Contrasting controls on convection at latitude zones near and away from the equator for the Indian summer monsoon

Understanding controls on convection on various timescales is crucial for improved monsoon rainfall forecasting. Although the literature points to vertically homogeneous vorticity signatures preceding rainfall during the Indian summer monsoon, we show using reanalysis data that, for rainfall associated with northward propagating intraseasonal oscillations (ISOs), different controls are present at different latitude zones. For the latitude zone close to the equator (5∘N–14∘N) and including the southern Indian region, a conventional dynamical control on rainfall exists with barotropic vorticity leading ISO rainfall by about five days. In contrast, for the latitude zone away from the equator (15∘N–24∘N; covering the central Indian region), thermodynamic fields control ISO rainfall, with barotropic vorticity following rainfall by two days on average. Over central India, the pre-moistening of the boundary layer (BL) yields maximum moist static energy (MSE) about four days prior to ISO rainfall. Analyzing the statistics of individual events verifies these observations. Similar thermodynamic control is also present for the large-scale extreme rainfall events (LEREs) occurring over central India. These high rainfall events are preceded by positive MSE anomalies arising from the moisture preconditioning of the BL. The resulting convection then leads to a maximum in barotropic vorticity 12 h after the rainfall maximum. Characterizing these influences on convection occurring over various timescales can help identify the dominant mechanisms that govern monsoon convection. This can help reduce climate model biases in simulating Indian monsoon rainfall.


Introduction
The South Asian monsoon is characterized by seasonal reversal of winds and northward shifting of the inter-tropical convergence zone along with enhanced precipitation in the South Asian region [1].Its prominent subseasonal features include low-frequency northward propagating intraseasonal oscillations (ISOs) [2][3][4] and westward-moving synoptic scale depressions and low-pressure systems (LPSs) [5][6][7][8].The ISOs modulate the active-break spells of monsoons, crucial for the hydrological cycle and the agriculture sector of the country [9,10].These ISOs also influence the prevalence of LPSs, with the latter occurring more frequently during the active phase [11].The LPSs are linked to extreme rainfall events (EREs) [12,13], which can lead to very heavy rainfall, often exceeding 300 mm day −1 , and can have adverse societal impacts.Despite their importance, the processes controlling convection at ISO timescales or during such EREs are not fully understood.
On intraseasonal timescales, various mechanisms based on dynamics and thermodynamics have been proposed for the northward propagating ISOs [14][15][16][17][18][19][20].Among these, the one that has received the most attention appeals to the generation of barotropic vorticity before convection [17].In this mechanism, in the presence of background vertical shear of the zonal winds, baroclinic divergence arising from previous convection results in barotropic vorticity generation, which then leads to convection moving northward.In an idealized model of this process, the additional inclusion of mean vertical shear of the meridional wind leads to improved ISO simulation [21].Although this mechanism provides a clear atmospheric pathway that gives dynamic control of convection, it is unclear how the initial convection is triggered.Several studies have pointed out the importance of moisture and its advection for ISO propagation [19,[22][23][24], suggesting a role for thermodynamic influences on ISO rainfall.In fact, there is a lot of diversity in the origin and spatial structure of individual ISO events, as well as in their average manifestations across different years [4,25,26].Therefore, it is unlikely that a single account of the northward propagation of convection is valid for all ISO events across the monsoon period, especially as convection moves northward during individual events.All these mechanisms have been reported to be influential to varying degrees at different locations in the monsoon domain [27].
Most studies in the past have examined the composite structure of ISO vorticity relative to the convection center, irrespective of where this convection center is present [17,19,28].However, it is notable that latitude zones close to and away from the equator experience very different wind shear and moisture distribution during the monsoon season, and these differences can influence the dominant mechanisms that are at work on various timescales.In fact, a recent modeling study pointed out the in-phase nature of barotropic vorticity and convection away from the equator [23].This suggests that the simplified view of barotropic vorticity preceding convection is not true at all latitudes.Thus, it is important to examine the ISO events occurring at different latitude zones separately.
The ISO events present away from the equator, associated with the active phase of monsoon, support the formation of LPSs and subsequent EREs [11].Past studies have reported higher (3.5 times) frequency of LPSs during this active phase [11,29] compared to the break phase.A significant number of LPSs are known to contribute to large-scale EREs (LEREs) in central India [13].A few different mechanisms governing the generation and evolution of these LPSs have been proposed in the literature.These can broadly be classified into vertical wind shear-related baroclinic instability [30,31], horizontal wind shear-related barotropic instability [32,33], and moisture gradient-related moist-vortex instability [34].In each case, an important condition governing the evolution of LPSs is the control and maintenance of the convection in these events.Despite various mechanisms, the conditions that lead to EREs still remain elusive.
In the past, studies have suggested that convection is a consequence of low-level convergence [35][36][37].During monsoons, this convergence can be achieved through the generation of barotropic vorticity [17].This dynamical pathway leading to convection has been criticized, with studies showing an insignificant role of convergence in controlling convection [38,39].An alternate view in which surface fluxes and boundary layer (BL) moist static energy (MSE) play a more important role in the formation of convection has recently gained prominence [40][41][42].This view puts more emphasis on the production and destruction of convective instability, many a time measured by the presence of anomalously high BL MSE, in influencing the evolution of convection.A systematic understanding of the pathways leading to convection at various timescales during the summer monsoon is missing.In this study, we seek to bridge this gap by examining the evolution of convection alongside the closely related dynamic and thermodynamic variables, using reanalysis data.
This study seeks to identify the influences on convection at different latitude zones and over a wide range of timescales during rainfall events associated with the summer monsoon.By focusing on the composite evolution of convection and associated dynamic (vorticity) and thermodynamic (BL MSE) fields for ISOs and LEREs at different latitude zones, we provide evidence of predominantly thermodynamic control on convection away from the equator.

Data
We use six hourly ERA5 reanalysis data from the European Centre for Medium-Range Weather Forecasts [43] for the months of May to October.The data was downloaded for vorticity, geopotential, specific humidity, and temperature at nine different pressure levels from 200 hPa to 1000 hPa.The mean total precipitation rate is taken as the proxy for convection in this study and is converted to typical rainfall units (mm day −1 ).This will be referred to as rainfall from here on.Using the top of the atmosphere outgoing longwave radiation (OLR) as the proxy for convection does not affect the main findings of this study.All the variables were downloaded at 1 • × 1 • horizontal resolution for the time period of 1979 to 2020.Although we use reanalysis rainfall (ERA5) for our analysis, we have verified through other datasets (including Tropical Rainfall Measuring Mission (TRMM) 3B42(V7) [44] and India Meteorological Department gridded rain-gauge product over land [45]) that ERA5 rainfall captures the observed seasonal mean climatology and the intraseasonal variation satisfactorily (figure S1).Throughout the paper, barotropic vorticity is defined simply as the vertically averaged vorticity from 1000 hPa to 200 hPa.
MSE is defined as: where C p is the specific heat capacity at constant pressure for dry air, T is absolute temperature, L v is the latent heat of vaporization of water, q is specific humidity, g is the gravitational constant, and z is geopotential height.BL MSE is considered as the vertical average of MSE between 900 hPa to 975 hPa pressure levels.

Identification of ISO events
All the ISO fields are obtained by applying the 30-70 day Butterworth band-pass filter to the corresponding variables similar to previous studies [17,20].
For intraseasonal variability, we focus on two latitude zones: one near the equator (70 ) and another away from it (70 . Day 0 at each latitude zone is defined as the day of maximum ISO rainfall averaged over that particular zone.We consider only strong ISO events that have rainfall exceeding 1.5 times the standard deviation at day 0 (figure S2).
Since the focus is on 30-60 day northward propagating ISOs, all the instances where the day 0 ISO rainfall is separated by at least 30 days are considered different ISO events for constructing composites.We only consider those events for which day 0 lies within the June to September months of our study period.In order to assess the lead-lag relation between convection and other variables, we subtract the day of maximum ISO rainfall (day 0 by construction) from the day on which the other variable is maximum for one complete ISO cycle.We call this value the lag.With this calculation, a negative (positive) lag would mean that the respective variable's maximum precedes (follows) the rainfall maximum, which we describe simply as the variable leading (following) rainfall.

Identification of LEREs
For LEREs, we choose the region from 75 • E to 85 • E and 15 • N to 25 • N, as this region has been reported to have the highest frequency of EREs in previous studies [13,46].Day 0 of an LERE is identified as the day at which the area averaged rainfall for our chosen region first exceeds a prescribed threshold.We prescribe the threshold to the 99.5 percentile value of averaged rainfall over this region for the entire study period of May to October for 42 years (1979-2020).We note here that this method of detection of EREs is different from the previous studies, most of which have used a gridpoint-wise classification [29,47,48] instead of area-averaged rainfall for classifying extreme events.Such an alternative way of classification of EREs is governed by our focus on the evolution of the wider dynamic and thermodynamic fields during these events, rather than quantifying the number of EREs.To avoid false identification of the same LERE multiple times, we set a criterion ensuring that two extreme events must occur at least three days apart from each other.

Control on convection
Throughout this manuscript, the dominant control on convection is established based on the variable that precedes rainfall during the lifecycle of the convection event (for both ISO and LEREs).For any event, if barotropic vorticity leads rainfall we term this as dynamic control on convection, whereas if BL MSE leads rainfall, it is termed as thermodynamic control.

Relationship between ISO rainfall and vorticity
The spatial maps of ISO rainfall and ISO barotropic vorticity composites on the day of maximum ISO rainfall (day 0) at the two latitude zones are shown in figures 1(a) and (c).The filtered rainfall shows a northwest-southeast tilt, a known feature of the northward propagating ISO events [4].This tilt is more prominent in the 15 • N-24 • N latitude zone.
The ISO barotropic vorticity is positive and maximum in the respective zones at day 0 but does not have a northwest-southeast tilted structure, unlike ISO rainfall.The vorticity amplitude strengthens away from the equator with maximum value in the northern Bay of Bengal adjacent to the coast.ISO rainfall also has its maximum value in this region.Now focusing on the temporal evolution of averaged ISO rainfall and barotropic vorticity in these two latitude zones (figures 1(b) and (d)), we notice a crucial difference.For the zone close to the equator, barotropic vorticity leads rainfall by around four days (figure 1(d)).This observation is in agreement with the leading hypothesis of northward propagating ISO in which the generation of positive barotropic vorticity leads to convection [17].One of the assumptions of this hypothesis is the presence of background easterly shear.The seasonal mean easterly wind shear (U 200 hPa -U 850 hPa ) in this region is −29.7 ms −1 .It has been previously reported that barotropic vorticity precedes convection when the rainfall is centered at 5 • N [27].In contrast, we find that this relationship reverses in the latitude zone away from the equator, where the rainfall precedes barotropic vorticity by about two days (figure 1(d)).This clearly suggests that ISO rainfall is controlled by a different mechanism in this region with the generation of barotropic vorticity possibly being an effect and not a cause of convection.The mean easterly shear in this region is much smaller (−19.8ms −1 , 33% lesser than near the equator).To ensure that this result is not merely an artifact of averaging, we examined the unfiltered composites for both latitude zones.Despite the high-frequency systems that are embedded in the ISOs, a similar leadlag is recovered from the unfiltered composites as well (figure S3).We also performed ISO filtering at each grid point and then calculated the difference in the day of maximum ISO rainfall and ISO barotropic vorticity.Most grid points show a positive lag north of 15 • N, which suggests that the rainfall maximum precedes the vorticity maximum to the north of 15 • N (figure S4).
From figures 1(b) and (d), it is clear that this leadlag relationship holds not only for the day of maximum rainfall and vorticity but also for the day when these two variables change sign.For the latitude zone near the equator, positive ISO rainfall is preceded by positive barotropic vorticity.Whereas away from the equator, ISO rainfall first becomes positive, and positive barotropic vorticity is then seen to appear two days later.A similar picture emerges when we perform the same analysis using TRMM rainfall for the period of 1998-2020 (figure S5).This further confirms that away from the equator, the generation of local positive vorticity is not the direct cause of ISO rainfall.

Statistics of individual events
In the previous section, we presented results based on the mean of all the events.Here we will examine if we obtain a similar sequence of evolution upon analyzing the statistics of individual events.Based on our identification criteria for ISOs (section 2), we found 28 events in the latitude zone: 5 • N-14 • N and 58 events in the latitude zone 15 • N-24 • N for the entire period of analysis.The temporal evolution of ISO barotropic vorticity for both latitude zones, along with their 95% confidence intervals, again reveals the differing lead-lag relationships described in the previous section (figure 2(a)).For the 15 • N-24 • N zone, most events have a positive lag (for the definition of lag, refer to section 2), with a mean of 2 day and a median of 2.25 day (figure 2(b)).Here, 48 out of the total 58 events have rainfall preceding the vorticity.On the other hand, for the latitude zone close to the equator, most events have a negative lag with vorticity preceding rainfall (figure 2(b)).The mean is −4.8 day, and the median for this zone is −3.75 day.In this region, for 24 out of the total 28 events, vorticity precedes rainfall.Both the mean and the median of lag for the two regions are statistically different using the two-sample t-test and Wilcoxon rank-sum test [49], respectively.Comparing the boxplot of lag values for the two regions also confirms this difference between the two regions.For the ISO events close to the equator, the interquartile range of lag lies between −8 and 0, whereas for the events away from the equator, it lies between 0 and 3 (figure S6).The cumulative distribution functions of lag for the two latitude zones are also significantly different (using the Kolmogorov-Smirnov test), further verifying our initial observation of the different lead-lag relationships of vorticity and rainfall between the two latitude zones (figure S7).

Thermodynamic control away from the equator
So far, we have established with reasonable confidence that barotropic vorticity does not lead to ISO rainfall in the latitude zone away from the equator.Thus, a different process controls the fate of convection in this region.The BL quasi-equilibrium theory suggests that this control could come from surface fluxes and BL processes [41,42].In this framework, convection quickly acts to remove any instability present in the atmosphere.This means that BL MSE is a potential candidate for controlling the convection.We now analyze the BL MSE to see if control on ISO rainfall can be attributed to thermodynamic processes.The evolution of ISO-filtered MSE composites shows a large variability pointing to the complex nature of thermodynamic fields (figure 3(c)).
The MSE peaks about four days before ISO rainfall for the 15 • N-24 • N latitude zone, whereas it is in phase for the zone close to the equator (figure 2(c)).Considering the individual events and the lag of each of these events, it is clear that there is a lot of variability in the value of lag of MSE for both regions (figure 2(d)).Despite such a large spread, the zone away from the equator has most events with a negative value of lag (41 out of 58), confirming the observation reported from the mean evolution of MSE composites.The BL MSE leading rainfall provides a thermodynamic control on convection away from the equator.Examining the evolution of individual terms of BL MSE (equation ( 1)), it is found that it is actually the pre-moistening (L v q) of the BL accompanied by a delayed cooling response (C p T) that results in this lead in BL MSE relative to ISO convection (figure S8).This pre-moistening of the atmosphere before the convection has been discussed in earlier studies [19,22,50].Although the latent heat contribution to MSE precedes convection for both the latitude zones, there are subtle differences in the timing of BL cooling (C p T) in the two regions that result in differences in the day of MSE maximum in the two regions.More specifically, for the 15 • N-24 • N latitude zone, the premoistening takes place a few days before convection.Additionally, there is a slower cooling response of the BL such that the temperature reaches its minimum Thus, we see different controls on convection in the two latitude zones on ISO timescales.Close to the equator, where the background easterly wind shear is high; there is a dynamic control on convection with positive barotropic vorticity preceding the positive rainfall anomaly.Away from the equator, where the background easterly wind shear is smaller, there is predominantly thermodynamic control on convection with the rise in BL MSE occurring before the enhancement of rainfall.The generation of barotropic vorticity follows after the presence of positive rainfall anomaly in this case.

Large-scale Extreme Rainfall Events (LEREs)
The latitude zone 15 • N-24 • N covers central India and the core monsoon region of the country, and the period of positive ISO phase in this region is often considered the active phase of the monsoon [51].The previous section revealed thermodynamic control on ISO rainfall in this latitude zone, and now we explore if a similar control is present for the LEREs occurring over central India.Based on our identification criteria for LEREs (section 2), we found 59 extreme events in central India for the whole study period.
We note here that the frequency of EREs in our study is significantly lower than reported previously [29,47,48].This is because we use area-averaged rainfall for classifying extreme events, unlike past studies that used a gridpoint-wise classification.This choice leads to the detection of spatially coherent large-scale extreme events that have been reported to be mostly present in central India [13].The composite mean of all the LEREs at the time of maximum rainfall shows a large convective event covering central-east India with rainfall exceeding 50 mm day −1 (figure S9).The temporal evolution of barotropic vorticity composites averaged over the same region shows a lag with respect to rainfall.The maximum vorticity, on average, is achieved after 12 hours of peak rainfall (figure 3(a)).Examining the individual events reveals that out of all the 59 LEREs, 49 have a rainfall maximum either coinciding with the vorticity maximum or preceding it (figure 3(b)).For 70% of these events, the maximum in vorticity is achieved within 24 h of rainfall maximum, suggesting a rapid circulation response to the convection in such events (figure S10).As in the previous section, we now assess the BL MSE to examine whether there might be a thermodynamic control on LEREs.The composite MSE anomaly is positive prior to the time of maximum rainfall and becomes negative after it (figure 3(c)).This is true for most of the events with BL MSE maximum appearing around 18 hours to a day prior to the rainfall maximum (figure 3(d)).Out of the 59 LEREs, 43 have a negative value of lag, confirming that for most events, the region has positive MSE prior to intense rainfall.The individual components of MSE again suggest a similar mechanism, where the moisture anomalies reach their maximum about half a day prior to the convection maximum leading to the pre-moistening of the BL.Temperature declines to its minimum value between 6 and 12 h afterward (figure S11).This could result from a combination of evaporative cooling of the falling precipitation and reduced insolation from significant cloud cover [52,53].
Thus we find that for LEREs, pre-moistening of the atmosphere and a positive BL MSE leads to high rainfall, again suggesting a thermodynamic control on convection at synoptic timescales north of 15 • N. The convection enhances the vorticity, possibly through vortex stretching from the BL convergence.This mechanism remains to be explored and is postponed for a later study.

Summary and discussions
Convection-circulation coupling is a striking feature of the tropics, with the interaction of the two occurring at various ranges of spatial and temporal scales [54].Indeed, convection-circulation coupling has been identified as the key missing piece in our understanding of the tropical climate [55].The separation of the two for causal inference is seldom possible.One reason for this is that various dynamic and thermodynamic processes work together in the initiation, growth, and demise of convective systems.Past studies have argued that for Indian summer monsoon ISOs, barotropic vorticity generation leads to convection.In this work, we show that this is not always the case, and it, in fact, is dependent upon the latitude zone at which convection occurs.Close to the equator (5 • N-14 • N), in the presence of strong background easterly shear, the barotropic vorticity leads ISO rainfall by five days.Thus, in this region, dynamic control over convection is predominant.On the other hand, away from the equator (15 • N-24 • N), the background easterly shear is weaker, and it is the thermodynamic processes that appear to have the dominant control over ISO convection.Here, the premoistening of the BL and the rise in MSE precede convection, while convection is followed by further enhancement of the barotropic vorticity after about two days (figure 4).This contrasting lead-lag relationship for the two regions is invariant to the choice of variable used as a proxy for convection, and similar results are obtained if the top of the atmosphere OLR is used instead of rainfall (figure S12).The moisture preconditioning of the atmosphere away from the equator could be caused by the advection of moisture.A recent modeling study has noted the significance of moisture advection for the propagation of ISO north of 10 • N [23].
The latitude zone away from the equator has also been reported to be favorable for large-scale extreme precipitation [13].Most of these extreme events are associated with westward-moving low-pressure synoptic systems, which are more frequent during the active phase of ISOs [11].Analyzing the composite of EREs reveals a similar picture to that of ISOs in this region, with the BL MSE leading convection by 36 h.Convection is followed by barotropic vorticity, which peaks 12 h after maximum precipitation.This thermodynamic control on convection is associated with the rise in BL moisture just before the LEREs.Numerous recent studies have discussed the important role that moisture plays in the dynamics of LPSs [13,34,56].The background moisture profile evolves northward throughout the monsoon season, which could be a potential reason for the dominant thermodynamic control on convection away from the equator.The precipitation rate has previously been shown to be sensitive to the background meridional gradient of moisture even in a simple model, with the meridional wind gradient further decoupling vorticity from convection [56].The consequent rise of barotropic vorticity, which is seen after the convection reaches its maximum, could result from latent heat release during convection and the associated BL convergence that leads to a large vortex stretching contribution to the vorticity budget [57].This control on the vorticity budget is more effective away from the equator, where the planetary vorticity is larger.
This study summarizes various dominant controls on monsoon convection at intraseasonal as well as synoptic timescales and important changes in these controls as one moves away from the equator.Although past literature points to the importance of barotropic vorticity for convection, by analyzing the chronological order of the evolution of convection and the associated barotropic vorticity, we have demonstrated that the relationship between the two is strongly latitude-dependent.Away from the equator, thermodynamic processes have a stronger influence on monsoon convection.Although a lead-lag analysis is not sufficient to prove causality, it identifies a sequence of critical processes that create favorable conditions for convection.For an improved mechanistic understanding of why MSE leads convection, a detailed MSE budget analysis is required.A recent study has shown that the MSE budget for understanding moist thermodynamics is most accurate for systems that have latent energy anomalies much larger than geopotential and kinetic energy anomalies [58].This condition is satisfied for both ISO convection and EREs.Furthermore, MSE budget analysis for the Indian monsoon has confirmed that MSE variations are mainly governed by moisture on intraseasonal timescales [59].Corresponding MSE budgets for the systems examined in the present paper can advance the understanding of thermodynamic controls on convection.
This study points to the latitudinal dependence of processes that control the convection on intraseasonal timescales.It has the broad implication that models need to simulate the latitude dependence of dominant controls on convection correctly, which is important for their ability to capture the variability of convection.Several generations of Coupled Model Intercomparison Project (CMIP) models, including the latest CMIP6, are unable to reproduce this variability [60][61][62].A recent study has reported an unduly weak relationship between MSE and precipitation in these models [63].It is also likely that these models do not capture the mechanisms influencing the build-up of MSE in the BL away from the equator, including over Indian landmass, and hence fail to promote off-equatorial precipitating convection.Given the decreasing trend of ISO rainfall [29] and the increasing trend of EREs [47] in recent decades, it is important to recognize these dominant controls and their variability in order to systematically apprehend how they might be influenced in a warming climate.Furthermore, thermodynamic control away from the equator points to the potential for improvements in rainfall forecasting from expanded assimilation of moisture and BL thermodynamic variables in weather forecast models.
The results of this study on latitudinal dependence of convection, in the context of the Indian summer monsoon, are likely to be applicable more broadly.Further analysis to uncover common processes and factors controlling convection across the global tropics is important.

Figure 1 .
Figure 1.(a) Composite of ISO filtered barotropic vorticity (color) and ISO rainfall (contour) when the ISO rainfall was maximum (day 0) at latitude zone 15 • N-24 • N. (b) Composite evolution of ISO rainfall (in mm d −1 ) and barotropic vorticity (in s −1 ) when rainfall is centered at 70 • E-90 • E; 15 • N-24 • N. (c) Same as (a) but for the latitude zone 5 • N-14 • N. (d) same as (b) but for the region 70 • E-90 • E; 5 • N-14 • N. Stippling in (a) and (c) represents the regions with significant anomalies at 95% level using Student's t-test.95% confidence interval is also indicated in (b) and (d).

Figure 2 .
Figure 2. (a) Evolution of ISO barotropic vorticity for latitude zone: 5 • N-14 • N (in blue) and 15 • N-24 • N (in red).(b) Histogram of difference in the day of maximum barotropic vorticity and maximum rainfall (day 0) for all the individual ISO events identified from 1979 to 2020 for the two latitude zones.The mean and sample standard deviation (in days) for the two zones are also indicated.(c) Same as (a) but for ISO-filtered boundary layer moist static energy (BL MSE).(d) Same as (b) but for BL MSE.The shading in (a) and (c) indicate the 95% interval based on Student's t-test.

3 .
(a) Evolution of composite mean of barotropic vorticity anomaly of all the EREs in the Central Indian region (15 • N-24 • N).(b) Histogram of difference in the day of maximum barotropic vorticity and maximum rainfall (day 0) for all the individual EREs identified from 1979 to 2020.The mean and sample standard deviation (in days) of the distribution are also indicated.(c) Same as (a) but for BL MSE.(d) Same as (b) but for BL MSE.value a few days after the day of maximum convection.Together, these two factors lead to the BL MSE increasing roughly four days before convection for the 15 • N-24 • N latitude zone.

4 .
Schematic showing the different processes that control ISO convection at the two different latitude zones.(a) Equatorward of 15 • N, the mean easterly shear is strong, and barotropic vorticity leads rainfall by 5 day.Here, the rainfall is in phase with boundary layer MSE (BL MSE).(b) Poleward of 15 • N, the mean easterly shear is weak, and BL MSE leads rainfall maximum by 4 day.The maximum rainfall then leads to barotropic vorticity maximum 2 day later.Thermodynamic processes control convection in this zone.