode on &')"the tropical Indian Ocean. Nature 401 : 360-363. &'*"

Sakal K, Kawamura R and Iseri Y. 2010. ENSO-induced tropical convection variability &(+"over the Indian and western Pacific oceans during the northern winter as revealed &(!"by a self-organizing map. Journal of Geophysical Research 115 : D19125, 19pp. &(#"

Schneider N. 1998. The Indonesian Throughflow and the Global Climate System. Journal &($"of Climate 11 : 676-689. &(%"

Schott FA, Xie SP and McCreary JP. 2009. Indian ocean circulation and climate &(&"variability. Reviews of Geophysics 47 : RG1002. &('"

Sen Gupta R and Desa E. 2001. The Indian Ocean – a perspective. A.A.Balkema, a &(("member of Swets & Zeitlinger Publishers. &()"

Sen PK. 1968. Estimates of the regression coefficient based on Kendall's tau. Journal of &(*"the American Statistical Association 63 : 1379-1389. &)+"

Singh P and Jain SK. 2002. Snow and glacier melt in the Satluj River at Bhakra Dam in &)!"the western Himalayan region. Hydrological Sciences Journal 47(1) : 93-106 &)#"

Vinayachandran PN and Kurian J. 2008. Modelling Indian Ocean circulation: Bay of &)$"Bengal fresh plume and Arabian Sea mini warm pool. Proceedings of the 12th &)%"Asian Congress of Fluid Mechanics. Daejeon, Korea. &)&"

Yadav RK, Rupa Kumar K and Rajeevan M. 2007. Role of Indian Ocean sea surface &)'"temperatures in modulating northwest Indian winter precipitation variability. &)("Theoretical and Applied Climatology 87 : 73-83. &))"

Yadav RK, Rupa Kumar K and Rajeevan M. 2009. Increasing influence of ENSO and &)*"decreasing influence of AO/NAO in the recent decades over northwest India &*+"winter precipitation. Journal of Geophysical Research 114 : D12112. &*!"

Yadav RK, Yoo JH, Kucharski F and Abid MA. 2010. Why is ENSO influencing &*#"northwest India winter precipitation in recent decades? Journal of Climate 23 : &*$"1979-1993. &*%"

Yamagata T, Behra SK, Luo JJ, Masson S, Jury MP and Rao SA. 2004. Coupled ocean-&*&"atmosphere variability in the tropical Indian Ocean. Book Published by American &*'"Geophysical Union. &*("

Table 1: Correlation coefficients of average MAMJ Bhakra total inflow and different climate indices in winter.

Climate indices Correlation coefficients (1978-2004) Nov Dec Jan Feb

SOI - -0.45 - -0.50 Nino4 - +0.33 +0.34 +0.34

Nino3.4 - +0.31 +0.35 +0.32 DMI +0.42 +0.42 +0.37 +0.20

Note: bold values are 95% significant and bold italics are 90% significant

Fig 1: Contour map of total inflow (z axis in cfs) of Bhakra reservoir across Satluj River in northern India

Fig 2: First derivatives of total daily inflow (logarithm of inflow to capture large variability) of Bhakra for 2001. Left (right) arrow shows onset of melt (monsoon) events

(a)

(b)

Fig 3: Correlations of MAMJ total Bhakra inflow with global winter (DJF) (a) SST (b) MSLP for 1978-2004. Shading indicates local statistical significance level at 90%

confidence that was determined by Monte Carlo method taking autocorrelations into account.

(a)

(b)

Fig 4: Composites of Bhakra inflow with MSLP (a) low inflow (b) high inflow. Shading indicates local statistical significance level at 90% confidence that was determined by the Monte Carlo method taking autocorrelations into account

(a)

(b) Fig 5: Composites of Bhakra inflow with T2/10m (a) low inflow (b) high inflow. Shading indicates local statistical significance level at 90% confidence that was determined by the Monte Carlo method taking autocorrelations into account.

(a)

(b)

Fig 6: Average winter (DJF) Latent Heat and moisture flux in (a) low inflow year (1989-1990) (b) high inflow year (2003-2004).

(a)

(b)

Fig 7: Composites of (a) low inflow with zonal wind stress at 700mb (b) high inflow with zonal wind stress at 500mb. Shading indicates local statistical significance level at 90% confidence that was determined by the Monte Carlo method taking autocorrelations into account.