Paradox in South Asian summer monsoon circulation change: Lower tropospheric strengthening and upper tropospheric weakening

In the literature, there exist contradictory conclusions on the South Asian summer monsoon (SASM) precipitation and circulation changes: whether the circulation change contributes positively by strengthening or negatively by weakening to the rainfall enhancement, on a background of moisture content increase. Based on Representative Concentration Pathway 4.5 simulations by 18 Coupled Model Intercomparison Project phase 5 models, this study explains these puzzles by illustrating that the SASM circulation changes oppositely between the lower and upper troposphere, with tipping point at 450 hPa. However, this indicates a new paradox, created by competing mechanisms. By analyzing the intermodel variability, we determine that the mean advection of stratification change mechanism weakens the upper tropospheric circulation, while the enhanced surface land‐sea thermal contrast strengthens the lower level and surface winds. Our moisture budget analysis shows that the SASM precipitation enhancement (8% K−1) attributes to moisture increase (5% K−1) and lower tropospheric circulation strengthening (3% K−1).


Introduction
Constituting the most spectacular manifestation of the global climate system resulting from land-sea thermal contrast and orographic features, the South Asian summer monsoon (SASM) provides a major percentage (up to 80%) of annual precipitation over the world's most densely populated regions, e.g., India, and has tremendous impacts on agriculture, health, water resources, economies, and ecosystems throughout South Asia [Webster et al., 1998]. As an important part of the large-scale atmospheric circulation, its vigorous upward motion dominates the boreal summer Hadley circulation [Trenberth et al., 2005] with profound remote influences [Zhang et al., 2005].
A number of assessments on the response of the SASM to global warming have been attempted using projections from the World Climate Research Program's (WCRP's) Coupled Model Intercomparison Project (CMIP). While consistent evidences are found for the increase of the equilibrium SASM precipitation [Douville et al., 2000;Ashrit et al., 2005], contradictions stand among former studies on the monsoon circulation change in relation to the moisture and rainfall increase.
Despite a larger continental-scale land-sea thermal contrast in summer, the intensity of the monsoon circulation is predicted to reduce by 14% per century [Tanaka et al., 2005]. This weakening can be explained by the thermodynamic constraint discussed in Held and Soden [2006] or by the mean advection of stratification change (MASC) mechanism raised in Ma et al. [2012]. The principle of MASC is unique to global warming: on a background of global increase in static stability, the anomalous isotherms are apparently distorted by the advection of climatological vertical motion, which is equivalent to cooling/heating in the convective ascending/ subsidence regions. Opposing the climatological thermal driving, this anomalous adiabatic forcing factor dynamically and effectively weakens any type of tropical tropospheric circulation, including the overturning and monsoon circulations. By definition, MASC is presumably an invariant independent of model or scenario but highly sensitive to the extent of the sea surface temperature SST warming. We confirm this property in the supporting information: in the SASM region, MASC forcing shows a high intermodel correlation coefficient (r) of À0.89 with the SST warming extent, in spite of differences in model physics ( Figure S1).
The seemingly contradiction between the SASM rainfall increase and circulation reduction [Kitoh et al., 1997] can easily be interpreted as a result of the moisture increase from the warmer Indian Ocean and the consequent larger moisture transport/convergence over the monsoon region [Ueda et al., 2006 , 2003;May, 2004]. However, the projected monsoon rainfall growth rate (up to~8.4% K À1 [Kripalani et al., 2007] or 12.5% K À1 [Annamalai et al., 2007]) is higher than that of column vapor (7% K À1 [Held and Soden, 2006]), which implies an increase in the monsoon circulation. Indeed, Hu et al. [2000] reported intensifications of both SASM precipitation and circulation due to the enhancement of the land-sea thermal gradient.

Arblaster
The present study aims to clarify the apparent contradictions of the SASM circulation change (strengthening or weakening) found in the literature. After an in-depth examination, we discovered that studies suggesting the monsoon circulation weakening tend to focus on the upper troposphere [e.g., Tanaka et al., 2005;Ueda et al., 2006], whereas monsoon wind strengthening are usually found in the lower troposphere near surface [Hu et al., 2000]. Using the CMIP5 output, we look separately into changes of SASM surface and 200 hPa winds in relation to surface and air temperature changes. Mechanisms controlling the circulation change are investigated. Eventually, the vertical profile of SASM circulation change is illustrated, and a rainfall change budget is calculated, attributing to moisture content increase and lower/upper level circulation change according to the profile.

Data and Method
Representative Concentration Pathway (RCP) 4.5 simulations by 18 CMIP5 models (Table 1) are used in this study. These models provide various lengths of simulations with the future radiative forcing stabilized at 4.5 W m À2 in 2100 [Thomson et al., 2011]. We analyze one realization (r1i1p1)  The SASM precipitation and circulation indices are defined based upon the multiple indices discussed in Wang and Fan [1999]. Our indices cover a larger area than the all-Indian summer rainfall and the Webster and Yang indices: 0-30°N, 70-110°E (the magenta square in Figure 1a). Since zonal wind is generally much stronger than meridional, it is used to calculate circulation indices at surface and upper level ( Figure 2) and in combination with scalar wind speed for vertical profile (Figure 3).
We perform a moisture budget analysis ( Figure 4) to decompose the atmospheric dynamic and thermodynamic contributions to rainfall change [Seager et al., 2010]. Once the atmospheric moisture equation is vertically integrated, one obtains (1)   where P is precipitation, E is evaporation, < > represents column mass integration throughout the troposphere (approximated as 200-1000 hPa), and the overbar denotes the monthly average. V denotes atmospheric velocity. The eddy term is due to submonthly variability.
In global warming, the perturbation of P À E can be linearly decomposed as where the first term on the right-hand side represents the contribution of moisture content change and the second/third term lower/upper level circulation change.

Results
Surface wind is important for the SASM lower level moisture transfer/convergence and convection trigger, which is baroclinically in thermal wind balance with the upper level wind. In order to investigate the possibly  However, Figure 1 apparently raises a new paradox between the SASM circulation changes in lower and upper troposphere, which cannot be explained by one single mechanism but probably by competing mechanisms behind the scene. From the literature, we learned that enhanced land-sea thermal contrast is possible to enhance the SASM circulation, while MASC tends to reduce any type of tropical circulation, including the SASM. In order to examine these mechanisms, we look into the intermodel relationship between the changes of zonal wind at surface/upper level and land-sea thermal contrast/MASC in the CMIP5 RCP4.5 simulations (Figure 2). The land-sea thermal contrast in Figures 2a and 2b are defined by the difference between the average warming of land and that of ocean in the SASM index region (0-30°N, 70-110°E; marked by the magenta square in Figure 1a). As mentioned in section 1 and illustrated in Figure S1, the MASC effect is much influenced by the SST warming extent, so the average SST warming in the SASM index region is used to represent MASC in Figures 2c and 2d. Figure 2a shows a linear relation between the changes of land-sea thermal contrast and the surface zonal wind with an intermodel correlation (r) of 0.46, passing the 95% significance level, while a low correlation (0.12) is found at 200 hPa (Figure 2b), which indicates that although the land-sea thermal contrast plays a key role in strengthening the SASM lower level wind, its effect cannot reach far to the upper troposphere. In contrast, the MASC effect is found to weaken the SASM circulation throughout the column, with high correlations with both surfaces (r = À0.42; Figure 2c) and upper level (r = 0.53; Figure 2d) zonal wind changes, passing the 90% and 99% significance levels, respectively. A sample illustration of the MASC effect on surface and upper level SASM circulation is shown in Figure S2, with a comparison with the latent heating effect. Nevertheless, in the lower troposphere near surface, the MASC effect is apparently overwhelmed by the enhanced land-sea thermal contrast, resulting in strengthened circulation. Figure 2 clarifies the paradox in the SASM circulation change by competing mechanisms: weakened upper tropospheric wind by the MASC effect, and strengthened lower tropospheric and surface wind by the enhanced surface land-sea thermal contrast. Figure 3 summarizes these effects by showing the vertical profiles of climatology and change of the SASM zonal wind and wind speed averaged in the index region and among the 18 CMIP5 RCP4.5 simulations. In fact, barotropically across all tropospheric levels, the zonal wind change appears to be westerly (consistent with Figure 1), with two peaks at 850 hPa and 250 hPa, indicating the maximum effects of land-sea contrast and MASC, respectively. The uncertainty range marked by ensemble mean ±1 standard deviation shows that the sign of the SASM circulation change profile is robust among the models. However, because of the baroclinic nature of the climatological SASM circulation, the anomalous westerly weakens the upper tropospheric easterly wind but strengthens the lower tropospheric westerly wind. As clearly shown by the wind speed change, the apparent paradox in the SASM circulation change is created, with the tipping point at 450 hPa.
The moisture budget analysis formulated by equation (2) decomposes the change in the SASM δ P À E À Á into contributions from moisture content increase and lower and upper level circulation changes separated by the 450 hPa tipping point (Figure 4). The total δ P À E À Á (Figure 4a) shows a "wet-get-wetter" pattern [Held and Soden, 2006], increasing in the North Indian Ocean (I), off the west coast of India (II), along the Himalaya ridge (III), and from the northeastern Bay of Bengal through southern China (IV). This change pattern is resulted from moisture increase (Figure 4b) in Regions I and III and from lower level wind strengthening (Figure 4c) in Regions I, II, and IV. Unsurprisingly, upper level wind weakening (Figure 4d) has little contribution due to lacking of moisture. Dominating δ P À E À Á , the percentage increase of the spatial mean rainfall in the SASM index region by 8% K À1 attributes to the moisture content increase by 5% K À1 and the lower tropospheric circulation enhancement by 3% K À1 , along with the tropical mean SST warming.

Conclusions
The present study has summarized four contradictions with explanations on the SASM precipitation and circulation changes in the literature: 1. Despite an enhancement in continental-scale surface land-sea thermal contrast in summer, the intensity of the monsoon circulation is predicted to reduce: Circulation weakened by the MASC mechanism; 2. Rainfall increases while circulation reduces: Precipitation enhancement is due to increase in water vapor content; 3. Rainfall growth rate is higher than that of column water vapor: Implication for circulation increase; 4. Circulation strengthening was reported, consistent with enhanced land-sea thermal contrast but inconsistent with other former research: Circulation weakening occurs in the upper troposphere, while strengthening occurs in the lower troposphere and near surface. More importantly, we have raised and clarified one new paradox with a related question: 5. Why is the circulation change opposite between the lower and upper troposphere? And how would this affect precipitation?
Confirming our findings in the literature, we have clearly shown that in global warming projections with the 18 CMIP5 models along RCP4.5, the SASM circulation change has a paradox: Wind increases at surface but deceases at upper level. Based on an intermodel variability analysis on possible mechanisms, we found that competing mechanisms are responsible for this paradox: The MASC effect weakens the upper tropospheric circulation, while the enhanced surface land-sea thermal contrast dominates the strengthening of the lower tropospheric and surface winds, though MASC tends to have a minor negative contribution.
The vertical profile of average zonal wind change in the SASM region shows a barotropic westerly consistent through all tropospheric levels. However, this barotropic westerly change has level-dependent meanings for the baroclinic climatology, strengthening the lower tropospheric westerly and weakening the upper tropospheric easterly, with tipping point at 450 hPa, as shown by the wind speed change profile. Our moisture budget analysis unsurprisingly suggests no significant contribution of the upper level circulation reduction to the SASM rainfall change due to low moisture content, and the precipitation change pattern attributes to both moisture content increase and lower level wind enhancement, which contribute 5% K À1 and 3% K À1 , respectively, to the total spatial mean precipitation increase of 8% K À1 , along with the tropical mean SST increase.