Where were the monsoon regions and arid zones in Asia prior to the Tibetan Plateau uplift?

12 The impact of the Tibetan Plateau uplift on the Asian monsoons and inland arid climates is an 13 important but also controversial question in studies of paleoenvironmental change during the 14 Cenozoic. In order to achieve a good understanding of the background for the formation of the 15 Asian monsoons and arid environments, it is necessary to know the characteristics of the 16 distribution of monsoon regions and arid zones in Asia before the plateau uplift. In this study, we 17 discuss in detail the patterns of distribution of the Asian monsoon and arid regions before the 18 plateau uplift on the basis of modeling results without topography from a global coupled 19 atmosphere-ocean general circulation model, compare our results with previous simulation studies 20 and available bio-geological data, and review the uncertainties in the current knowledge. Based on 21 what we know at the moment, tropical monsoon climates existed south of 20°N in South and 22 Southeast Asia before the plateau uplift, while the East Asian monsoon was entirely absent in the 23 extratropics. These tropical monsoons mainly resulted from the seasonal shifts of the Inter-24 Tropical Convergence Zone. There may have been a quasi-monsoon region in central-southern 25 Siberia. Most of the arid regions in the Asian continent were limited to the latitudes of 20-40°N, 26 corresponding to the range of the subtropical high pressure year-around. In the meantime, the


INTRODUCTION 38
The Asian monsoons and arid climates are closely related to global change and, to a large 39 extent, determine the formation and development of various Asian environments [1,2]. The with previous studies, we examine the distribution patterns of the Asian monsoons and arid 144 climates before the TP uplift and then review and discuss relevant issues and associated 145 uncertainties, which allow us to identify some of the problems or issues that are worth further 146 study in the future. 147 148

RESULTS FROM A COUPLED ATMOSPHERE-OCEAN MODEL 149
We first analyze simulation results from a fully coupled AOGCM by focusing on the spatial atmosphere and ocean are coupled once every day. FAMOUS is structurally almost identical to 158 HadCM3, and produces climate and climate-change simulations that are reasonably similar to 159 HadCM3 but runs much faster. This characteristic is particularly useful for long runs of ppmv. One experiment has the present-day land-ocean mask and topography (abbreviated as 165 "OROG" for the name of the experiment); and the other has the same land-ocean mask and 166 idealized, uniform land surface characteristics as the OROG, but with global orography height set 167 to 0 (abbreviated as "FLAT" for the name of the experiment). Both of these experiments have 168 highly idealized, globally uniform land surface characteristics (albedo, roughness length etc.) to 169 highlight the impact of the orographic changes. Both experiments have been run for 1000 years 170 and the last 100 year mean results are used in this paper. We mainly focus on the distributions of 171 the Asian monsoon regions and arid zones in the following analysis.

173
The characteristics of precipitation of monsoon climates are mostly reflected in the seasonal 175 cycle of alternating rainy and dry seasons during the year. In reference to Wang and Ding [47], we 176 first define the monsoon regions in the Eastern Hemisphere for the OROG and FLAT experiments 177 using the rainfall seasonality. Specifically, we define monsoon regions as places where the 178 difference in rainfall between summer (rainy season, as June-July-August (JJA) for Northern 179 Hemisphere (NH) and December-January-February (DJF) for Southern Hemisphere (SH)) and 180 winter (dry season, as DJF for NH and JJA for SH) is greater than 200 mm, and where the 181 percentage of summer to annual total rainfall is greater than 40%. Regions with summer-winter 182 rainfall difference greater than 400 mm can be considered as the typical monsoon regions. Based 183 on this definition, for the OROG experiment representing the present-day condition with global 184 topography (Fig. 1a), the simulated typical monsoon regions are mostly found in the northern intensities and spatial extents, as long as the land-ocean configuration remains the same as today. 211 Additionally, weak monsoon climates can be found in parts of southwestern, southern, and eastern 212 China. It should also be noted that beside the tropical and subtropical monsoons, there exists a 213 quasi-monsoon region in Siberia of the upper mid-latitudes of the Asian continent (~50-65°N) 214 where summer rainfall accounts for 45-60% of the annual total (Fig. 1b). 215 According to the traditional definition of monsoons, they should be characterized by a 216 seasonal reversal of dominating winds [21,48], which lead to wet summers and dry winters. In 217 order to visually interpret the seasonal changes of the dominant winds from the FLAT experiment 218 representing the zero-topography condition, we mapped the simulated NH winter (DJF) and 219 summer (JJA) 1000 hPa wind vectors (Fig. 2). In the NH winter (Fig. 2a), the wind field from 220 Africa to East Asia is very similar to typical planetary wind belts, with the regions south of 30°N 221 mostly being dominated by northeasterly winds. In the NH summer (Fig. 2b), however, the 222 regions south of 20°N from Africa to South and Southeast Asia and those south of 30°N in East 223

Continental arid zones 248
Aridity results from the presence of dry descending air and a lack of moisture, which lead to 249 the lack of clouds and precipitation. Aridity arises from a number of general causes acting 250 individually or working together, for example, atmospheric high pressure zones, continentality, 251 rain shadows, and cold ocean currents [49,50]. At the regional scale, the causes of aridity mainly 252 include continentality that depends on distances from large water bodies or oceans, rain 253 barrier/rain shadow effects of mountains, or cold oceanic surface currents that create stable Therefore, SLP fields can be used to represent distributions of persistent high-pressure systems 262 and the associated arid climate zones. 263 before the TP uplift (Fig. 3), there is a zonal pattern of SLP isobars in the NH winter across the 265 entire Eurasian continent, with SLP values decreasing from the south to north in the mid-to 266 high-latitude regions north of 30°N (Fig. 3a). This distribution pattern indicates the dominant 267 control of the westerly circulation in the mid-to high-latitudes with the center of the subtropical 268 high-pressure zone located close to 30°N from Southwest to East Asia, while the low-pressure During the NH summer (Fig. 3b), due to the heating of the Eurasian continent, there is a strong 278 warm-core surface low pressure over the land mass, with the center located near the Lake Baikal 279 at approximately 55°N. This low pressure over land extends to the south and interrupted the 280 continuity of the subtropical-high belt over the Eurasian continent, while over the oceans the 281 northern extent of the subtropical high-pressure cells can reach 40°N or even further north in the 282 Pacific (Fig. 3b). At the same time, the subtropical high-pressure belt in the SH is centered near 283 30°S, extending from southern Africa to central Australia. For the annual average SLP (Fig. 3c), 284 the NH subtropical high-pressure belt extends from North Africa to East Asia, centered near 30°N 285 and there is a low-pressure zone extending W-E centered near 55°N. The latitudes in-between are 286 controlled by the westerlies, while the latitudes north of 55°N are dominated by the polar 287 easterlies. In the SH, the subtropical high-pressure belt is centered near 30°S. It is worth noting that there may have been a narrow quasi-monsoon zone running across the 336 upper mid-latitude Eurasia (50-65°N), especially in central and southern Siberia before the TP 337 uplift according to our simulation results (Fig. 1b). This region is characterized by a summer rainy 338 season with prominent seasonality. In its core region at 55-60°N, the directions of the winter and 339 summer dominating winds are nearly opposite (Fig. 2), fitting the traditional definition of the 340 monsoon climate. Therefore, to certain extent this region can be considered as having a weak 341 monsoon climate. However, it should be pointed out that at these relatively high latitudes, the 342 winds that bring moisture to produce summer precipitation are not originated from the tropical  (Fig. 1b). Typical monsoons 375 are also found in the tropical Africa and northern Australia. These monsoon phenomena have been 376 resulted from the insolation-induced seasonal shifts of the ITCZ. On the contrary, a 377 similar-to-present monsoon pattern is observed in the model outputs corresponding to a 378 modern-topography scenario with an elevated Tibetan Plateau (Fig. 1a). These results are highly 379 consistent with the available geological evidence in South Asia (e.g., [69,70] (Fig. 4a,b). dry belt before the TP uplift (Fig. 3c) is highly consistent with the semi-arid zone defined by the 398 bio-geological indicators with Paleogene ages within China (Fig. 4a,b)  Miocene.

414
The results of numerical experiments in this study for the scenario with global topography are 415 also in agreements with the Cenozoic paleoenvironmental patterns (Fig. 4)  East Asia (Fig. 1a). This is strongly supported by the boi-gelogical data [15] showing a drastic 419 humidification in eastern China since the early Neogene (Fig. 4c). Similarly, the simulated dry 420 climate in central Asia is consistent with the dry conditions documented by the bio-geological data 421 controlled aridity in Asia before the TP uplift also radically differs, in both origin and concept, 432 from the present-day dry lands in central Asia, which are independent of the subtropical high. Our 433 results suggest that any studies of the subject should consider these crucial conceptual differences. 434 Otherwise, controversies could arise simply because the discussed concepts of the monsoons and 435 aridity are different. 436 patterns over the Asian continent has also drawn much attention from the paleoclimate community. 438 Our model outputs, in association with the geological data, may provide a significant insight to 439 this issue. Although modeling results themselves are not actual chronological events, geological 440 data (Fig. 4)  sites where the history of the monsoon can be traced back to the late Oligocene [69]  precipitation may increase in the Asian monsoon regions under the scenario of increased 501 greenhouse gas concentrations, mostly caused by enhanced moisture convergence, the monsoon 502 circulation intensity itself will be weakened. This means that the effects of changing atmospheric 503 CO2 concentration on the Asian monsoons before the TP uplift should also be examined in detail. 504 Additionally, making things even more complex, the palaeogeography related to plate tectonics 505 has been recognized as a key factor controlling the long-term evolution of the atmospheric CO2 506 through its capability of modulating the efficiency of silicate weathering and the climate 507 sensitivity to atmospheric CO2 [83]. Consequently, the modulation of changing atmospheric CO2 508 to the development of the Asian monsoons during geologic time has been identified as an 509 important area for further research. 510 Finally, in view of the current limited knowledge and the importance in understanding the 511 distribution of the Asian monsoon and arid regions before the TP uplift as the foundation for the 512 study of the impact of the TP uplift on spatial patterns of climates in Asia, there is much more to 513 accomplish in areas of numerical simulation and analysis of geological records. This is especially 514 true in terms of cross-comparison and integration of the simulation results and geological evidence. 515 For example, regarding the question of whether the EAM existed before the Miocene, or if it 516 indeed existed back then, its spatial range and intensity are still open for answers. From the 517 modeling perspective, the emergence of the SAM should be earlier than that of the EAM, which, 518 however, still lacks supporting geological evidence at the present. Unless the paleoelevation of the 519 TP was sufficiently high in the Paleocene-Eocene period, the EAM system would not have been 520 established at a much earlier time than the Miocene. For future studies, in the Tertiary before the 521 TP uplift or when the plateau elevation was still low, the question of whether there existed a 522 quasi-monsoon zone in the mid-to high-latitudes of Eurasia, separated from the tropical and 523 subtropical circulations, also requires cross-validation using relevant geological records. Although 524 15 there should have been no strong aridity in the mid-latitude Asian inland north of 40 o N before the 525 TP uplift according to numerical simulations, the actual timing of formation and spatio-temporal 526 evolution history of the mid-latitude arid regions in central Asia remain as important scientific 527 questions that require validation using reliable high-resolution geological records. 528 529

FUNDING 530
The work was supported by the Strategic Priority Research Program of the Chinese Academy 531 of Sciences (XDB03020601) and the National Natural Science Foundation of China (41290255) 532 with numerical modelling support from NCAS-Climate and NCAS-CMS in the UK. 533 534