• Open Access

Duration of the South America summer monsoon is increasing

Authors

  • V. Brahmananda Rao,

    Corresponding author
    1. Centro de Previsão de Tempo e Estudos Climáticos (CPTEC), Instituto Nacional de Pesquisas Espaciais (INPE), São José dos Campos, SP, Brazil
    • Correspondence to: V. B. Rao, Centro de Previsão de Tempo e Estudos Climáticos (CPTEC), Instituto Nacional de Pesquisas Espaciais (INPE), C.P. 515, 12245-970 São José dos Campos, SP, Brazil.

      E-mail: raovtz@yahoo.com.br

    Search for more papers by this author
  • Sergio H. Franchito,

    1. Centro de Previsão de Tempo e Estudos Climáticos (CPTEC), Instituto Nacional de Pesquisas Espaciais (INPE), São José dos Campos, SP, Brazil
    Search for more papers by this author
  • Manoel A. Gan,

    1. Centro de Previsão de Tempo e Estudos Climáticos (CPTEC), Instituto Nacional de Pesquisas Espaciais (INPE), São José dos Campos, SP, Brazil
    Search for more papers by this author
  • Renato O. P. Gerolamo


Abstract

We studied the interannual variability of the beginning, end and duration of South American monsoon system (SAMS) for the period 1956–2006. The mean beginning date was 10 October, whereas the mean end date was 6 April, and the mean duration was 174 days. The beginning of SAMS showed a tendency to start earlier and the end date showed a tendency to delay; thus, the duration of SAMS showed a tendency to increase by about 7 days in 10 years. We found that the beginning and end dates of SAMS have significant positive correlation with sea surface temperature (SST) of an earlier period at southwest Africa in the South Atlantic Ocean. This suggests the potential of SST in the prediction of the beginning and end dates of SAMS.

1. Introduction

South America was not considered to have a monsoon climate (Ramage, 1971). This is mainly because the low-level winds do not reverse in direction from summer to winter as it happens, for example, in Indian monsoon (Ramage, 1971). However, availability of high-quality data, such as National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR) reanalysis data, revealed that a large part of subtropical South America indeed exhibits typical monsoonal features, including a seasonal reversal of low-level wind direction (Zhou and Lau, 1998; Gan et al., 2004). Similar to other monsoon regions, more than 50% of the total annual precipitation over most of tropical and subtropical South America occurs during the austral summer December–January–February (DJF) (Rao and Hada, 1990; Rao et al., 1996).

The beginning, end and duration of the monsoon rainy season in the region dominated by South America monsoon system (SAMS) have a profound influence in many aspects of social life such as agriculture and hydroelectric power generation. Several studies have been made to define the beginning and the end dates of monsoon systems using different criteria. Fasullo and Webster (2002) and Raia and Cavalcanti (2008) used vertically integrated moisture transport. Kousky (1988) and Marengo et al. (2001) used either outgoing longwave radiation (OLR) or rainfall data to define the onset of the rainy season. However, Franchito et al. (2008) noted errors that can occur if OLR data are used to determine the dates of the beginning and end of the rainy season. Further, the skill of the current numerical models to predict the precipitation in the low latitudes is still low. Gan et al. (2006) suggested the use of wind instead of precipitation because the numerical model's skill in predicting wind components is better. As the monsoon onset and withdrawal is defined mainly by wind changes, a monsoon index based on wind changes is useful in identifying the beginning, end and interannual variability of the monsoon activity. Thus, the purpose of this study is to investigate the interannual variability in the beginning, end and duration of SAMS. Earlier, Carvalho et al. (2011), using a large-scale index for South American monsoon, suggested that the duration of SAMS may be increasing. For this purpose, we use the 850-hPa zonal wind and precipitation index, which is verified to have promise in SAMS by Gan et al. (2004). This index is defined for west central Brazil (WCB) (60°W–50°W, 20°S–10°S). Gan et al. (2004) showed that wind changes over this region represent well the characteristics of SAMS and more importantly the summer time rainfall. In some years, the lack of rain in the waters of the major rivers in WCB leads to dramatic situations, such as one that occurred during the summer of 2000/2001, when the rare power rationing was introduced in the State of Sao Paulo (Franchito et al., 2008).

2. Data and methodology

In this study, the wind data at various pressure levels were obtained from the NCEP/NCAR (Kalnay et al., 1996) for the period 1956–2006. Rainfall data for the same period were obtained from Willmott and Matsuura (2001). For the identification of the beginning and the end of the monsoon, we applied a similar method used by Gan et al. (2004). The method consists in identifying the date when the zonal wind at 850-hPa over the area 10°S–20°S and 60°W–50°W changes direction from easterly to westerly, which is taken as the beginning, and westerly to easterly, which is taken as the end. The beginning (end) of the rainy season is also taken when the rainfall is more (less) than 4 mm day−1 for at least subsequent 6 pentads (5 days averages). This index was called as 850-hPa zonal wind and precipitation index. Sea surface temperature (SST) for the period 1956–2006 was obtained from Reynolds et al. (2002).

3. Results

Table 1 provides the beginning, end and duration of the South America monsoon for the period 1956–2006. The last column (MA5) provides the 5-year moving averages of the duration. Also given are the mean values. The mean onset date was 10 October, whereas the mean ending date was 6 April. The mean duration of the monsoon was 174 days, that is, approximately 6 months. In the period considered, the earliest onset and end days were 12 September and 8 February, respectively, whereas the corresponding latest dates were 9 January and 26 May.

Table 1. Dates of the beginning, end and duration of the SAMS obtained using the 850-hPa zonal wind index and duration of the SAMS filtered with time variation of 5-year running averages (MA5)
YearBeginningEndDuration (days)MA5
56/579 January5 March56
57/5820 November14 April146
58/5926 September14 May231156
59/6026 September14 March171168
60/615 October29 March175168
61/629 December2 April115153
62/634 November29 March146156
63/6430 October2 April156163
64/6514 October22 April190176
65/6624 September17 April206179
66/6729 September28 March181178
67/6819 October27 March161171
68/6918 October20 March154158
69/7018 October17 March151162
70/717 November27 March141162
71/7223 October10 May201167
72/7327 October5 April161166
73/745 October5 April183170
74/7517 October11 March146157
75/7617 October25 March161153
76/7726 September8 February136153
77/7826 October15 March141159
78/7911 October9 April181158
79/8026 October18 April176163
80/8130 October3 April156174
81/8230 October8 April161174
82/835 October18 April196176
83/8410 October7 April181180
84/8519 October22 April186176
85/864 October28 March176175
86/8718 Nov7 April141170
87/8819 October26 April191170
88/8918 October22 March156172
89/9018 October21 April186182
90/913 October6 April186187
91/922 November10 May191191
92/9317 September20 April216193
93/9417 October10 April176195
94/9517 October30 April196198
95/967 October19 April196198
96/976 October29 April206196
97/981 October4 May216189
98/996 October20 March166182
99/0016 October24 March161182
00/0120 October29 March161173
01/0220 September13 April206178
02/0330 September19 March171186
03/0420 October27 April191181
04/059 October7 April201
05/0614 November29 March136
Mean10 October6 April174
Latest9 January26 May
Earlier12 September8 February

Figure 1 shows the variation of the beginning, end and duration of the South American monsoon using 850-hPa zonal wind and precipitation index. Also, this figure shows the linear tendencies. It can be seen from the figure that there is a tendency for the beginning to start early, while the end to get delayed. Thus, it can also be seen that there is a tendency, significant at more than 95% by Student's t-test, for the South American monsoon duration to increase (by about 7 days in 10 years).

Figure 1.

Variation of the beginning (a), end (b) and duration (c) of the SAMS determined using the 850-hPa zonal wind index. Also shown are the linear tendencies.

Figure 2 shows the time variation of 5-year running averages. It can be clearly seen that the duration of the South American monsoon is indeed increasing. Earlier, Gan et al. (2006) have shown that the 850-hPa zonal wind index [this index is different from that used by Gan et al. (2004) because they also used the precipitation] is highly significantly correlated with precipitation (their Figures 1 and 2). Thus, the South America monsoon duration of precipitation is increasing.

Figure 2.

Time variation of 5-year running averages.

The regions including the states of Sao Paulo, Mato Grosso do Sul, Goias and Minas Gerais and the regions northwest of Parana and south of Tocantins exhibit the monsoon regime with high summer rainfall and scanty winter rainfall. These states are crucial for the agricultural production of Brazil. The planting season in these states begins with the onset of monsoon rainfall. Thus, it is interesting to find out how well the 850-hPa zonal wind index is correlated with the beginning of rainfall in these regions and consequently can be useful for agricultural purposes. We calculated the correlation between the onset dates determined using the 850-hPa zonal wind index and rainfall amounts. For the period 1966–2004, the correlation coefficient is 0.295, significant at 95% level, whereas for the period 1977–2004 the value is 0.463, significant at 99% level. This shows the utility of the wind index for the agricultural purposes.

To find out how well the SST is correlated with the onset and end dates, we calculated the correlation between these parameters for three consecutive months (e.g. JFM, FMA, etc.). Figure 3(a) and (b) shows these maps. We calculated the correlation with all the previous three consecutive months both for the onset and the end dates and selected highest correlations. The onset dates are highly significantly correlated with SST of June, July and August near the west coast of Africa. The value of correlation coefficient is 0.45, significant at 99% level. This area is darkened in the figure. In the case of the end dates, the 3-month period of September, October and November shows again high correlation near southwest coast of Africa. The correlation value is 0.35, significant at 99% level. Also a region of 0.25 is seen in north Atlantic near the west coast. This shows the potential of SST in the Atlantic for the prediction of the beginning and end of SAMS.

Figure 3.

(a) Isolines correlation coefficient between beginning dates and SST in JJA; (b) isolines correlation coefficient between end dates and SST in SON.

4. Conclusion

SAMS on an average starts on 10 October and ends on 6 April. The duration of SAMS was 174 days during the period 1956–2006. This result is in agreement with Carvalho et al. (2011) who found 170 days for the period from 1948 to 1972, but it is shorter for the period from 1972 to 1982 where they found 195 days. The beginning showed a tendency to start early and the end showed a tendency to delay; thus, we concluded that the duration of SAMS shows a tendency to increase on an average by about 7 days in 10 years. We found that the beginning and end dates of SAMS show a significant positive correlation with the SST at the Southwest Atlantic Ocean near the west coast of Africa. This shows the potential of SST in the prediction of the beginning and end dates of SAMS.

In this study, we found that over South America during the summer the wind change from easterly to westerly over the area 10°S–20°S and 60°W–50°W is occurring early and reverse wind change is occurring later. These changes are probably linked to the changes in the air temperature. Because of the general global warming the duration of the westerly wind regime that represents the summer monsoon may be increasing. In any case, this hypothesis should be tested with further studies.

Ancillary