Abstract

The Indian Ocean monsoon (IOM) exhibits considerable year-to-year variations that have previously been attributed to a number of forcing mechanisms including the El Niño–Southern Oscillation (ENSO) and Eurasian snow cover anomalies. In this study, spatial data of Eurasian spring land surface temperatures are analyzed as well as proxies for soil moisture, summer IOM precipitation, and summer IOM 850-mb zonal winds for the 1979–99 period to isolate correlated modes of variability. The results indicate the existence of a prominent mode that appears to be related to the boreal winter Arctic Oscillation (AO); this mode projects strongly on the June precipitation and 850-mb zonal wind fields in the vicinity of the IOM region. Its projection on spatial fields of temperature and proxies for soil moisture shows springtime surface warming and drying in the region to the north and west of the Indian subcontinent and cooling over the higher Eurasian latitudes during years of anomalously intense June monsoon rainfall. Such surface signatures are consistent with the negative phase winter AO. It is hypothesized that the preconditioning of the spring season surface characteristics may be associated with an AO-induced quasi-stationary tropospheric circulation anomaly: the impact of this anomaly is to displace the mid-Eastern jet poleward during AO-negative phases, resulting in anomalous surface heating and drying that persist into the later spring season and finally affect the rainfall over the IOM region in June.

1. Introduction

Much of the earth’s population lives in climates dominated by monsoons—seasonally reversing circulations typically characterized by alternating wet and dry seasons. The basic economic livelihood of those inhabiting regions influenced by monsoons often depends on the amount of rainfall accompanying the wet-season phase (Webster et al. 1998). In fact, the term monsoon has become synonymous with this wet-season rainfall (Trenberth et al. 2000). Monsoon failure can lead to drought, famine, and shortages of water. On the other hand, abnormally intense monsoon precipitation may induce flooding, crop loss, and disease epidemics. Given the socioeconomic impact of monsoons, there is a great need to understand the underlying dynamics of monsoon regions (Webster et al. 1998). Moreover, from a public policy perspective, reliable predictions of monsoon intensity before wet-season onset can facilitate the implementation of proactive policies designed to mitigate the negative impacts of adverse monsoon conditions.

Over the past century, much scientific effort has been devoted to the study of monsoon systems, especially the south Asian monsoon system and the portion of this monsoon that influences the Indian subcontinent, hereafter referred to as the “Indian Ocean monsoon” (IOM). A large body of work has focused on understanding the interannual variations of monsoon intensity. The role of the El Niño–Southern Oscillation (ENSO) as a modulator of year-to-year fluctuations of monsoon intensity has received considerable attention. During the late nineteenth and early twentieth centuries, Sir Gilbert Walker observed a phase relationship between monsoon rainfall and the Southern Oscillation (Walker 1924; Walker and Bliss 1932). Walker noted a tendency for below-normal monsoon rainfall to occur with negative phases of the Southern Oscillation index (i.e., El Niño events). Later studies have supported the connection between weak monsoons and warm central Pacific SSTs (Angell 1981; Rasmusson and Carpenter 1983; Ropelewski and Halpert 1987; Shukla 1987). It has been argued that anomalous subsidence over the Indonesian archipelago and south Asian regions associated with a reduction in the strength of the Walker circulation during El Niño events may inhibit monsoon development (Webster and Yang 1992). Significantly, though, the relationship between ENSO and the monsoon is not stationary and has been rather weak since the mid- to late 1970s (Torrence and Webster 1999).

The apparent breakdown of the phase relationship between monsoon intensity and ENSO over the last several decades highlights the need to consider forcing mechanisms beyond ENSO. One potential source of monsoon forcing is the state of the Eurasian land surface (Hahn and Shukla 1976; Dickson 1984). The land surface–monsoon interaction follows from the underlying dynamics of the IOM system, in particular the differential heating of the continental and oceanic surface (Ramage 1971). Changes in the land surface albedo or thermal inertia (or both) can alter the differential land–sea contrast. Through a snow–albedo feedback, for example, more extensive snow cover leads to a higher albedo and greater reflection of shortwave radiation at the surface, thereby slowing the rate of land surface heating. Deeper snow accumulation may also inhibit land surface heating via snow–hydrology feedbacks: the excess snowpack requires energy input for melting, thereby reducing net heating, and the groundwater generated by snowmelt may further limit surface heating by evaporative cooling (Yasunari et al. 1991).

Although a land surface–monsoon connection was first suggested over 100 yr ago by Blanford (1884), there is still much that is uncertain about the nature of the coupling. Hahn and Shukla (1976) found a strong connection between the extent of Eurasian snow cover and monsoon intensity. Later, Dickson (1984) and Barnett (1984, 1985) argued that greater snowfall and snow accumulation precedes drier-than-normal summertime monsoon conditions. Modeling studies have provided some support for these ideas (Barnett et al. 1989; Vernekar et al. 1995; Meehl 1994; Douville and Royer 1996). Meehl (1994), for example, describes a series of GCM simulations from various model frameworks that all show stronger monsoon conditions associated with a greater land–sea thermal contrast, lower land surface pressure, and less snow cover. Meehl (1994) also suggests a positive feedback between soil moisture and precipitation, with positive soil moisture anomalies providing a moisture source for subsequent monsoon rainfall, although the occurrence of positive soil moisture anomalies in the months preceding monsoon onset may represent (as previously noted) a negative feedback (Yasunari et al. 1991).

Not all studies, however, have reached similar conclusions concerning the coupling of the land surface and monsoons or the mechanisms involved. Bamzai and Shukla (1999), for example, suggest that the location of the most sizeable monsoon-related snow cover anomalies in regions remote to the monsoon region (e.g., western Eurasia) points to low-frequency changes in planetary-scale circulations as a source of monsoon variability. It is possible that such remote snow cover anomalies are not directly responsible for modulating monsoon intensity via the snow–albedo or snow–hydrology feedbacks. Rather, they may represent a passive response to circulation changes (like ENSO), or they actively induce circulation changes that force monsoon intensity variations. Bamzai and Shukla (1999) also emphasize the importance of soil moisture anomalies as the source of interseasonal memory. Subsequent studies, including Shinoda (2001) and Robock et al. (2003), find only a limited role for soil moisture anomalies.

The present study aims to shed some light on the land surface–monsoon connection. In exploring this connection, special emphasis is placed upon understanding the role of the boreal winter AO, the leading mode of interannual variability in the Northern Hemisphere (NH) extratropics, as a source of surface variability. We begin by presenting an overview of IOM variability as represented by a discrete, area-averaged precipitation rate index [the “large-area monsoon” index (LAMI), or its variant LAMI*]. While the LAMI is characterized by considerable variability on both interannual and intraseasonal time scales, the June LAMI exhibits a significant relationship to the boreal winter Arctic Oscillation. Using simple correlation analysis, the June precipitation index is observed to project similar spatial patterns onto the late winter/early spring surface temperature and soil moisture proxy fields as the boreal winter AO.

We next describe the results of a technique—canonical correlation analysis (CCA)—that isolates coupled patterns of variability between two fields. The CCA framework represents a valuable tool for understanding how the development of anomalous, early-season (June) IOM region precipitation variability may be related to the land surface conditions over Eurasia during the preceding late winter and spring periods. Two different combinations of meteorological fields are analyzed: late winter/early spring temperatures and June precipitation, and late winter/early spring temperatures and June zonal winds at 850mb. In both cases, CCA isolates a coupled ENSO mode and a coupled AO (or mixed AO–ENSO) mode. The latter mode supports the relationship between the boreal winter AO and June LAMI. It is speculated that the AO induces springtime surface warming and drying of the land surface as well as warming of the ocean surface in the vicinity of south Asia, possibly via an anomalous tropospheric stationary wave pattern. These surface conditions are associated with an anomalous southwesterly flow in the Arabian Sea in early summer (June) that enhances airmass convergence and cross-equatorial moisture transport into the monsoon region.

2. Data and methods

a. Data

The principal fields examined in this study are precipitation, 850-mb horizontal wind, land surface temperature, and the Palmer Drought Severity Index (PDSI). To further diagnose the results, we also utilized 500-mb zonal wind and snow cover data. The analysis is restricted to the 21-yr period from 1979 to 1999 because of the existing overlap of all datasets used. Two precipitation datasets, the Climate Prediction Center (CPC) Merged Analysis of Precipitation (CMAP; Xie and Arkin 1997) and National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis (Kalnay et al. 1996) products, are utilized, although particular emphasis is placed on the CMAP dataset. Both the CMAP and NCEP–NCAR reanalysis precipitation fields consist of gridded global monthly means of precipitation rate (in units of millimeters per day) with resolutions of 2.5° × 2.5°. The CMAP data are obtained observationally from both rain gauge and satellite measurements and have been supplemented by numerical model outputs. NCEP–NCAR reanalysis precipitation rates, on the other hand, are diagnostically determined and depend on the reanalysis model convective parameterization. The 850- and 500-mb wind data are also obtained from the NCEP–NCAR reanalysis and have a resolution of 2.5° × 2.5°. While interpolated to a uniform spatial grid by the reanalysis procedure, the variability in the wind field is largely observation driven. The monthly mean temperature data utilized in this study are from Hansen et al. (1999). These gridded, land-only temperature data have a resolution of 2° × 2° and are obtained from station measurements, typically in sparsely populated areas or towns.

To express variability in soil moisture, we utilized a CMAP-based cumulative rainfall index in the form of the standardized precipitation index (SPI; McKee et al. 1993) and the 2.5° × 2.5° PDSI record from Dai et al. (2004). The SPI provides a standardized measure of anomalies on atmospheric moisture supply over a specified time period leading up to and including the month of interest. For this study, a 4-month SPI [SPI(4)] is computed for May to capture the variability of accumulated precipitation over the late winter/spring period (i.e., February through May). The PDSI is computed through a two-layer bucket-type model that incorporates prior and present precipitation, surface air temperature, and information on local field water-holding capacity (Dai et al. 2004). Hence, the PDSI considers both cumulative atmospheric moisture supply and demand via surface temperature, in contrast to the SPI that is based on precipitation alone. Recently, variability in the PDSI was shown to be comparable to observed soil moisture over many regions including those that are relevant for the present study (e.g., China, Mongolia, and the former USSR; Dai et al. 2004). The satellite-derived snow cover data used stem from the National Oceanic and Atmospheric Administration/National Environmental Satellite, Data, and Information Service (NOAA/NESDIS). This snow cover dataset is the 2° × 2° gridded Northern Hemisphere digitized version, indicating absence/presence of snow at weekly resolution.

Comparisons are made between the various meteorological fields and indices of ENSO and the AO. The measure of ENSO employed here is the Niño-3 index (Trenberth 1997). This index represents an area average of sea surface temperature anomalies over the region 5°S–5°N, 150°–90°W (Reynolds and Smith 1994). Positive (negative) anomalies of Niño-3 correspond to El Niño (La Niña) conditions. The AO index represents the leading principal component of an empirical orthogonal function (EOF) decomposition of NH extratropical surface pressure anomalies (Thompson and Wallace 1998).

b. Methods

Two analysis techniques are applied in this study. First, simple regression/correlation analysis is applied between measures of monsoon intensity and continuous temperature fields as well as those that represent soil moisture variations in NH winter and spring before the monsoon. Such an approach is typical of monsoon studies in which monsoon intensity is defined in terms of a discrete index over a specified region. For the purposes of this study, we construct an index of monsoon intensity, which we refer to as LAMI, that consists of the area-averaged precipitation rate anomaly for the region spanning 10°–20°N, 60°–100°E. LAMI, as well as a more zonally elongated index spanning 60°–120°E (denoted LAMI*) is constructed to reflect the large-scale variability of the IOM. While LAMI (or LAMI*) is somewhat arbitrarily defined here, it will be shown in section 4 that its interannual behavior is consistent with an “objectively” determined mode of large-scale precipitation variability derived from a more sophisticated technique. Also, it should be noted that, in contrast to some more common measures of monsoon intensity, such as the All-India Rainfall Index (AIRI; Parthasarathy et al. 1995), LAMI includes both land and ocean precipitation variability; nevertheless, AIRI and LAMI/LAMI* are significantly positively correlated with one another.

In addition to simple regression/correlation analysis, we applied CCA to isolate coupled modes of spatiotemporal variability in two fields (Barnett and Preisendorfer 1987). CCA seeks to determine the features in one field that are maximally correlated with the features of the other field using an algorithm analogous to EOF analysis. Similar to EOF, the CCA modes must satisfy an orthogonality constraint, although (in contrast to EOF) orthogonality is required only in the time domain. Consequently, the canonical factor time series (CFTs) of a specified CCA mode is orthogonal to the CFTs of all other modes, while the spatial components of the modes are, in general, nonorthogonal.

To highlight a possible (land) surface–IOM connection, CCA analysis is performed on standardized, area-weighted spring Eurasian/North African land surface temperature anomalies (0°–90°N, 20°W–180°) and standardized area-weighted June precipitation as well as June 850-mbar zonal wind anomalies over the IOM spatial domain, respectively. The IOM spatial domain is defined as the regions between 10°–50°N, 30°–130°E and 10°S–10°N, 30°–95°E. Indonesia is excluded from the analysis to avoid strong ENSO variability in the precipitation fields (Dai and Wigley 2000). Even with the ad hoc exclusion of Indonesia, however, considerable ENSO variability remains.

The first step in CCA involves an EOF decomposition of each input field to limit the amount of noise present in the data. Such “EOF screening” is applied to remove between 30%–50% of the total variance from each field. Since the objective of this analysis is an understanding of the propagation of a late winter/early spring signal into late spring/early summer, the temperature EOF is constructed to carry the individual months of February, March, April, and May as an additional spatial dimension. This technique, referred to as a “sliding EOF” analysis, can ascertain patterns with a persistent temporal signature but with spatial patterns that may vary intraseasonally. Motivation for the use of this form of sliding EOF is provided by the results of the simple correlation analysis (see section 3d).

A subset of the first six EOFs from each input field is then subjected to the CCA algorithm. The CCA output consists of two weighting matrices, one for the temperature EOFs and one for the precipitation (or u850mb) EOFs that are used to reconstruct the CFTs from the original EOF time series. It also produces an eigenvalue matrix representing squared correlations of the CFTs of each field (Table 1). In both the temperature–precipitation and temperature–u850mb CCA analyses, these eigenvalues (Table 2) suggest that the first two factor correlations are very high (r > 0.78). The robustness of these canonical factors (CFs) was assessed by systematically varying the number of EOFs retained prior to CCA. This test suggests that the first two CFs are also fairly robust, though it must be noted that if the number of retained EOFs becomes too small (<5), CCA fails to separate the first two CFs. In contrast, higher-order but less correlated CFs exhibit sensitivity to the effects of EOF truncation and also, in some instances, appear to isolate more localized modes of coupled variability that explain little of the total variance in the respective fields.

Table 1.

Summary of CCA decompositions.

Summary of CCA decompositions.
Summary of CCA decompositions.
Table 2.

Correlations of LAMI and LAMI* with selected AO indices. Months are denoted by their first letter. Values in italics are significant at the 95% level (r = 0.43; two-tailed Student’s t test).

Correlations of LAMI and LAMI* with selected AO indices. Months are denoted by their first letter. Values in italics are significant at the 95% level (r = 0.43; two-tailed Student’s t test).
Correlations of LAMI and LAMI* with selected AO indices. Months are denoted by their first letter. Values in italics are significant at the 95% level (r = 0.43; two-tailed Student’s t test).

3. Regression/correlation analysis

a. Indian Ocean monsoon climatology

An overview of the climatology of the south Asian summer monsoon circulation is presented in Fig. 1. The south Asian summer monsoon system consists of multiple components, including the Indian Ocean, east Asian, and western Pacific monsoons (Webster and Yang 1992). Our principal focus in this study is the portion of the monsoon that directly influences the Indian subcontinent (the IOM). The principal features of the IOM include centers of heavy precipitation rates (>10 mm day−1) over the equatorial central Indian Ocean, the western Ghats/eastern Arabian Sea, the Bay of Bengal, and southeast Asia (Ramage 1971). In addition, the wind field at 850 mb is characterized by a horseshoe-shaped pattern, with easterly flow near the equator and westerly flow across south Asia. A jet-like flow—the Findlater jet—exists along the western margin of the Indian Ocean basin and connects the near-equatorial easterlies and Arabian Sea/Bay of Bengal westerlies (Findlater 1969).

Fig. 1.

Summertime (Jun–Aug) IOM climatology for 1979–99. Vectors are horizontal 850-mb wind (u, υ) components, and filled contours are precipitation rates. The LAMI and LAMI* regions are also highlighted.

Fig. 1.

Summertime (Jun–Aug) IOM climatology for 1979–99. Vectors are horizontal 850-mb wind (u, υ) components, and filled contours are precipitation rates. The LAMI and LAMI* regions are also highlighted.

b. LAMI interannual and intraseasonal variations

Time series of the deseasonalized LAMI, calculated from both the CMAP and NCEP–NCAR reanalysis datasets, are presented in Fig. 2. Both the seasonally averaged [June–August (JJA)] and individual monthly time series are illustrated. Several important features are evident in these time series. First, the time series computed from both datasets exhibit qualitatively similar interannual variations: Pearson correlation moments of 0.67, 0.72, 0.60, and 0.66 are noted for correlations of the June, July, August, and JJA-averaged CMAP and NCEP–NCAR LAMI, respectively. This agreement suggests that the NCEP–NCAR precipitation field over the large-area monsoon index region reflects much of the observed precipitation variability; in other words, the reanalysis precipitation variations in this region are driven, to a large degree, by observations rather than the reanalysis model framework. The consistency of the NCEP–NCAR and CMAP results suggest that it may be possible to use the reanalysis data to extend the present study to earlier time periods. Such an extension may prove useful since the coupling of the monsoon and other components of the climate system, such as ENSO, appear to be nonstationary over longer (e.g., interdecadal) time scales (Kumar et al. 1999).

Fig. 2.

Time series of the deseasonalized LAMI for both CMAP (red) and NCEP (blue) precipitation data. The LAMI is shown for individual months and for the JJA seasonal mean.

Fig. 2.

Time series of the deseasonalized LAMI for both CMAP (red) and NCEP (blue) precipitation data. The LAMI is shown for individual months and for the JJA seasonal mean.

Moreover, the time series of individual months are observed to show relatively little agreement with one another. That is, the seasonal persistence of area-averaged rainfall anomalies during a given year is weak. The lack of intraseasonal persistence may point to different forcing mechanisms of the early and late monsoon season variability. Alternatively, intraseasonal variability may be strongly influenced by high-frequency synoptic-scale weather events that are essentially unrelated to the relatively low frequency forcing mechanisms (Palmer 1994; Becker et al. 2001). Regardless of the source of intraseasonal variability, the interpretation of relationships between the monsoon and its forcing mechanisms may be sensitive to the data preparation, for example, whether seasonally averaged or individual monthly data are utilized.

c. Association between the wintertime AO and LAMI

Recently, the AO has received widespread attention because of its dominant influence on the NH winter and spring climate (Thompson and Wallace 1998). This predominantly zonally symmetric hemispheric mode of variability is characterized by a seesaw of atmospheric mass between the NH polar regions and the midlatitudes. The AO positive phase is associated with lower-than-normal pressure over the polar regions and higher-than-normal pressure at ∼45°N latitude, thereby inducing a poleward displacement of oceanic storm tracks and favoring zonal advection of relatively warm and wet air deep into continental interiors (Thompson and Wallace 2000).

Table 2 summarizes Pearson correlation moments between the AO index and the LAMI (and LAMI*) calculated from the CMAP and the NCEP–NCAR reanalysis products. A boreal winter (especially JFM) AO signal is evident in the June indices derived from both datasets. The phase relationship is such that negative AO events are associated with enhanced June precipitation over the LAMI region. The wintertime AO signal, however, does not appear to extend beyond June into the later monsoon season: correlations of wintertime AO and July and August, as well as JJA seasonal mean precipitation, respectively, are negligible. This result suggests that the wintertime AO influences only the early-season evolution of the IOM. Although only the winter AO–June LAMI/LAMI* correlations are consistently significant with different datasets and index representations, the April and May AO indices show some evidence (albeit generally weak) of a negative relationship to precipitation variability during July. Furthermore, the sign of the April/May AO–August LAMI/LAMI* correlations for the NCEP precipitation dataset tend to be of opposite sign.

d. The roles of temperature/soil moisture in connecting the wintertime AO and June LAMI

The existence of a robust boreal winter AO–early season monsoon connection suggests that indices of the AO and monsoon variability should exhibit similar footprints in the large-scale fields (such as temperature and proxies for soil moisture) during the late winter/spring period. To explore the characteristics of these footprints, covariances of the monthly February–May (FMAM) temperature and correlations of the May SPI(4) and PDSI fields computed with respect to the inverse wintertime AO index and the June LAMI (based on CMAP) are shown in Fig. 3.

Fig. 3

a. Covariance maps for NH land surface temperature anomalies based upon the standardized JFM AO index for the period 1979–99: the regression patterns between the inverse JFM AO index and (i) Feb, (ii) Mar, (iii) Apr, and (iv) May temperature anomalies. Land areas not contoured indicate missing data. Thick black contour line indicates regions that are significant at the 90% level (r = 0.37; two-tailed Student’s t test).

Fig. 3

a. Covariance maps for NH land surface temperature anomalies based upon the standardized JFM AO index for the period 1979–99: the regression patterns between the inverse JFM AO index and (i) Feb, (ii) Mar, (iii) Apr, and (iv) May temperature anomalies. Land areas not contoured indicate missing data. Thick black contour line indicates regions that are significant at the 90% level (r = 0.37; two-tailed Student’s t test).

A classic AO surface signature is obtained when the FMAM temperature fields are regressed upon the wintertime AO index (Fig. 3a), namely AO-negative phase cooling over Eurasian high latitudes and warming over North Africa and Eurasian low latitudes (Thompson and Wallace 1998). The February and March AO-negative phase warming over the North Africa/eastern Mediterranean/Red Sea region has been previously noted and has been attributed to anomalous warm air advection from the southwest (Thompson and Wallace 2000). The April warming, for which direct near-surface AO-induced advection effects are not clearly evident, occurs to the east of the February (and March) warming. It is possible that this warming occurs in response to changes in the near-surface energy balance induced by anomalous snow cover/soil moisture anomalies or changes in circulation. In May, the wintertime negative AO phase-warming signal over the land surface is quite weak. However, regression of the winter AO index onto the May SST fields reveals the presence of a significant warming signal over the northern portion of the Arabian Sea (not shown).

The projection of the June LAMI onto the FMAM temperature fields show a similar meridional structure as in the wintertime AO case, with cool temperatures to the north and warm temperatures to the south in years of anomalously wet June IOM conditions (Fig. 3b). There is, however, some mismatch in the location of the centers of strongest covariance between Figs. 3a and 3b: in February and March, for example, the maximum covariance between June LAMI and temperature occurs over western Eurasia, whereas the winter AO/temperature covariance pattern extends farther eastward. For April, the intensity and spatial extent of the warming signal over the southern Eurasian latitudes is considerably smaller in the June LAMI/temperature covariances. Also, in contrast to the winter AO/temperature covariance pattern, a more pronounced (though still weak) wet LAMI/warm temperature anomaly is observed over the Indian subcontinent in May.

Fig. 3

b. Same as in Fig. 3a, but the standardized Jun LAMI is used instead of the inverse JFM AO index.

Fig. 3

b. Same as in Fig. 3a, but the standardized Jun LAMI is used instead of the inverse JFM AO index.

The spatial features of May water surplus/deficit correlations with respect to each of the inverse wintertime AO and the June LAMI also broadly resemble one another in their meridional structure (Fig. 3c). Negative AO phases and anomalously wet June conditions in the monsoon region are associated with a tendency toward wetter conditions throughout the Eurasian midlatitudes and drier conditions over Eurasian low latitudes, including the Persian Gulf region, the Indian subcontinent, and parts of southern China. As with the temperature covariances (Figs. 3a,b), the spatial features of the SPI(4) and PDSI correlation maps are more pronounced in the vicinity of the respective centers of action of the LAMI and the AO (e.g., the June LAMI correlations show a more prominent May water deficit over the Persian Gulf–Indian subcontinent region). It should be noted that effects of snowmelt and frozen soils on soil moisture are not considered in the Palmer model (Dai et al. 2004), which renders the PDSI during the month of May over high latitudes and high elevations of the midlatitudes less reliable.

Fig. 3

c. Correlation maps of Eurasian May SPI(4) and May PDSI associated with the JFM AO index and the Jun LAMI for the period 1979–99. (i), (ii) May SPI(4) and May PDSI correlations with respect to the inverse JFM AO index; (iii), (iv) May SPI(4) and May PDSI correlations with respect to the Jun LAMI. Land areas not contoured indicate missing data.

Fig. 3

c. Correlation maps of Eurasian May SPI(4) and May PDSI associated with the JFM AO index and the Jun LAMI for the period 1979–99. (i), (ii) May SPI(4) and May PDSI correlations with respect to the inverse JFM AO index; (iii), (iv) May SPI(4) and May PDSI correlations with respect to the Jun LAMI. Land areas not contoured indicate missing data.

4. Canonical correlation analysis

In section 3, a JFM AO signature in the June IOM precipitation field is discussed. Furthermore, the wintertime AO and the June LAMI are observed to project similar spatial patterns onto the late winter/early spring surface temperature and soil moisture proxy fields. In the present section, CCA is employed with the objective of further exploring how the variability of the early summer monsoon may be related to boreal winter AO variability. As noted in section 2b, one of the powerful advantages of CCA relative to simple correlation analysis is its ability to isolate the coupling between two fields without specification of discrete indices.

a. FMAM temperatures and June precipitation

The results of the CCA on Eurasian/North African FMAM temperature and IOM June precipitation (CMAP) anomalies suggest the existence of two robust coupled modes (Table 1). The use of the June-only precipitation field follows from the simple correlation analysis, which suggests that the boreal winter AO signal in the precipitation field is quite weak in July and August (Table 1). Inclusion of the entire June–August period is, in fact, found to weaken the AO-like coupled pattern described below, although qualitatively similar leading-order modes (especially ENSO) are still obtained.

1) DJF ENSO mode

The leading CCA mode is ENSO-like both temporally and spatially. A good correlation is observed between the CFT (r = 0.50; first CFT of temperature) of this mode and the December–February (DJF) Niño-3 index [Fig. 4a(i)]. The patterns of El Niño phase warming over the land regions of southeast Asia and southern India (in April) are consistent with the analysis of Buermann et al. (2003) using seasonally averaged fields [Fig. 4a(ii)– (v)]. A tongue of enhanced June precipitation oriented along a northwest–southeast (NW–SE) axis from the Arabian Sea into the central Indian Ocean is observed during positive phases of this mode [Fig. 4a(vi)]. These precipitation anomalies suggest that El Niño phase forcing maintains a more southerly position of the ITCZ, thereby inducing anomalously large precipitation rates at latitudes closer to the equator in June and possibly delaying the onset of monsoon conditions at higher latitudes until later in the year. Indeed, El Niño events are associated with anomalous warming of the entire tropical troposphere relative to adjacent latitudes farther poleward. This meridional pattern of warming creates an anomalous temperature gradient favoring an anomalous equatorward displacement of intense convection.

Fig. 4

a. (i) Normalized time series of the first canonical factor of FMAM Eurasian/North African land surface temperature (black solid) and Jun precipitation (black dashed) anomalies over the IOM spatial domain for 1979–99. The standardized DJF average Niño-3 index time series is overplotted (red dashed–dotted). The corresponding regression patterns for (ii) Feb, (iii) Mar, (iv) Apr, and (v) May temperature and (vi) Jun precipitation anomalies based upon the standardized first CFT are also illustrated. Thick black contour line indicates regions that are significant at the 90% level.

Fig. 4

a. (i) Normalized time series of the first canonical factor of FMAM Eurasian/North African land surface temperature (black solid) and Jun precipitation (black dashed) anomalies over the IOM spatial domain for 1979–99. The standardized DJF average Niño-3 index time series is overplotted (red dashed–dotted). The corresponding regression patterns for (ii) Feb, (iii) Mar, (iv) Apr, and (v) May temperature and (vi) Jun precipitation anomalies based upon the standardized first CFT are also illustrated. Thick black contour line indicates regions that are significant at the 90% level.

Fig. 4

b. Same as in Fig. 4a, but for the second canonical factor. (i) The inverse standardized Feb–Apr average AO (blue dotted), the standardized Jun LAMI* (red dashed–dotted), and the standardized inverse JJA average Niño-3 (green dotted) time series are also shown.

Fig. 4

b. Same as in Fig. 4a, but for the second canonical factor. (i) The inverse standardized Feb–Apr average AO (blue dotted), the standardized Jun LAMI* (red dashed–dotted), and the standardized inverse JJA average Niño-3 (green dotted) time series are also shown.

2) Mixed FMA AO/JJA ENSO mode

Although the second mode appears to contain an admixture of contemporaneous (i.e., JJA) ENSO variability (r = 0.47; second CFT of temperature), its spatiotemporal features are suggestive of AO-like variability. As can be seen in Fig. 4b(i), there is good correspondence between the interannual variations of the second CFT and the inverse February–April (FMA) AO; the correlation is r = 0.67 (second CFT of temperature). Spatially, the meridional structure of warming at lower latitudes and cooling at higher latitudes associated with the second mode resembles the pattern of surface warming associated with AO negative phases [Fig. 4b(ii)– (v) and Fig. 3a]. Also, in contrast to the leading CCA mode, the temperature anomalies of the second mode extend farther to the north and west of Eurasia and exhibit more stability in their temporal evolution. Additionally, there is some suggestion of an eastward “propagation” of temperature anomalies from the North Africa/Mediterranean Sea/Red Sea sector in February to the Persian Gulf/Indian subcontinent sector in April.

This second CFT is also observed to correlate well with both the June LAMI (r = 0.54; second CFT of precipitation) and June LAMI* (r = 0.69; second CFT of precipitation), respectively. The spatial distribution of the precipitation anomalies associated with this second mode reveals why its correlation with the LAMI is strong [Fig. 4b(vi)]: uniformly wetter conditions are evident over the area used to define LAMI and LAMI*, namely, the latitude band from ∼10°–20°N and longitude band from the Arabian Sea into southeast Asia (contrast this to the more dipole-like nature of precipitation anomalies associated with the ENSO mode). Anomalously dry conditions are noted over the eastern portion of the Mediterranean as well as the near-equatorial central Indian Ocean. Repeating the CCA with the NCEP–NCAR reanalysis precipitation fields yields qualitatively similar results, although some differences in the spatial pattern of the precipitation CFs are evident, especially over oceanic regions. Divergence between the CMAP and NCEP–NCAR reanalysis results may be attributed to differences in the observed and reanalysis climatologies. Given the overall agreement, however, it should be possible to use the reanalysis to extend the analysis backwards in time. In this way, issues such as the stationarity of AO–monsoon coupling may be addressed.

3) Higher-order CCA modes

Although none of the higher-order modes are found to be robust with respect to EOF truncation effects, the CCA does provide some weak evidence for the influence of an “independent” April AO influence on the monsoon region precipitation field. In contrast to the boreal winter AO mode, the center of the April AO signal appears to lie in the east Asian monsoon region. Such a result is especially intriguing since Gong and Ho (2003) have recently noted an inverse relationship between the May AO and the east Asian monsoon associated with an anomalous poleward displacement of the western Pacific jet. It remains to be seen, however, what connection (if any) the weak April AO mode identified here may have to the results of Gong and Ho (2003).

b. FMAM temperatures and June u850mb

Although an AO-like second mode is evident in the FMAM temperature–June precipitation CCA, a boreal summer ENSO signal is shown to contaminate the mode. One source of this mixing may be related to a possible influence of the polarity of ENSO on the structure of the AO (Quadrelli and Wallace 2002). Another reason may be the brevity of the datasets: the short duration of the temperature and precipitation time series may not be sufficient to unmix the AO and ENSO signals. The noisiness and the more local characteristics of the spatial scales of precipitation fields may also contribute. In this section, the results of a temperature–850-mb zonal wind CCA are described. In contrast to the temperature–precipitation analysis, the temperature– zonal wind CCA yields better separated ENSO and AO modes.

1) ENSO mode

Like the FMAM temperature–June precipitation CCA, the leading FMAM temperature–June u850mb CCA mode is ENSO-dominated. In contrast to the first temperature–precipitation mode, however, the leading temperature–wind CCA mode is correlated with both the boreal winter and summer ENSO variability: correlations of the DJF and JJA Niño-3 with the reconstructed temperature CFT [Fig. 5a(i)] are r = 0.64 and 0.60, respectively. The FMAM temperature signal [Figs. 5a(ii)– (v)] is largely confined to the eastern portion of the Eurasian continent and is especially strong in southeast Asia. The June zonal wind signature is characterized by a meridional dipole over the western Indian Ocean [Fig. 5a(vi)]: anomalous easterlies to the north—present during El Niño conditions—are associated with anomalously dry June conditions over the northeastern Indian subcontinent (not shown).

Fig. 5

a. (i) Same as in Fig. 4a(i), except for Jun u850mb anomalies (black dashed). (ii)–(vi) Same as in Fig. 4a(ii)– (vi), except for (vi) Jun u850mb anomalies. The standardized DJF (red dashed–dotted) and JJA (green dashed–dotted) average Niño-3 index time series are overplotted.

Fig. 5

a. (i) Same as in Fig. 4a(i), except for Jun u850mb anomalies (black dashed). (ii)–(vi) Same as in Fig. 4a(ii)– (vi), except for (vi) Jun u850mb anomalies. The standardized DJF (red dashed–dotted) and JJA (green dashed–dotted) average Niño-3 index time series are overplotted.

Fig. 5

b. Same as in Fig. 5a, but for the second canonical factor. (i) The inverse standardized Feb–Apr average AO (blue dotted) and the standardized Jun LAMI (red dashed–dotted) time series are also shown.

Fig. 5

b. Same as in Fig. 5a, but for the second canonical factor. (i) The inverse standardized Feb–Apr average AO (blue dotted) and the standardized Jun LAMI (red dashed–dotted) time series are also shown.

2) FMA AO mode

Because both boreal winter and summer ENSO variability is captured in the leading temperature–wind CCA mode, the second mode is more purely AO-like than its counterpart from the temperature–precipitation CCA [Fig. 5b(i)]. The second CFTs project strongly on both the FMA AO (r = 0.70; second CFT of temperature) and the LAMI/LAMI* (r = 0.69/r = 0.64; second CFT of u850mb), respectively. In Figs. 5b(ii)– (v), a pattern of late winter/early spring temperature anomalies consistent with a wintertime AO influence is evident (see also Fig. 3a). Analogous to the leading mode, the second mode reconstructed wind field consists of a meridional dipole, although the center of the dipole is shifted slightly to the north compared to the ENSO mode [Fig. 5b(vi)]. As a consequence of this displacement of the wind field pattern, there is a more zonally oriented response in the precipitation field across the entire IOM region (see below).

A closer inspection of the anomalous June 850-mb wind field associated with the FMA AO mode suggests a possible source of the enhanced monsoon region precipitation during negative-AO/wet June years (Fig. 6). In particular, an anomalous southwesterly flow is evident in the western and central Indian Ocean basin. This anomalous flow field, which is located to the east of the climatological lower-troposphere monsoon jet, is part of an anomalous low pressure center located over the Northern Arabian Sea—some associated easterly flow over south Asia and northeasterly flow over the Arabian peninsula can also be discerned. It is speculated that this anomalous flow pattern may provide anomalous moisture transport into the IOM region, which follows from the work of Dube et al. (1990) and Cadet and Reverdin (1981). The latter study, based on observations, suggests that 70% of the climatological water vapor transport to the Indian subcontinent by the boreal summer monsoon originates to the south of the equator. Thus, the anomalous flow establishes another pathway (in addition to the climatological Findlater jet) for Southern Hemisphere water vapor to reach the monsoon region. Once advected into the climatological westerly flow over south Asia, the anomalous moisture can feed “downstream” portions of the monsoon system.

Fig. 6.

Regression maps for Jun precipitation (filled contours) and Jun 850-mb horizontal wind (vectors) anomalies based upon the second standardized CFT (u850mb) for 1979–99. The second CFT (u850mb) used stems from the temperature–u850mb CCA analysis. Thick black contour line indicates regions that are significant at the 90% level.

Fig. 6.

Regression maps for Jun precipitation (filled contours) and Jun 850-mb horizontal wind (vectors) anomalies based upon the second standardized CFT (u850mb) for 1979–99. The second CFT (u850mb) used stems from the temperature–u850mb CCA analysis. Thick black contour line indicates regions that are significant at the 90% level.

c. Consistency of CCA and simple correlation analysis

Comparison of Figs. 3, 4 and 5 suggests favorable qualitative agreement between the CCA of temperature and precipitation/u850mb and simple correlation analysis. That is, the AO-like mode of coupled temperature–precipitation (or temperature–u850mb) variability isolated by the CCA technique largely resembles both the forward projection of wintertime AO variability onto the temperature field of subsequent months and the backward projection of June LAMI variability onto the temperature field of preceding months. It should be emphasized, however, that the CCA technique produces the coupled patterns without the choice of specific indices. In that sense, the CCA modes represent more objectively determined coupled modes.

5. Potential mechanisms of wintertime AO–June LAMI coupling

The simple correlation and canonical correlation analyses described in sections 3 and 4 highlight characteristic patterns of variability that are suggestive of a boreal winter AO influence on the June monsoon intensity in the vicinity of south Asia. It is hypothesized that anomalous surface conditions associated with the occurrence of negative AO phases induce circulation anomalies that are favorable for intense early monsoon season development. What remains to be determined is how the influence of the winter AO is propagated and maintained throughout the spring season. With a characteristic time scale of 10 days, the AO lacks a long atmospheric memory (Thompson and Wallace 2001). However, Thompson and Wallace (2001) also observed that the AO exhibits a tendency to prevail more frequently in a particular phase over longer periods (i.e., at seasonal time scales), for reasons that are not yet fully understood. Furthermore, Buermann et al. (2003) report that NH spring land surface temperatures are better correlated with the preceding winter AO index than with the (contemporaneous) spring index, suggesting that some form of persistence at longer time scales is operational.

There are several pathways through which the winter AO may be connected to the early summer IOM. For example, snow cover (or snow depth) and/or soil moisture variations generated by the AO may impact the surface energy balance and thereby impact monsoon onset and intensity. SST anomalies represent another possible mechanism for linking AO and monsoon variability: because of the large heat capacity of the upper ocean, SST anomalies induced in one season may persist through subsequent seasons. Finally, low-frequency modulation of atmospheric circulation patterns, such as storm-track displacements, may serve to couple the winter AO and the summer monsoon. It is likely that some or all of these mechanisms act together to produce the observed AO forcing of early-season monsoon variability. For example, an anomalous storm-track displacement may alter regional snow cover patterns, thereby affecting soil moisture distributions and the surface energy budget.

From the analyses presented in sections 3 and 4, wet June conditions in the IOM region are observed to occur with warmer and to some extent drier conditions to the northwest of India during the preceding spring months. The temperature signal is especially pronounced over the south Asian land surface in April but less so in May; however, a stronger May temperature signal is evident over the northern portion of the Arabian Sea. The warming of the SST field over the Arabian Sea may be especially significant for the development of the intense rainfall in the LAMI region. Flatau et al. (2004) suggested that the delayed onset (and weak intensity) of the IOM in June of 2002 may have resulted from anomalous cooling of nearby ocean regions during May. Using a numerical model, Flatau et al. (2004) simulated a stronger monsoon by increasing the heat content of the upper layers of the Bay of Bengal.

We have considered how the presence of anomalous spring surface conditions may influence monsoon intensity in the LAMI region. But how do these surface conditions come about? More specifically, how might the late-spring surface state be forced by winter AO variability? As has been noted before, anomalous advection of warm air has been invoked to explain the warming in the mid-East region during February and March. However, we find little evidence for a direct linkage between anomalous surface advection and the later (April and May) warming that occurs farther to the east. We suggest that the connection between the April and May surface conditions and the winter AO may be mediated, at least in part, by middle- and upper-tropospheric planetary wave anomalies and the effect of these anomalies on cloudiness and surface heating. To illustrate this, covariances of FMAM NCEP-derived 500-mb zonal wind anomalies, computed with respect to the second temperature CFT (from the temperature–u850mb CCA analysis), are presented in Fig. 7. The pattern of zonal wind anomalies is associated with a quasi-stationary wave train in 500-mb geopotential height anomalies oriented along a northwest–southeast axis with poles of the same sign over western Europe and the northern Indian Ocean and a pole of opposite sign over the mid-East (not shown). This meridionally oriented wave train, as well as a zonally oriented wave train spanning Eurasia, appear to be associated with wintertime AO variability. Chang et al. (2001) noted similar large-scale wave patterns that they suggest might be related to changes in IOM intensity. It is possible that anomalous surface conditions induced by the winter AO, such as changes in snow cover, may force such wave trains, in a manner analogous to the mechanism suggested by Cohen and Entekhabi (1999).

Fig. 7.

Regression maps for (a) Feb, (b) Mar, (c) Apr, and (d) May 500-mb zonal wind anomalies based upon the second standardized CFT (temperature), derived from the temperature–u850mb CCA analysis, for 1979–99. Thick black contour line indicates regions that are significant at the 90% level. The 15 m s−1 monthly climatological contour line (thick gray) is also indicated.

Fig. 7.

Regression maps for (a) Feb, (b) Mar, (c) Apr, and (d) May 500-mb zonal wind anomalies based upon the second standardized CFT (temperature), derived from the temperature–u850mb CCA analysis, for 1979–99. Thick black contour line indicates regions that are significant at the 90% level. The 15 m s−1 monthly climatological contour line (thick gray) is also indicated.

The pattern of springtime zonal wind anomalies preceding wet June conditions tends to decelerate (accelerate) the equatorward (poleward) side of the climatological spring jet over North Africa and the mid-East; consequently, the storm track is displaced to the north in the spring season prior to an anomalously intense early monsoon. By contrast, the zonal-mean subtropical jet is shifted anomalously equatorward during negative AO/wet June phases. In the context of the evolution of the surface anomalies during spring, a northward displacement of the storm track over the Middle East is associated with reduced spring cloudiness over North Africa, the Arabian peninsula, the northern Indian Ocean, and adjacent land regions of south Asia (not shown). Negative cloud cover anomalies are consistent with enhanced insolation at the surface, resulting in springtime warming of the land and ocean surfaces (and drying of the land surface) in years of anomalously intense June monsoon conditions.

The role of snow cover anomalies in forcing monsoon variations and the connections of these anomalies to the boreal winter AO also warrants special consideration. Regression of the second temperature CFT from the temperature–u850mb CCA analysis (i.e., the FMA AO mode) onto the FMAM snow cover fields reflects an occurrence of somewhat less extensive local snow cover over the Himalayan/Tibetan plateau region but more extensive remote snow cover at higher latitudes prior to wet June conditions (Fig. 8). The spatial pattern of local and remote snow cover anomalies is broadly consistent with the action of the negative AO phase, which induces cooling at higher latitudes and warming at lower latitudes. However, while the temporal behavior of the remote snow cover anomalies is consistent with boreal winter AO forcing, the local snow cover anomalies appear to be less strongly related to boreal winter AO variability. Instead, these anomalies appear to be more strongly tied to DJF ENSO variability (not shown).

Fig. 8.

Correlation maps for (a) Feb, (b) Mar, (c) Apr, and (d) May snow cover anomalies based upon the second CFT (temperature), derived from the temperature–u850mb CCA analysis, for 1979–99. Regions of permanent snow cover (light gray) and no snow cover (dark gray) are also contoured.

Fig. 8.

Correlation maps for (a) Feb, (b) Mar, (c) Apr, and (d) May snow cover anomalies based upon the second CFT (temperature), derived from the temperature–u850mb CCA analysis, for 1979–99. Regions of permanent snow cover (light gray) and no snow cover (dark gray) are also contoured.

Differing sources of and roles for local and remote snow cover anomalies associated with monsoon intensity variations in the LAMI region may contribute to the heterogeneity of the correlation map. Indeed, snow cover anomalies in the Himalaya/Tibetan Plateau region may influence the monsoon directly via a snow–albedo feedback: more snow cover leads to greater reflection of solar radiation at the surface and reduced heating. The upstream snow cover anomalies in western Eurasia, on the other hand, may influence the monsoon indirectly via a snow–circulation feedback, as suggested by Cohen and Entekhabi (1999). In this way, the remote snow cover anomalies may be significant in establishing or maintaining the quasi-stationary circulation pattern that fosters the development of the south Asian land surface temperature/northern Indian Ocean SST anomalies described above.

6. Conclusions

The Indian Ocean monsoon (IOM) exhibits considerable year-to-year variations that have previously been attributed to a number of forcing mechanisms including ENSO and Eurasian snow cover anomalies. In this paper, we have described the relationship between the boreal winter AO and early-season IOM precipitation over the period 1979–99. Correlation analysis of the AO and an index of area-averaged IOM precipitation—the Large Area Monsoon Index (LAMI) or its variant LAMI*—documents a statistically significant inverse relationship between the JFM AO and June LAMI/LAMI*. Confirmation of a JFM AO–June LAMI/LAMI* connection is found in the linear regressions of these indices with temperature fields and those that represent soil moisture variations in the period “bridging” boreal winter and early summer: forward (backward) projections of the JFM AO (June LAMI/LAMI*) onto the NH February–May temperature and May SPI and PDSI fields are observed to be in qualitative agreement with one another. In particular, these projections show springtime surface warming and drying in the region to the north and west of the Indian subcontinent during years of anomalously intense June monsoon rainfall/negative winter AO phases.

To highlight the coupling of the monthly February–May North African/Eurasian land surface temperature and June IOM precipitation fields, CCA has been performed. Two significant modes of coupled temperature–precipitation variability are identified: a leading ENSO mode and a next-to-leading AO (or mixed AO–ENSO) mode. A separate CCA analysis, using February–May temperature anomalies and June IOM zonal 850-mb wind fields, more clearly decouples the ENSO and the AO signals. As in the simple correlation analysis, intense June LAMI region rainfall is preceded by springtime warming and drying to the north and west of the Indian subcontinent. The “preconditioning” of the springtime land and ocean regions in the vicinity of the monsoon region appears to be associated with an AO-induced quasi-stationary tropospheric circulation anomaly: the impact of this anomaly is to displace the mid-Eastern jet poleward during AO-negative phases, resulting in anomalous surface heating and drying. These surface conditions are associated with an anomalous southwesterly flow to the east of the lower-troposphere monsoon jet that enhances the airmass convergence (and moisture transport) into the monsoon region.

Acknowledgments

We thank Inez Y. Fung and John Chiang for their helpful comments and suggestions. This research was funded by the NASA EOS-IDS Grant NAG5-9514 (PI: Fung).

REFERENCES

REFERENCES
Angell
,
J. K.
,
1981
:
Comparison of variations in atmospheric quantities with sea surface temperature variations in the equatorial eastern Pacific.
Mon. Wea. Rev.
,
109
,
230
243
.
Bamzai
,
A.
, and
J.
Shukla
,
1999
:
Relation between Eurasian snow cover, snow depth, and the Indian summer monsoon: An observational study.
J. Climate
,
12
,
3117
3132
.
Barnett
,
T. P.
,
1984
:
Interaction of the monsoon and Pacific trade wind system at interannual time scales. Part III: A partial anatomy of the Southern Oscillation.
Mon. Wea. Rev.
,
112
,
2388
2400
.
Barnett
,
T. P.
,
1985
:
Variations in near-global sea level pressure.
J. Atmos. Sci.
,
42
,
478
500
.
Barnett
,
T. P.
, and
R.
Preisendorfer
,
1987
:
Origins and levels of monthly and seasonal forecast skill for United States surface air temperatures determined by canonical correlation analysis.
Mon. Wea. Rev.
,
115
,
1825
1850
.
Barnett
,
T. P.
,
L.
Dümenil
,
U.
Schlese
,
E.
Roeckner
, and
M.
Latif
,
1989
:
The effect of Eurasian snow cover on regional and global climate variations.
J. Atmos. Sci.
,
46
,
661
683
.
Becker
,
B. D.
,
J. M.
Slingo
,
L.
Ferranti
, and
F.
Molteni
,
2001
:
Seasonal predictability of the Indian Summer Monsoon: What role do land surface conditions play?
Mausam
,
52
,
175
190
.
Blanford
,
H. F.
,
1884
:
On the connexion of Himalayan snowfall and seasons of drought in India.
Proc. Roy. Soc. London
,
37
,
3
22
.
Buermann
,
W.
,
B.
Anderson
,
C. J.
Tucker
,
R. E.
Dickinson
,
W.
Lucht
,
C. S.
Potter
, and
R. B.
Myneni
,
2003
:
Interannual covariability in Northern Hemisphere air temperatures and greenness associated with El Nino–Southern Oscillation and the Arctic Oscillation.
J. Geophys. Res.
,
108
.
4396, doi:10.1029/2002JD002630
.
Cadet
,
D. L.
, and
G.
Reverdin
,
1981
:
Water vapor transport over the Indian Ocean during summer 1975.
Tellus
,
33
,
476
487
.
Chang
,
C-P.
,
P.
Harr
, and
J.
Ju
,
2001
:
Possible roles of Atlantic circulations on the weakening Indian monsoon rainfall–ENSO relationship.
J. Climate
,
14
,
2376
2380
.
Cohen
,
J.
, and
D.
Entekhabi
,
1999
:
Eurasian snow cover variability and Northern Hemisphere climate predictability.
Geophys. Res. Lett.
,
26
,
345
348
.
Dai
,
A.
, and
T. M. L.
Wigley
,
2000
:
Global patterns of ENSO-induced precipitation.
Geophys. Res. Lett.
,
27
,
1283
1286
.
Dai
,
A.
,
K. E.
Trenberth
, and
T.
Qian
,
2004
:
A global dataset of Palmer Drought Severity Index for 1870–2002: Relationship with soil moisture and effects of surface warming.
J. Hydrometeor.
,
5
,
1117
1130
.
Dickson
,
R. R.
,
1984
:
Eurasian snow cover versus Indian monsoon rainfall—An extension of the Hahn–Shukla results.
J. Climate Appl. Meteor.
,
23
,
171
173
.
Douville
,
H.
, and
J.
Royer
,
1996
:
Sensitivity of the Asian summer monsoon to an anomalous Eurasian snow cover within the Meteo-France GCM.
Climate Dyn.
,
12
,
449
466
.
Dube
,
S. K.
,
M. E.
Luther
, and
J. J.
O’Brien
,
1990
:
Relationships between interannual variability in the Arabian Sea and Indian summer monsoon rainfall.
Meteor. Atmos. Phys.
,
44
,
153
165
.
Findlater
,
J.
,
1969
:
A major low level air current over the Indian Ocean during the northern summer.
Quart. J. Roy. Meteor. Soc.
,
95
,
280
362
.
Flatau
,
M. K.
,
P.
May
, and
J.
Cummings
,
2004
:
Numerical investigation of the 2002 Monsoon onset. Extended Abstracts, 26th Conf. on Hurricanes and Tropical Meteorology, Miami, FL, Amer. Meteor. Soc., 108–109
.
Gong
,
D-Y.
, and
C-H.
Ho
,
2003
:
Arctic oscillation signals in the East Asian summer monsoon.
J. Geophys. Res.
,
108
.
4066, doi:10.1029/2002JD002193
.
Hahn
,
D.
, and
J.
Shukla
,
1976
:
An apparent relationship between Eurasian snow cover and Indian monsoon rainfall.
J. Atmos. Sci.
,
33
,
2461
2462
.
Hansen
,
J.
,
R.
Ruedy
,
J.
Glascoe
, and
M.
Sato
,
1999
:
GISS analysis of surface temperature change.
J. Geophys. Res.
,
104
,
30997
31022
.
Kalnay
,
E.
, and
Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis Project.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
.
Kumar
,
K. K.
,
B.
Rajagopalan
, and
M. A.
Cane
,
1999
:
On the weakening relationship between the Indian monsoon and ENSO.
Science
,
284
,
2156
2159
.
McKee
,
T. B.
,
N. J.
Doesken
, and
J.
Kleist
,
1993
:
The relationship of drought frequency and duration to time scales. Preprints, Eighth Conf. on Applied Climatology, Anaheim, CA, Amer. Meteor. Soc., 179–184
.
Meehl
,
G. A.
,
1994
:
Influence of the land surface in the Asian summer monsoon: External conditions versus internal feedback.
J. Climate
,
7
,
1033
1049
.
Palmer
,
T. N.
,
1994
:
Chaos and predictability in forecasting the monsoons.
Proc. Indian Natl. Sci. Acad.
,
60A
,
57
66
.
Parthasarathy
,
B.
,
A. A.
Munot
, and
D. R.
Kothawale
,
1995
:
Monthly and seasonal rainfall series for All-India homogeneous regions and meteorological subdivisions: 1871–1994. Contributions from Indian Institute of Tropical Meteorology, Research Rep. RR-065, Pune, India
.
Quadrelli
,
R.
, and
J. M.
Wallace
,
2002
:
Dependence of the structure of the Northern Hemisphere annular mode on the polarity of ENSO.
Geophys. Res. Lett.
,
29
.
2132, doi:10.1029/2002GL015807
.
Ramage
,
C. S.
,
1971
:
Monsoon Meteorology.
Academic Press, 296 pp
.
Rasmusson
,
E. M.
, and
T. H.
Carpenter
,
1983
:
The relationship between the eastern Pacific sea surface temperature and rainfall over India and Sri Lanka.
Mon. Wea. Rev.
,
111
,
354
384
.
Reynolds
,
R. W.
, and
T. M.
Smith
,
1994
:
Improved global sea surface temperature analyses using optimum interpolation.
J. Climate
,
7
,
929
948
.
Robock
,
A.
,
M.
Mu
,
K.
Vinnikov
, and
D.
Robinson
,
2003
:
Land surface conditions over Eurasia and Indian summer monsoon rainfall.
J. Geophys. Res.
,
108
.
4131, doi:10.1029/2002JD002286
.
Ropelewski
,
C. F.
, and
M. S.
Halpert
,
1987
:
Global and regional scale precipitation patterns associated with the El Nino/Southern Oscillation.
Mon. Wea. Rev.
,
115
,
1606
1626
.
Shinoda
,
M.
,
2001
:
Climate memory of snow mass as soil moisture over central Asia.
J. Geophys. Res.
,
106
,
33393
33403
.
Shukla
,
J.
,
1987
:
Interannual variability of monsoons.
Monsoons, J. S. Fein and P. L. Stephens, Eds., John Wiley and Sons, 399–464
.
Thompson
,
D. W. J.
, and
J. M.
Wallace
,
1998
:
The Arctic oscillation signature in the wintertime geopotential height and temperature fields.
Geophys. Res. Lett.
,
25
,
1297
1300
.
Thompson
,
D. W. J.
, and
J. M.
Wallace
,
2000
:
Annular modes in the extratropical circulation. Part I: Month-to-month variability.
J. Climate
,
13
,
1000
1017
.
Thompson
,
D. W. J.
, and
J. M.
Wallace
,
2001
:
Regional climate impacts of the Northern Hemisphere Annular Mode and associated climate trends.
Science
,
293
,
85
89
.
Torrence
,
C.
, and
P. J.
Webster
,
1999
:
Interdecadal changes in the ENSO–monsoon system.
J. Climate
,
12
,
2679
2690
.
Trenberth
,
K. E.
,
1997
:
The definition of El Niño.
Bull. Amer. Meteor. Soc.
,
78
,
2771
2777
.
Trenberth
,
K. E.
,
D. P.
Stepaniak
, and
J. M.
Caron
,
2000
:
The global monsoon as seen through the divergent atmospheric circulation.
J. Climate
,
13
,
3969
3993
.
Vernekar
,
A. D.
,
J.
Zhou
, and
J.
Shukla
,
1995
:
The effect of Eurasian snow cover on the Indian monsoon.
J. Climate
,
8
,
248
266
.
Walker
,
G. T.
,
1924
:
Correlation in seasonal variations of weather, IV. A further study of world weather.
Mem. Indian Meteor. Dep.
,
24
,
275
332
.
Walker
,
G. T.
, and
E. W.
Bliss
,
1932
:
World weather. V.
Mem. Roy. Meteor. Soc.
,
4
,
53
84
.
Webster
,
P. J.
, and
S.
Yang
,
1992
:
Monsoon and ENSO—Selectively interactive systems.
Quart. J. Roy. Meteor. Soc.
,
118
,
877
926
.
Webster
,
P. J.
,
V. O.
Magana
,
T. N.
Palmer
,
J.
Shukla
,
R. A.
Tomas
,
M.
Yanai
, and
T.
Yasunari
,
1998
:
Monsoons: Process, predictability, and the prospects for prediction.
J. Geophys. Res.
,
103
,
14451
14510
.
Xie
,
P.
, and
P. A.
Arkin
,
1997
:
Global precipitation: A 17-year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs.
Bull. Amer. Meteor. Soc.
,
78
,
2539
2558
.
Yasunari
,
T.
,
A.
Kitoh
, and
T.
Tokioka
,
1991
:
Local and remote responses to excessive snow mass over Eurasia appearing in the northern spring and summer climate—A study with the MRI GCM.
J. Meteor. Soc. Japan
,
69
,
473
487
.

Footnotes

Corresponding author address: Dr. Wolfgang Buermann, Department of Earth and Planetary Sciences, University of California, Berkeley, McCone Hall, Berkeley, CA 94720. Email: buermann@atmos.berkeley.edu