El Niño signal variability in the precipitation over southeastern South America during austral summer

Authors


Abstract

[1] Some aspects of El Niño (EN) events signal in the precipitation over southeastern South America (SESA) during austral summer are examined. In February–March this signal covers most of the southeast part of SESA and has a wide range of variability in its intensity. The rainiest EN events are characterized by an intensification of the moisture flow from the amazonic forest to SESA jointly with an enhancement of the subtropical jet over South America. The dynamics in upper-levels is associated with a circulation field in the Southern Hemisphere dominated by the presence of the ENSO pattern described in previous studies. During less rainy EN events the moisture flow towards SESA diminishes; and, the structure of the ENSO pattern of circulation over South America implicates a weakening of the subtropical jet. Therefore, the characteristics of the low and upper levels circulation modulate the intensity of the rain over SESA during these events.

1. Introduction

[2] The signals of El Niño (EN) and La Niña (LN) in the precipitation over southeastern South America (SESA), the region to the east of the Andes between 15°S and 40°S, have been extensively analyzed. Aceituno [1988], Ropelewski and Halpert [1987, 1989], Kiladis and Diaz [1989], Pisciottano et al. [1994] and Grimm et al. [1998, 2000] made a detailed description of their characteristics. These studies showed that the most important feature of the precipitation during ENSO events is a positive (negative) anomaly in the period October-November-December of EN (LN) years in the region formed by the northeast of Argentina, south of Brazil, north of Uruguay and east of Paraguay.

[3] Rainfall variability among EN events and among LN events during austral spring does not depend on sea surface temperature (SST) in the equatorial Pacific. In fact, SSTs in the subtropical South Pacific in EN cases and in the subtropical South Atlantic in LN years are the ones which apparently modulate precipitation over most part of SESA [Barros and Silvestri, 2002]. This non-fully linear relation between the equatorial Pacific SST and precipitation over SESA suggests that the analysis of climatic anomalies or of differences EN minus LN may not be the most appropriated method to explore the signal of ENSO events in this region. An alternative methodology is the study of the anomalies respect to the mean of neutral (NEU) cases, that is to say, the differences EN minus NEU and LN minus NEU. This type of analysis for EN events occurred in February–March (Feb–Mar) between 1953 and 1999 is shown in Figure 1. EN years were determined according to Trenberth [1997] and correspond to EN cases during the austral summer of the following year to the starting of the event (year +). The analysis is focused in SESA, although the differences for the entire South America are shown in Figure 1. Monthly precipitation data were taken from the University of Delaware (UDel) dataset produced by Willmott and Matsuura (available online from http://climate.geog.udel.edu/∼climate/).

Figure 1.

EN minus NEU composite of precipitation in February–March. Areas where values are statistically significant at the 80, 90 and 95% level are shaded in dark, intermediate and light gray, respectively. The contour interval is 10 mm. Negative contours are dashed and the zero contour is omitted. The straight lines between 15°S and 40°S delimit SESA.

[4] Figure 1 shows that during Feb–Mar EN events are significantly rainier than NEU cases in a region that covers the east of Argentina, east of Paraguay and south of Brazil. In LN cases there are significant positive anomalies in northeastern and central part of Argentina (not shown).

[5] The same analysis was made for the previous January–February bimonthly period (not shown). In this case, the significant positive differences in EN events are circumscribed to a little region centered in 29°S–58°W, approximately, while in the northeast of SESA, over the continental extension of the South Atlantic Convergence Zone, there are significant negative values in LN composite. Therefore, the period Feb–Mar better depicts the EN signal in the precipitation over SESA and therefore the analysis is focused in these two months.

[6] The positive values during Feb–Mar in EN and LN cases are describing a non-linear relation between the precipitation in the southeastern area of SESA and the equatorial Pacific SST. This raises the question if the lack of linearity is also a feature of the rainfall response in EN events as it occurs in austral spring. Thus, the objective of this study is to analyze the variability of the precipitation in SESA during EN events in austral summer. The methodology adopted is described in the following section.

2. Variability Among EN Events

[7] Figure 1 shows that in Feb–Mar there are positive differences EN minus NEU in the southeast of SESA with maximum significance in the region limited by 25°S–31°S and 52°W–59°W. Hereafter this region will be referred as REN and the precipitation over it is used as an index to analyze the variability among EN events.

[8] The analysis of the Southern Hemisphere (SH) circulation features during EN events was made using data of streamfunction at 0.2 σ-level and winds at 850-hPa of the reanalysis NCEP-NCAR [Kalnay et al., 1996]. In South America, radiosonde observations were started in 1958, but were completed with Brazilian radiosonde from 1963 on. Therefore, the reanalysis NCEP-NCAR of the circulation fields in South America after this year are more reliable. For this reason, the variability of EN events was analyzed considering EN and NEU cases occurred between 1966 (year of the first EN event in Feb–Mar after 1958) and 1999 (last year available in the UDel dataset of precipitation).

[9] Figure 2 shows precipitation in REN during the twelve EN events occurred in Feb–Mar between 1966 and 1999. The horizontal lines indicate the magnitudes of MNEU and VNEU = [MNEU + σNEU] being MNEUNEU) the mean (standard deviation) during the fourteen NEU cases produced in this period. In five EN events precipitation is higher than VNEU (years 1966, 1983, 1992, 1995, and 1998, hereafter referred as STRONG cases) and in seven events precipitation is lower than that value (years 1969, 1970, 1973, 1977, 1980, 1987 and 1993, hereafter referred as WEAK cases).

Figure 2.

February–March precipitation in REN region during EN events. The dotted (solid) line indicates the magnitude of MNEU (VNEU).

[10] The differences of precipitation STRONG (STR) minus NEU and WEAK (WEA) minus NEU are shown in Figure 3. STR events are significantly rainier than NEU cases in a wide region that covers the northeast of Argentina, Paraguay and south of Brazil. Significant negatives anomalies are observed in the western extreme of SESA near the Andes between 20°S and 30°S. In WEA events the most important differences are produced in the northeast and northwest of SESA, in both cases there are negative differences. Therefore, the value VNEU separates one group of EN events significantly rainier than NEU cases in the east of SESA (STR cases) from other group which mean precipitation is not significantly different from the mean of NEU years (WEA cases).

Figure 3.

(a) STR minus NEU composite of precipitation; (b) as in Figure 3a but for WEA minus NEU. The contour interval is 20 mm and areas with statistically significant values as in Figure 1. Negative contours are dashed and the zero contour is omitted.

[11] The NCEP-NCAR SST fields associated with the precipitation anomalies described in the previous paragraph are shown in Figure 4. In the equatorial central Pacific the warm anomalies are higher in STR events (Figure 4a) but the region with significant differences is very similar in both composites. Moreover, in STR cases there are significant negative anomalies in the subtropical south Pacific and in the western equatorial Pacific which are not seen in WEA composite.

Figure 4.

(a) STR minus NEU composite of SST; (b) as in Figure 4a but for WEA minus NEU. Areas with statistically significant values as in Figure 1. The contour interval is 0.3°C. Negative contours are dashed and the zero contour is omitted.

[12] The STR minus NEU and WEA minus NEU composites of 0.2 σ-level streamfunction are shown in Figure 5. Observed in STR cases is the ENSO pattern of circulation described by Karoly [1989] and Kidson [1999], among others: a wave train extending to the southeast from the equatorial central Pacific turning to the northeast near the Antarctic Peninsula and reaching subtropical latitudes over the Atlantic Ocean. Another wave train, with some similar characteristics to the Pacific-South American 2 (PSA2) mode described by Mo and Higgins [1998], can be distinguished between the northwest of New Zealand and middle latitudes of SH. Over SESA, the cyclonic anomaly in the south together with the anticyclonic anomaly in the north produce an enhancement of the subtropical jet and a cyclonic vorticity advection over SESA, two favorable conditions for the enhancement of rainfall over the region.

Figure 5.

As Figure 4 but for streamfunction at 0.2 σ-level. Contour interval is 1 × 106 m2 s−1. Negative contours are dashed and the zero contour is omitted.

[13] In WEA composite (Figure 5b) upper-level circulation anomalies are different from the STR composite, specially over SESA: the ENSO pattern is shifted nearly 30° to the west of its respective position in STR cases and over the south of South America there is an anticyclonic center in 35°S–45°W, approximately, that provoke a weakening of the subtropical jet over SESA. In other words, the upper-level circulation field over South America presents dynamical conditions not favorable for the development of precipitation over SESA. This is in accordance with the differences of precipitation shown in Figure 3b.

[14] Since UDel precipitation dataset is independent of the NCEP-NCAR reanalysis, the consistency between Figures 3 and 5 strengthens the credibility of the described signal.

[15] Figure 6 shows the anomalies of the low-level circulation over South America for STR and WEA cases. Observed in STR composite is an enhancement of the wet and warm flow to SESA from the amazonic tropical forest associated with the trade winds that enter in the equatorial part of the continent and that then are channeled southward by the Andes similarly as it was described by Berbery and Collini [2000]. The trade winds feed this flow by the enhancement of the south Atlantic anticyclone (Figure 6a).

Figure 6.

As Figure 4 but for wind at 850-hPa. A reference vector of 1 m s−1 is shown at the lower right-hand side of each panel.

[16] In WEA composite, the field of circulation does not show enhancement of the flow to SESA from the northwest. There is an anticyclonic anomaly centered in 20°S–55°W, approximately, that weakens the advection of moisture from the northwest.

[17] Figure 4 indicates that the equatorial central Pacific is warmer in the rainier EN events in SESA. To analyze this result in more detail, Figure 7 shows the scatter diagram between the precipitation in REN and the SST of EN3.4 region. A relation of the type warm-wet/cold-dry between these variables is not observed (the correlation coefficient is 0.22) and there are even LN years with the same precipitation as EN years. However, a clear linear relation exists among EN events (the correlation coefficient considering only EN events is 0.69, significant at 95% level). Therefore, in Feb–Mar the REN region is rainier during EN events if the equatorial central Pacific is warmer. However, no similar relation exists between these variables when the whole period or only LN events or only NEU cases are considered.

Figure 7.

Scatter diagram between the precipitation in REN region and the SST in EN3.4. Both variables in February–March.

[18] The characteristics of the relation between the precipitation in SESA and the SST of EN3.4 described by Figure 7 are similar to those obtained using the SST of EN3.4 in the previous period November–December (Figure 8). In this case the correlation coefficient is 0.20 in the entire period 1966–99 and 0.75 (significant at 99% level) considering only EN cases. Thus, the SST of equatorial central Pacific in the previous spring seems to be a good predictor of the precipitation over the eastern part of SESA during EN events of austral summer.

Figure 8.

Scatter diagram between the precipitation in REN during February–March and the SST in EN3.4 in the previous November–December.

3. Summary and Conclusions

[19] The most significant differences between EN and NEU years in the precipitation over SESA during austral summer are focused in February–March in a region that covers the east of Argentina, east of Paraguay and south of Brazil. There is an important variability in the intensity of these differences and it is possible to define a significantly rainier group of events than NEU cases (STR events) and another group with precipitation not significantly different from the rainfall average of NEU cases (WEA events).

[20] The upper-level circulation in STR cases exhibits the typical ENSO pattern extending between the equatorial central Pacific and middle latitudes of SH jointly with another wave train that is propagated southeastward from the northwest of New Zealand. This structure generates dynamical conditions that favor the development of rain over the eastern part of SESA due to it produces an enhancement of the subtropical jet and cyclonic vorticity advection over this region. In WEA cases, the ENSO pattern is shifted to the west of its respective position in STR composite and the upper-level circulation presents an anticyclonic anomaly at the southeast of South America that produces a weakening of the subtropical jet. This feature contributes to inhibit the rain over SESA. The field of low-level circulation shows that the moisture flow from the Amazon to SESA is enhanced (weakened) in STR (WEA) cases. Therefore, the differences in the circulation patterns of low and upper-levels cause the rainfall differences between these two groups of EN events.

[21] The equatorial central Pacific is warmer in STR than in WEA events. This relation is also observed in the SST of the previous spring, suggesting a possible tool for forecasting austral summer precipitation in SESA during EN events.

[22] Due to its influence in the precipitation over SESA, further studies must be done to determine the causes of the variability observed in the anomalies of SH circulation during EN events in austral summer. It is necessary to analyze the propagation of the wave trains generated over the equatorial Pacific taking into account that the wave propagation from tropical regions to higher latitudes depends not only on the characteristics of the ocean-atmosphere system in the excitation region but also on the conditions that the wave finds along its posterior trajectory [Trenberth, 1993]. Another atmospheric characteristic that must be studied is the interaction between the principal modes of SH circulation variability. Although the structure and dynamics of these principal modes have been extensively documented so far [e.g., Karoly, 1989; Kidson, 1999; Limpasuvan and Hartmann, 2000; Mo, 2000; Renwick, 2002], the possible response of the SESA precipitation to the interaction between these modes has not yet been analyzed.

Acknowledgments

[23] This work was funded by ANPCyT (PICT 07-09950) and CONICET (PIP 02339). The author is grateful to Dr. Vicente Barros for his comments and suggestions.

Ancillary