Volume 39, Issue 9
Free Access

Late Miocene upward and outward growth of eastern Tibet and decreasing monsoon rainfall over the northwestern Indian subcontinent since ∼10 Ma

Peter Molnar

Corresponding Author

Peter Molnar

Department of Geological Sciences, University of Colorado at Boulder, Boulder, Colorado, USA

Cooperative Institute for Research in Environmental Sciences, Boulder, Colorado, USA

Corresponding Author: P. Molnar, Department of Geological Sciences, University of Colorado at Boulder, Boulder, CO 80309, USA. ([email protected])Search for more papers by this author
Balaji Rajagopalan

Balaji Rajagopalan

Cooperative Institute for Research in Environmental Sciences, Boulder, Colorado, USA

Department of Civil, Environmental, and Architectural Engineering, University of Colorado at Boulder, Boulder, Colorado, USA

Search for more papers by this author
First published: 01 May 2012
Citations: 37


[1] Many geologic observations suggest that the eastern portion of the Tibetan Plateau has expanded eastward and grown in height over the past 15-10 Ma, and other observations suggest that climate over the northwestern Indian subcontinent (“NW India”) became more arid between ∼11 and 7 Ma. We suggest that they are linked: higher terrain increased orographic precipitation over eastern Tibet, which diabatically heated the atmosphere there; that heating then forced subsidence to the west via the mechanism proposed by Rodwell and Hoskins; and subsidence suppressed rainfall. Simple Gill-model calculations for a heat source displaced from the equator show such subsidence, and a regression of July-August outgoing long-wave radiation (OLR) over southeastern Tibet with rainfall shows a negative relationship between eastern Tibet and NW India, consistent with the proposed link between growth of eastern Tibet and diminished South Asian monsoon rainfall over the northwestern Indian subcontinent.

Key Points

  • The growth of Tibet has affected south Indian climate on geological time scales
  • The effect is not a strengthened, but a weakened monsoon
  • We must understand the monsoon as being more than All-India-Rainfall

1. Introduction

[2] A long tradition associates the South Asian (or Indian) summer monsoon with heating over Tibet and a reversal in meridional temperature gradients, which are accompanied by ascent over Tibet, cross-equatorial circulation aloft, and strong, also cross-equatorial, low-level southwesterly to westerly winds that bring moisture to the Indian subcontinent [e.g.,He et al., 1987; Webster et al., 1998; Yanai et al., 1992]. Recently, however, Boos and Kuang [2010] challenged this view and argued that the only part of the Tibetan Plateau that is vital for a strong monsoon is its southern edge, the Himalaya. Exploiting theory [e.g., Boos and Emanuel, 2008; Plumb, 2007; Privé and Plumb, 2007] and the observation that ascent occurs over the northern Indian subcontinent and the southern edge of the Himalaya [Boos and Emanuel, 2009; Webster et al., 1998], they argued that the Himalaya blocks southward flow of cold dry air onto the Indian subcontinent, which, if allowed to mix with hot moist air, would delay and weaken the monsoon.

[3] The apparent irrelevance of heating over Tibet for monsoon rainfall, however, presents a conundrum for paleoclimatic interpretations of monsoon history and geologic evidence for the growth of Tibet; many have correlated, if only crudely, changes in climate over the South Asian monsoon region with the growth of Tibet near 10 Ma [e.g., Harrison et al., 1992; Molnar et al., 1993; Prell et al., 1992; Prell and Kutzbach, 1997]. Conversely, given that the dimensions of the Tibetan Plateau have evolved over the past 50 Ma and that climate over part of the monsoon region has changed, paleoclimatic data may allow an assessment of Tibet's role in the monsoon.

[4] For more than 20 years, paleoclimatic evidence from the northwestern Indian subcontinent (northwestern India and northern Pakistan, but hereafter “NW India”) has accumulated to suggest that the climate of that region began to change near ∼10 Ma (Figure 1). Paleo-ecological evidence (fossil leaves, wood, roots, etc.) suggests that a relatively humid climate prevailed in NW India in mid-Miocene time (ca. 15-10 Ma) [e.g.,Prasad, 1993; Retallack, 1995]. Subsequent changes in stable isotopes from carbonate sediment [e.g., Behrensmeyer et al., 2007; Quade et al., 1989, 1995; Stern et al., 1997], in both large and small mammals [e.g., Barry, 1995, Barry et al., 2002; Flynn and Jacobs, 1982], and in pollen [Hoorn et al., 2000] point toward a shift toward a more arid climate, perhaps a seasonably more arid climate like that of monsoons (Figure 1).

Details are in the caption following the image
Map of Tibet and surroundings showing studies that suggest growth of eastern Tibet since 15-10 Ma and climate change in NW India since ∼10 Ma. White boxes show regions for which modern rainfall in July–August are correlated (Figure 3b). For eastern Tibet, observations include evidence for acceleration of incision of deep canyons or of erosion of high terrain [Clark et al., 2005; Godard et al., 2009; Kirby et al., 2002; Ouimet et al., 2010; Duvall et al., submitted manuscript, 2012], of exhumation of rock within thrust belts [Lease et al., 2011; Zheng et al., 2006, 2010], and in sedimentation rates and onsets of coarse sediment deposition in basins [Bovet et al., 2009; Fang et al., 2003, 2005; Lease et al., 2007, 2011; Wang et al., 2011; Zhang et al., 2012]. Evidence for climate change includes a change from forest- to savannah-dwelling mammals [Barry, 1995; Barry et al., 2002; Flynn and Jacobs, 1982], and time series showing shifts of δ18O in pedogenic carbonates [Quade et al., 1989], grass pollen [Hoorn et al., 2000], and percentages of G. bulloides in cores from the Arabian Sea [Kroon et al., 1991; Prell et al., 1992]. In time series, the present is on the right, and red dashed lines mark 5-Myr intervals.

[5] Regarding the tectonic evolution of eastern Asia, a wealth of evidence suggests that styles and rates of crustal deformation within Tibet and on its surroundings underwent a change since ∼15-10 Ma (Figure 1) [e.g., Harrison et al., 1992; Molnar and Stock, 2009; Molnar et al., 1993]. This evidence includes accelerated folding of the Indian Ocean sea floor south of India [Cochran, 1990], the growth of high terrain on the eastern and northeastern margins of Tibet [e.g., Lease et al., 2011; Zheng et al., 2006, 2010], and most relevant here, deep incision of eastern Tibet beginning between 15 and 8 Ma [e.g., Clark et al., 2005; Godard et al., 2009; Kirby et al., 2002; Ouimet et al., 2010]. Moreover, A. Duvall et al. (Widespread Late Cenozoic increase in erosion rates across the interior of eastern Tibet constrained by detrital low-temperature thermochronometry, submitted toTectonics, 2012) showed that erosion accelerated since 11 to 4 Ma across a swath of high Tibet drained by rivers flowing east (Figure 1); they argued that this accelerated erosion calls for an eastward tilting of the plateau to give the rivers additional stream power.

[6] Although the significance of individual observations noted in Figure 1might be challenged, collectively they suggest that eastern Tibet began to grow upward and eastward since ∼15-10 Ma, and concurrently or a bit more recently, NW India's climate changed. FollowingRodwell and Hoskins [1996] we argue that the growth of eastern Tibet can be related to the climate change in NW India.

2. The Influence of Subtropical Heating on Regions to the West

[7] Building on Gill's [1980] model of atmospheric circulation in the tropics, Rodwell and Hoskins [1996] argued that a heat source in the subtropics would send Rossby waves westward, such that in a steady state, a standing wave pattern would develop with subsidence over the latitude band of the heat source and west of it. Adiabatic descent would induce radiative cooling of descending air, which in turn would cause drying and further, diabatic, cooling of it. This would suppress the development of clouds and allow direct radiative heating of the land surface. They suggested that both the aridity of the Sahara and the weak summer rainfall over the Mediterranean result from intense monsoon heating of the atmosphere over the northern Bay of Bengal.

[8] To quantify this effect, they exploited a relatively simple global model that uses the primitive equations with linear drag in the lower levels and Newtonian cooling to maintain the basic state that they imposed. The model does not include moisture, but instead a heat source where latent heating would be large. With this model Hoskins and Rodwell [1995] obtained a circulation reminiscent of that for the South Asian monsoon. As another test of the model, Rodwell and Hoskins [1996] showed that the circulation forced by a heat source closer to the equator (centered at 10°N) matched that of Gill [1980]. A heat source centered at 25°N induced subsidence to the west over the Sahara and Mediterranean regions. After examining the effects of a background circulation, with westerly winds appropriate for summer and, also later, with realistic topography, Rodwell and Hoskins [1996, p. 1395] reported: “It would appear that remote monsoon heating can induce a general region of descent associated with the linear Gill-type Rossby-wave pattern to its west and that the interaction of the mid-latitude westerlies with the Rossby thermal structure intensifies this descent region north-eastward.”

[9] Given the agreement of Hoskins and Rodwell's [1995] calculations with Gill's [1980] simple model, we extended Gill's approach (in the auxiliary material) and show results of a simple calculation of the vertical component of velocity for a heat source centered at ∼30°N (Figure 2). Our purpose is not to make a quantitative comparison with observations, but merely to show that heating induces descent to the west. Gill's simplifying assumptions are many, not least of which is the representation of the Coriolis effect as varying linearly with latitude. Moreover, Gill combined the two likely dissipative mechanisms, Rayleigh friction and Newtonian cooling, using a single decay constant for both and with a value that is many times larger than commonly assigned to each. These limitations are widely known, and Sarachik and Cane [2010] devote several pages to discussing why despite its weaknesses, the Gill model works so well. Different values for the decay constants lead to different rates of descent and different westward extents of descent, but the general pattern of descent west of the heat source is a common feature of heating in regions displaced from the equator (Figure 2). As shown in the auxiliary material, the zonal extent of the region where descent occurs depends (i) exponentially on the value of the rate of dissipation, with high rates enhancing localized rapid descent just west of the heat source, and (ii) linearly on the zonal width of the region of diabatic heating. Rodwell and Hoskins [1996] used a relatively wide heat source, which contributes to its distant effects as far west as the Mediterranean in their calculations.

Details are in the caption following the image
Calculated dimensionless vertical component of velocity forced by a heat source given by cos [(y − 3)π/2] between latitudes 2.5 < y < 3.5, and zero elsewhere, where dimensions are in Rossby radii of deformation (∼10°) as used by Gill [1980] and also cosine tapered over a longitude range of one radius of deformation (See auxiliary material). Note ascent over the region that is heated and in the region to its north and south, and descent to the west. Contour intervals are −0.01, −0.005, −0.0025, 0, 0.01, 0.02, 0.05, and 0.1 dimensionless units that scale with heating and dissipation rates.

3. Present-Day Correlations

[10] Condensation of vapor, which forms clouds and then falls as precipitation, releases latent heat and therefore heats the atmosphere diabatically. If heating of the air over eastern Tibet led to descent over NW India, and if such descent suppressed rainfall there, then present-day rainfall over eastern Tibet should correlate negatively with rainfall over NW India. Because rainfall can vary over short distances, especially in rugged terrain where rain gauges are sparse, we use outgoing long-wave radiation (OLR) as a surrogate for rainfall and related latent heating over eastern Tibet [e.g.,Hartmann, 1994, p. 158].

[11] Rainfall over NW India is dominated by the monsoon, with intense rainfall over the western edge of India and over the northern Bay of Bengal and adjacent land to the north. As the monsoon evolves, a band of rainfall follows the southwestern margin of the Himalaya and although not as intense as to the southeast, it marks a seasonal cycle that peaks in July-August in NW India (Figure 3b). Rainfall over southeastern Tibet peaks earlier, in June, but remains high through July and August. Thus, a correlation of July-August rainfall with OLR over Eastern Tibet allows a sensible test of the proposed link between them (Figure 3). We used the Aphrodite gridded (0.5°) daily precipitation data over monsoon Asia for 1961–2004 [Xie et al., 2007] to form monthly averages, but tests with datasets of Mitchell and Jones [2005] and Rajeevan et al. [2006] show similar results.

Details are in the caption following the image
Relationships of precipitation and OLR over NW India and eastern Tibet. (a) Map of correlation coefficients for July-August rainfall over eastern Asia [Xie et al., 2007] and OLR [Liebmann and Smith, 1996] over Eastern Tibet (region in red box). (b) Monthly averages and standard deviations of rainfall [Xie et al., 2007] from NW India (in blue box in Figure 3a. (c) July-August rainfall over NW India vs. OLR over eastern Tibet. Solid black and red lines show linear and LOC regressions [Warton and Weber, 2002; Warton et al., 2006], and dashed lines show 95% confidence intervals.

[12] A spatial correlation of rainfall over eastern Asia with OLR, based on an interpolation of satellite data [Liebmann and Smith, 1996], over eastern Tibet shows the expected negative correlation over that region with r < −0.7 (Figure 3a). Low OLR, associated with radiation by high, cold clouds, which also block the earth's surface, correlates with relatively high rainfall. OLR in eastern Tibet, however, correlates positively with rainfall in NW India with r = 0.45 (Figure 3c), significant at p < 0.01 [Helsel and Hirsch, 2002].

[13] A linear regression of rainfall over NW India against OLR over eastern Tibet (Figure 3c) yields Precip = −8.7 (±4.4) mm/day + 0.06 (±0.02) OLR (W/m2), where the values in the parenthesis are standard errors of the regression coefficients. If errors in both precipitation and OLR are allowed in the regression model fitting, also known as standardized major axis (SMA) estimation [Warton and Weber, 2002; Warton et al., 2006] or line of organic control (LOC) [Helsel and Hirsch, 2002, chap. 10], the resulting line is Precip = −24.0 (±5.3) mm/day + 0.13 (±0.02) OLR (W/m2). Thus if OLR over eastern Tibet were 20–30 W/m2 higher than it is there today, but similar to what it is today over eastern China in summer, we would predict that that summer rainfall over NW India would have been 30–50% greater than it is today.

4. Discussion and Conclusions

[14] We exploit a series of largely qualitative arguments to suggest that the growth of the eastern margin of the Tibetan Plateau led to an aridification of NW India. A number of geologic observations suggest such aridification some time between roughly 11 and 7 Ma, but none offers tight quantitative constraints (Figure 1). Similarly, several observations suggest that eastern Tibet has expanded eastward and grown higher since 15-10 Ma (Figure 1), but again, none derives from quantitative paleo-altimetry. In what is the weakest link in the argument, we assume that a higher eastern Tibet led to increased orographic precipitation over that region. This assumption carries the implicit assumption that the effect of orography farther west or northwest, along Tibet's eastern margin before 10–15 Ma, was modest. Enhanced precipitation, manifesting itself as diabatic heating of the atmosphere, would then induce descent farther west over NW India. The association of such heating with descent to the west rests on a quite solid foundation, from the work ofRodwell and Hoskins [1996], and can be illustrated qualitatively using the Gill [1980] model (Figure 2). Finally, a positive correlation between July-August OLR over eastern Tibet and precipitation over NW India (Figure 3) supports such a relationship between heating over eastern Tibet and descent over NW India.

[15] This train of logic differs from what many of us have argued: that the rise of Tibet near 10 Ma led to a strengthening of the monsoon [e.g., Harrison et al., 1992; Molnar et al., 1993; Prell et al., 1992; Prell and Kutzbach, 1997]. The association of a higher eastern Tibet with a drier NW India, however, implies that rainfall over NW India decreased near 10 Ma and, if this rainfall were a measure of the strength of the South Asian monsoon, that an already strong monsoon weakened. This is opposite to the inference that the monsoon strengthened at that time; the rapid increase in the planktonic foraminifer Globigerina bulloidesover the western Arabian Sea at 8–10 Ma suggests increased upwelling of cold nutrient-rich water, as occurs during monsoons today [e.g.,Kroon et al., 1991; Prell et al., 1992]. These different views of the monsoon need not be inconsistent, for as is widely recognized, the monsoon manifests itself differently in different regions, and changes in one need not require changes in another. Unfortunately, paleoclimatic data over the parts of India where monsoon rainfall is most intense today do not seem to exist yet. In any case, the arguments given here that summer rainfall over NW India decreased as eastern Tibet rose do not require a weakening of the South Asian monsoon, at least as manifested in its clearest form by heavy rain over parts of the Indian subcontinent and strong winds over the western Arabian Sea, but they do suggest that heating over at least part of Tibet does affect monsoon rainfall over part of NW India.


[16] We were stimulated by preliminary General Circulation Model results of G.-s. Chen, Z.-y. Liu, and J. E. Kutzbach. Without K. Julien's encouragement and guidance, we would not have dug sufficiently deeply into Gill's model. J. A. Collins and A. C. Molnar gave necessary programming help, and W. R. Boos, G. H. Roe, and an anonymous reviewer offered constructive criticism of early drafts. This research was supported in part by the National Science Foundation under grant EAR-0909199.

[17] The Editor thanks Warren Prell and an anonymous reviewer for assisting with the evaluation of this paper.