Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
It is well known that the South American monsoon system (SAMS) develops at low latitudes in response to seasonal changes of the continent-ocean thermal contrast, and is the major modulator of summer rainfall over tropical South America (Nogués-Paegle et al., 2002; Vera et al., 2006). It exhibits most of the continental monsoon features, such as the annual rainfall cycle with wet summer and dry winter (Rao et al., 1996; Zhou and Lau, 1998; Gan et al., 2004; Grimm et al., 2005). The SAMS wet season begins in the equatorial Amazon and propagates rapidly eastward and southeastward during austral spring (Kousky, 1988; Horel et al., 1989; Rao et al., 1996; Marengo et al., 2001; Gan et al., 2004; Vera et al., 2006; Garcia and Kayano, 2009). The monsoon pattern is well developed during summer, when the major convective activity over Central Brazil is linked to that in the South Atlantic convergence zone (SACZ). The SAMS demise starts during austral autumn, when the intense precipitation over the Amazon weakens and migrates northwestward. So, the SAMS cycle includes the convection propagation in the northwest–southeast direction during both onset and demise stages. However, Janowiak and Xie (2003) found that the rainfall over most of southwest Brazil and northern Bolivia propagates bidirectionally northeastward and southwestward during the SAMS onset. Thus, the regions where the rainfall is strongly modulated by the SAMS are the Central Amazon (CAM) Basin, western Central Brazil (WCB) and the SACZ area.
Most of the above-mentioned papers analyze the monsoon-related convection or precipitation annual variations. However, for a better understanding of the thermodynamic processes involved during the monsoon life cycle, it is important to analyze the moisture and heat budgets in the main areas encompassing the SAMS or the SAMS-related systems. For these analyses, it is worthwhile to recall that Yanai et al. (1973) introduced the concepts of apparent heat source (Q1) and apparent moisture sink (Q2) as the residual terms of the heat and moisture budgets, respectively. So, the vertical and horizontal distributions of Q1 and Q2 might indicate the nature of the thermodynamic processes involved. From the equations to estimate Q1 and Q2, presented by Luo and Yanai (1984), one can draw an inference on the consistency of these estimates and the processes involved. If the heating in a given area, besides the radiative heating (QR), is mainly due to the condensation process, the vertically integrated values of Q1–QR and Q2 are similar. However, the occurrence of a strong sensible heat flux or evaporation from the surface might yield differences between the horizontal distribution of the vertically integrated heat source and moisture sink (Luo and Yanai, 1984). Yanai and Tomita (1998) found a cumulus convective vertical transport over Central Brazil during austral summer, spring and autumn, as indicated by the positive vertical profiles of Q1 and Q2 with a Q1 peak in the 400–600 hPa layer and a Q2 peak in the 700–850 hPa layer.
Rao et al. (1996) calculated the divergence of vertically integrated water vapor flux over the South American region for the 1985–1989 period, and showed that the tropical Atlantic is the main moisture source for the Amazon Basin, which in turn is the main moisture source for Central Brazil during the austral spring and summer months. Drummond et al. (2008), based on a Lagrangian diagnostic method, also found that the tropical South Atlantic is the main moisture source for Central Brazil during austral summer. Two main low-level systems are involved in this process: the continental low and the South Atlantic subtropical high (SASH). While the SASH plays an important role by transporting water vapor (Lenters and Cook, 1995; Rao et al., 1996; Doyle and Barros, 2002; Raia and Cavalcanti, 2008), the continental low is directly related to the large-scale low-level wind convergence over the Amazon and along the SACZ (Lenters and Cook, 1995). On the other hand, Drummond et al. (2008) showed that the tropical North Atlantic might be an additional moisture source for Central Brazil, depending on the considered months. In this context, Arraut and Satyamurty (2009) showed that this oceanic sector is the main moisture source for the austral summer rainfall in the Amazon and South American subtropics.
Recently, Garcia and Kayano (2010), hereinafter referred to as GK10, showed that the convective activities in the SAMS and Atlantic ITCZ areas are closely related. They determined the seasonal modes of the convection in the South American region through empirical orthogonal function (EOF) analysis of filtered Outgoing Longwave Radiation (OLR) for the 0.4–1.2 year scale. They found an equatorially anti-symmetric pattern for the first mode which is monsoon related, and a pattern dominantly in the Atlantic ITCZ area for the second mode, both varying in an annual scale. Using the lagged correlation analyses, they found that the first and the second modes evolve into each other within a 1-year interval. This result confirms for the SAMS region a previous finding that the monsoon is just part of the seasonal poleward displacement of the ITCZ obtained from numerical models by Chao and Chen (2001).
In order to get a better understanding of the relationship between the SAMS and the Atlantic ITCZ, the present work complements the preliminary results presented by GK10 by searching for the thermodynamic reasoning of this relationship. This will be done from the analyses of heat and moisture budgets in the SAMS and ITCZ areas. These analyses are necessary because the convective activity in these areas are linked through the monsoon cycle in such a way that as the convective activity grows strong in one area it reduces in the other, as illustrated in Figure 3 by GK10.
It is worthwhile recalling that most papers on the moisture budget are greatly based on the vertically integrated water vapor flux and its convergence, while neglecting the other terms of the water vapor conservation equation. Furthermore, only the residual term in the heat budget equation is usually considered. So, the present work aims to provide a more complete diagnostic view of the moisture and heat budgets related to the SAMS and the Atlantic ITCZ. Thus, all terms of the heat and moisture budget equations are calculated and analyzed separately at several vertical levels for the CAM, WCB, SACZ and ITCZ areas. The locations of these areas are illustrated in Figure 1. The next sections are organized as follows: description of data and methodology in Section 2 and analyses of the moisture and heat budget terms in Section 3. Conclusions are given in Section 4.
2. Data and Methodology
Daily relative humidity (RH), air temperature (T), vertical velocity in pressure coordinate (ω), geopotential height (Z) and zonal and meridional winds (u and v, respectively) in a grid of 2.5° resolution in latitude and longitude are obtained from the National Centers for Environmental Prediction–Department of Energy (NCEP-DOE). Reanalysis-II dataset is available at http://www.cdc.noaa.gov/(Kanamitsu et al., 2002). These variables are obtained for ten standard pressure levels at 1000, 925, 850, 700, 600, 500, 400, 300, 250 and 200 hPa, except for the RH which is available up to 300 hPa. All variables are selected for the 1979–2006 period. The principal component (PC1) of the SAMS mode, identified by GK10 as the first EOF mode of pentad filtered (0.4–1.2 year) OLR series in the domain bounded at 20°N, 20°S, 80°W and 10°W, is also used. The EOF pattern and PC1 of this mode previously discussed by GK10 are shown in Figures 2 and 3. This mode features a nearly anti-symmetric pattern in relation to the equator with significant positive loadings over tropical South America to the south of the equator and in the equatorial Central Atlantic, and the negative ones elsewhere (Figure 2). The temporal variations of this mode show a well-defined annual cycle with negative amplitudes from November to the end of April (the end of the rainy season over tropical South America) and the positive ones approximately from pentad 30 to 60 (Figure 3). This mode describes a monsoon dipole pattern for the filtered OLR over the SAMS region, as a response to the solar forcing.
Since the specific humidity (q) in the Reanalysis-II dataset is available only at the surface level, this variable at pressure levels is calculated using the saturation vapor pressure (es) equation given in Goff and Gratch (1946), as
where T is the temperature in K, Ts = 373.16 K and ews = 1013.246. The vapor pressure (e) is calculated as . Finally, q is given as
where lev, varying from 1000 to 300 hPa, refers to the level in which q is calculated.
The terms of the moisture budget are obtained at the standard levels from 1000 to 300 hPa using the water vapor conservation equation given as (Peixoto and Oort, 1992)
where S indicates the storage of water vapor and can be expressed as the difference between the evaporation and condensation rates per unit mass, E − C. The assumption that the condensed water (P) falls immediately results in S = E − P (Palmén and Holopainen, 1962). So the water vapor budget per unit mass, in a given point and instant t, can be expressed as
where is the horizontal wind vector. Equation (4) gives the moisture budget for an air parcel, in which term 1 is the local rate of q change; term 2, the horizontal advection of q; term 3, the vertical advection of q and term 4, the residue which reflects sources (E − P > 0) or sinks (E − P < 0) of atmospheric q.
For the heat budget at the standard levels from 1000 to 200 hPa, the thermodynamic energy equation is expressed as (Newell et al., 1974)
where cp is the specific heat at constant pressure , p is the pressure, α is the specific volume and J is the diabatic heating. Since is the vertical velocity in the pressure coordinate and , where ϕ is the geopotential, Equation (5) can be written as
This equation shows the heat budget for an air parcel, in which term 1 gives the local rate of T change; term 2, the horizontal advection of T; term 3, T variations due to parcel adiabatic expansion or compression processes (hereafter referred to as adiabatic term); and term 4, the heat sources and sinks due to diabatic processes such as radiative warming or cooling and atmospheric warming due to the latent and/or sensible heating release (hereinafter referred to as diabatic term).
All terms of the moisture and heat budgets are calculated. Sources and sinks of moisture and heat are the residual terms. It is worthwhile recalling that rather than analyzing the quantitative aspects of the moisture and heat budgets, the vertical structures of these budgets under the SAMS establishment are the main aspects dealt with here. As mentioned above, the calculations are done for four regions where the rainfall is strongly modulated by the SAMS, or encompasses a SAMS-related system: CAM, limited at 2.5°S, 10°S, 62.5°W and 55°W; WCB, limited at 10°S, 20°S, 60°W and 50°W; SACZ area, limited at 15°S, 25°S, 45°W and 35°W; and the ITCZ area, limited at 2.5°N, 7.5°S, 40°W and 30°W (Figure 1).
Each term in Equations (4) and (6) is obtained on a daily basis. In order to remove the influence of high-frequency transients in the analyses, non-overlapping 5-day (pentad) mean values are calculated for each term. As in GK10, pentad series are filtered in the 0.4–1.2 year scale to isolate the annual and semi-annual period oscillations. A wavelet analysis is used as a band-pass filter, which is given by Equation 29 in Torrence and Compo (1998). The wavelet used here is the Morlet wavelet. This wavelet is a complex exponential modulated by a Gaussian, ee, with η = t/s, where t is the time, s is the wavelet scale and ωo is a non-dimensional frequency. The computational procedure of the wavelet analysis described by Torrence and Compo (1998) is used here.
The time series of each term in Equations (4) and (6) is lagged and simultaneously correlated with PC1 of the SAMS mode. This is done at each vertical level and for each grid point of the selected regions. The correlations are calculated with PC1 lagging from lag 0 to 73 pentads the budget terms for all areas. In order to assess the statistical significance of the correlations, the number of degrees of freedom is the number of years, which is 28. The Student's t-test for 28 degrees of freedom gives the thresholds of 0.40 for the correlations to be significant at the 95% confidence level. In order to facilitate the visualization, area averages of simultaneous and lagged correlations are displayed in vertical versus lag plots. Correlations between PC1 and horizontal convergence of moisture flux (HCMF) have also been calculated for the CAM, WCB and SACZ areas. Since the correlations are obtained between the heat and moisture budget terms and the PC1 series, only the areas and the terms are mentioned in the discussions.
In order to facilitate the interpretation of the correlations, it is worthwhile to recall that in Figure 2, for positive (negative) amplitudes of PC1, the positive (negative) OLR anomalies correspond to the absence (presence) of convective activity over tropical South America to the south of the equator and in the equatorial Central Atlantic. Furthermore, as the maximum positive amplitude occurs approximately at pentad 43, the correspondence between the lag of the lagged correlation analysis and the pentad within the 1-year cycle (with 73 pentads) is obtained by adding 43 to the lag. For positive amplitudes, the sign of the correlation coincides with the sign of the anomalies. Thus, positive amplitudes are assumed and the correlations are also referred to as anomalies in the discussion. GK10 analyzed the convective activity in the SAMS area by correlating the PC1 with the filtered OLR for the 0.4–1.2 year scale from lag 0 to 36 pentads (Figure 3). Examining this figure, the onset dates of convective activity in the CAM, WCB, SACZ and ITCZ areas occur in lags 24, 20, 20 and 32 pentads, respectively. In order to determine the demise dates for these areas, the PC1 has been correlated with filtered OLR from lag 36 to 72 (Figure 4). Analysis of this figure indicates that the demise dates of convective activity in the CAM, WCB, SACZ and ITCZ occur in lags 60, 56, 56 and 68 pentads, respectively. Hereinafter, these onset and demise dates will be referred to as OLR-onset and OLR-demise dates to avoid confusion with the onset and demise dates determined in the present analysis.
3.1. Moisture budget
Figures 5, 6 and 7 illustrate the vertical structures of the lagged correlations for each term of the moisture budget in the CAM, WCB and SACZ areas, respectively. In these areas, the local rate of q change shows significant positive (negative) correlations at all levels in most of the first (latest) 36 lags, with the largest magnitudes above 700 hPa for the CAM area, and in the low and upper levels for the SACZ (Figures 5(a), 6(a) and 7(a)). The horizontal advection of q displays significant correlations centered in a lower layer for the CAM and SACZ areas and in an upper layer for the WCB with negative values from lag 20 to 48 pentads and the positive ones for most of the other lags (Figures 5(b), 6(b) and 7(b)). The vertical advection of q for the three areas shows significant correlations at most levels of the 900–300 hPa layer with positive values from lag 20 to 44 pentads for the SACZ and WCB areas and from lag 26 to 50 for the CAM area, and the negative ones for most of the other lags (Figures 5(c), 6(c) and 7(c)). For each area, the residue and the vertical advection of q terms show approximately anti-symmetric patterns (Figures 5(d), 6(d) and 7(d)).
The significant negative (positive) correlations for the vertical advection of q and the opposite sign anomalies of the residue indicate a dry (wet) period and the establishment of a source (sink) of moisture for the atmosphere in these areas (Figures 5(c), (d); 6(c), (d); and 7(c), (d)). The dry (wet) period encompasses that with positive (negative) correlations for the horizontal advection of q. Furthermore, the first date with signals of dry (wet) conditions coincides approximately with the OLR-demise (OLR-onset) dates. In addition, the local rate of q change at most levels starts to increase (decrease) approximately 16 pentads before the OLR-onset (OLR-demise) dates of the three analyzed areas (Figures 5(a), (b); 6(a), (b) and 7(a), (b)). So the local rate of q change seems to be an appropriated parameter for monitoring purposes of the convective activity onset and demise in these areas.
Comparisons of the moisture budget terms among the analyzed areas show some important differences. The onset date of the significant anomalies for the local rate of q, vertical advection of q and the residue terms in the WCB area occur 4 pentads before those in the CAM area. This result confirms the time delay of monsoon OLR-onset (OLR-demise) dates for the CAM in relation to those of the WCB previously noted (GK10; Janowiak and Xie, 2003). Due to the early onset of convection in the WCB area, the vertical structure of local rate of q change average for the 1979–2006 for this area is analyzed in more detail (Figure 8). According to the results presented by GK10 and in Figure 4, it is worthwhile recalling that the onset (demise) of convective activity in the WCB occurs at pentad 63 (26). In Figure 8, the largest magnitude of the positive (negative) values of the local rate of q change in the 1000–900 hPa layer noted from pentad 48 to 64 (20–36) are indicative of local q increase (reduction) prior (after) the convective activity onset (demise) date in the WCB. A possible explanation for these results is the proximity of the WCB to the Bolivian high system which acts to provide (maintain) moisture for the WCB prior (after) the convective activity onset (demise) date.
Different from the other areas, the source and sink of moisture for the atmosphere in the SACZ area are not found on the surface and low levels; rather, they are confined to the 700–300 hPa layer. A possible explanation for this result lies in the differences in the HCMF. The significant correlations between PC1 and the HCMF present comparable magnitudes for the CAM and WCB areas, and are smaller for the SACZ (Figure 9). For the three areas, the dry (wet) period shows positive (negative) values in the upper levels and the negative (positive) ones in the low levels. So, the HCMF anomalies are larger in the CAM and WCB than in SACZ. Since the moisture coming from the tropical Atlantic reaches first the CAM and WCB areas and later the SACZ, more moisture is available at low levels in the CAM and WCB areas than in the SACZ. This difference is relevant in the characterization of the rainy season of these regions because the HCMF field plays an important role in defining the maximum precipitation observed over the Amazon Basin and the SACZ regions (Labraga et al., 2000; Herdies et al., 2002).
Figure 10 illustrates the vertical structures of the lagged correlations for each term of the moisture budget in the ITCZ area. The significant positive (negative) correlations for the local rate of q change occur in time intervals which vary according to the height from lag 16 to 38 (lag 52–73) pentads at 1000 hPa, from lag 26 to 38 (lag 62–73) pentads at 550 hPa and from lag 10 to 30 (lag 46–68) pentads at 300 hPa (Figure 10(a)). The horizontal advection of q shows significant positive (negative) correlations at most levels of the 700–400 hPa layer from lag 0 to 20 and lag 66 to 73 (from lag 30 to 56) pentads (Figure 10(b)). The vertical advection of q shows significant negative (positive) correlations at all levels from lag 0 to 20 and lag 68 to 73 (from lag 34 to 56) pentads (Figure 10(c)). The residue term shows significant positive (negative) correlations in time intervals which vary according to the height from lag 0 to 21 (lag 36–58) pentads at 850 hPa, from lag 6 to 20 (lag 42–56) pentads at 600 hPa and from lag 0 to 18 (lag 32–56) pentads at upper levels (Figure 10(d)). So, local rate of q change starts to increase (decrease) at the low and upper levels 16 and 22 pentads, respectively, before the OLR-onset (OLR-demise) of convective activity. The dry (wet) season in the ITCZ area features a source (sink) of moisture for the atmosphere, as indicated by the dominance of negative (positive) anomalies of the vertical advection of q, and positive (negative) anomalies of residue at all levels approximately from lag 0 to 20 (lag 34–56) pentads.
3.2. Heat budget
Figures 11, 12 and 13 display the vertical structures of the lagged correlations for the heat budget terms in the CAM, WCB and SACZ areas, respectively. The positive (negative) correlations or anomalies of the local rate of T change represent a warming (cooling) tendency of the atmosphere. Significant positive (negative) correlations for this term occur in the CAM area in the 1000–850 hPa layer from lag 0 to 4 and from lag 54 to 73 (lag 18–40) pentads and in the layer 800–200 hPa from lag 12 to 32 (lag 52–73) (Figure 11(a)). The largest significant correlations for the local rate of T change in the WCB are noted in the 700–200 hPa layer, with positive values from lag 6 to 30 pentads and the negative ones from lag 44 to 68 pentads (Figure 12(a)). For the SACZ, positive (negative) correlations for this term are found at most levels from lag 6 to 30 (lag 42–68) pentads (Figure 13(a)).
While the horizontal advection of T presents quite a complicated vertical structure for the CAM area, with vertical reversions of the correlation sign in three layers (1000–840, 800–600 and 400–200 hPa), it shows the largest correlations in the upper levels for the WCB, and in the middle and upper levels for the SACZ (Figures 11(b), 12(b) and 13(b)). The adiabatic term remains at most levels significantly and positively (negatively) correlated from lag 56 to 73 and lag 0 to 8 (lag 20–44) pentads for the WCB and SACZ, and from lag 60 to 73 and lag 0 to 14 pentads (lag 24–48) for the CAM (Figures 11(c), 12(c) and 13(c)). For this term, the positive (negative) anomalies are related to atmospheric warming (cooling) due to the adiabatic compression (expansion) process caused by the descending (ascending) motion during the dry (wet) season. While the diabatic and adiabatic terms feature approximately anti-symmetric patterns for the CAM and WCB areas, the anti-symmetry for these terms is confined approximately in the 840–220 hPa layer for the SACZ area (Figures 11(d), 12(d) and 13(d)). In the WCB area, the significant correlations for the adiabatic and diabatic terms at 925 hPa occur approximately 4 pentads earlier than in the other levels (Figure 12(c) and (d)). These correlations become significant at lag 16 (52) pentads for the wet (dry) season (Figure 12(c) and (d)). This result indicates that the 925 hPa level is an important level for monitoring the adiabatic and diabatic terms in this area.
The opposite signs of the adiabatic and diabatic patterns reflect the relationship between vertical motion and the latent heat release in the tropics and subtropics (Newell et al., 1974). The inclusion of a heat source in the tropics causes ascending motion to balance the thermal source (Hoskins and Karoly, 1981). The negative (positive) anomalies for the diabatic term indicate a heat sink (source) for the atmosphere during dry (wet) season.
Comparisons among the areas of heat budget terms reveal some interesting differences in the timing of the correlations. Significant positive (negative) correlations for the local rate of T change in the 700–200 hPa layer occur 4 pentads earlier in the WCB area than in the CAM area (Figures 11(a) and 12(a)). Consistently, the time lag of 4 pentads is also noted for the significant correlations for the adiabatic and diabatic terms between the CAM and the two other areas. The fact that the significant correlations for these terms in the CAM area lags by 4 pentads the corresponding correlations in the WCB strongly suggests that the rainy season onset and demise start first in the WCB than in the CAM area.
Another interesting aspect noted for the CAM and WCB areas is the time lag between low and upper levels of the significant correlations for the local rate of T change. In order to examine this aspect, the vertical structure of filtered T average for the 1979–2006 period is constructed for the CAM and WCB areas and displayed in Figure 14. This Figure shows a lagged warming in the upper levels in relation to the low levels. The values greater than 0.02 K (less than − 0.02 K) occur in the 1000–850 hPa layer during late winter and spring (late summer and autumn) when the rainy season onset (demise) occurs, and in the upper levels during late spring and summer (winter and early spring) when the wet (dry) season is established (Figure 14(a) and (b)). This warming time lag between low and upper levels might be explained as the following: the low-level warming due to sensible heat release associated with an underlying warm surface creates instable conditions triggering the onset of convective activity; as this activity continues and intensifies, some pentads later, the water vapor condensation and latent heat release occur at the upper levels.
Figure 15 displays the vertical structures of the lagged correlations for each term of the heat budget in the ITCZ area. In this area, the local rate of T change shows significant positive (negative) correlations in time intervals which vary according to the height, with the strongest increase (decrease) of T occurring at 925 and 600 hPa levels during the period 20–24 lag (60–64 lag) (Figure 15(a)). This period includes the SAMS OLR-onset (OLR-demise) dates as indicated by the enhancement (weakening) of convective activity in the CAM, WCB and SACZ areas. The horizontal advection of T features significant correlations in the 1000–850 hPa layer with negative values (cold advection) approximately from lag 0 to 24 pentads and positive values (warm advection) from lag 34 to 60 pentads (Figure 15(b)). The adiabatic term shows significant positive (negative) correlations in most levels from lag 0 to 20 and lag 70 to 73 (from lag 32 to 56) pentads, when convection is absent (present) in the ITCZ area (Figure 15(c)). Therefore, the atmospheric warming (cooling) is due to the adiabatic compression (expansion) caused by downward (upward) motions. The diabatic term shows a pattern approximately anti-symmetric to that of the adiabatic term (Figure 15(d)). For the diabatic term, the significant negative (positive) correlations noted in most levels from lag 0 to 18 (lag 34–56) pentads indicate a heat sink (source) for the atmosphere. The negative (positive) anomalies of horizontal advection of T, the positive (negative) anomalies of the adiabatic term, and the negative (positive) anomalies of the diabatic term indicative of the dry (wet) season in the ITCZ area are noted up to lag 20 (56) pentads, when a warming (cooling) tendency starts with maxima at 925 and 600 hPa. Thus, the local rate of T change might be examined at 925 and 600 hPa for monitoring purposes of the convective activity onset and demise in the ITCZ area.
The vertical structures of moisture and heat budget terms are investigated in the CAM Basin, WCB, SACZ, and in the Atlantic ITCZ for the annual time-scale during the 1979–2006 period. Important differences in these budgets have been found among these areas. Indications of the wet and dry conditions appear 4 pentads earlier in the WCB than in the CAM. Different from the other areas, the SACZ area does not show source and sink of moisture for the atmosphere at the surface level. This might be because the SACZ is far from the moisture source region. Indeed, the anomalies of the HCMF are larger for regions closer to the moisture source region, as the CAM and WCB, than for the SACZ. So, low-level moisture available to be advected upwards is more abundant in the CAM and WCB areas than in the SACZ. This difference is relevant in characterizing the rainy season of these regions because the HCMF field plays an important role in defining the maximum precipitation observed over the Amazon Basin and the SACZ regions (Labraga et al., 2000; Herdies et al., 2002).
In order to support the discussion above, the vertical structures of the residue and diabatic terms average for the 1979–2006 are constructed for the SAMS and ITCZ areas (Figure 16). Several aspects noted from the correlation analyses are confirmed in Figure 16. Positive (negative) values of the residue term and the negative (positive) values of the diabatic term noted in the CAM, WCB and SACZ areas during the austral winter (summer) indicate moisture source (sink) and heat sink (source) for the atmosphere in these areas. These periods correspond to dry (wet) season in the CAM, WCB and SACZ areas. A time lag of 4 pentads for the moisture sink (source) and heat source (sink) between the CAM and WCB areas during the wet (dry) season of both regions is confirmed. Furthermore, the absence of moisture and heat sources and sinks for the atmosphere at the surface level in the SACZ area, and the presence of the strongest moisture and diabatic heat sources and sinks for the atmosphere at the middle tropospheric levels for the CAM and WCB areas are also noted in Figure 16. The ITCZ area in relation to the other areas shows differences in the timing and in the vertical locations of the maximum and minimum residue and diabatic terms. This might be due to the oceanic location of the ITCZ area. For this area, the occurrences of the strongest moisture sources (sinks) for the atmosphere in the surface and low levels, and of the strongest heat sinks (sources) in the middle tropospheric levels are conspicuous. This result confirms the previous finding that the ITCZ is under a vertical convective transport regime (Yanai and Tomita, 1998).
Comparisons between Figures 5 and 10 and between Figures 11 and 15 show that the ITCZ and the SAMS are closely related in terms of moisture and heat budgets. Indeed, the largest anomalies of the vertical advection of q and the residue term of the moisture budget and the adiabatic and diabatic terms of the heat budget during the wet (dry) period for the CAM are found approximately at lag 38 (2), and for the ITCZ, 8 pentads later. Consistently, a time lag of 8 pentads is noted for the moisture sink (source) and heat source (sink) for the atmosphere between the CAM and ITCZ regions during their wet (dry) seasons (Figure 16). In fact, the largest negative (positive) values of the residue term and the largest positive (negative) values of the diabatic term occur at pentad 12 (44) for the CAM and at pentad 20 (52) for the ITCZ. This result provides an additional observational support to the relationship between the SAMS and the Atlantic ITCZ and confirms Chao and Chen (2001) findings.
In general, the warming in the analyzed areas is due to the condensation process because these areas are under convective regime. So, for these areas the heat source coincides with the moisture sink, which is coherent with the Yanai et al. (1973) and Yanai and Tomita (1998) results. The most important terms in the moisture and heat budgets are the vertical advection of q and the adiabatic term, respectively. The corresponding residual terms show an anti-symmetric pattern in relation to the vertical advection of q and the adiabatic term. These aspects are typical of the convective areas, as noted by Veiga et al. (2005) in their study of moisture and heat budget for the Walker circulation in the zonal belt from the equator to 10°S.
To sum up, the present study complements previous results found by GK10, through the analyses of the moisture and heat budgets at several vertical levels in the SAMS and ITCZ regions. It provides thermodynamic reasoning for the relationship between the SAMS and ITCZ found previously. Therefore, these results might have important implications in the modeling studies and monitoring activities of the SAMS and ITCZ.
The authors thank the two anonymous reviewers for their helpful comments and suggestions. The authors were partially supported by the Conselho Nacional de Desenvolvimento Científico e Tecnológico of Brazil. This paper is part of the doctoral thesis of the first author.