Comparison between winter precipitation in southeastern South ...

GEOPHYSICAL RESEARCH LETTERS, VOL. 32, L05709, doi:10.1029/2004GL021749, 2005

Comparison between winter precipitation in southeastern South America during each ENSO phase Gabriel E. Silvestri Consejo Nacional de Investigaciones Cientı´ficas y Tećnicas (CONICET), Buenos Aires, Argentina Received 14 October 2004; revised 31 January 2005; accepted 9 February 2005; published 8 March 2005.

[1] Some aspects of winter precipitation over Southeastern South America (SESA) during El Ninõ (EN) and La Ninã (LN) events are analyzed in this paper. In June – July, the precipitation over a vast area in the north of SESA is similar in both ENSO phases and higher than in neutral cases. On the other hand, there is a wide rainfall variability among EN and among LN cases, which is not associated with SST of EN regions. Similar characteristics are observed in the precipitation over the south of SESA. The obtained results describe a nonlineal relationship between the SST of the equatorial Pacific and the precipitation over most of the southeast of South America during austral winter. Citation: Silvestri, G. E. (2005), Comparison between winter precipitation in southeastern South America during each ENSO phase, Geophys. Res. Lett., 32, L05709, doi:10.1029/ 2004GL021749.

1. Introduction [2] Southeastern South America (SESA) has a significant El Ninõ-Southern Oscillation (ENSO) signal on precipitation, and as a consequence, on the streamflow of its rivers. SESA extends over Uruguay, Paraguay, southern Brazil and eastern Argentina and it is the region where great part of agricultural activities and hydroelectric power generation of these countries are concentrated. Therefore, the climatic variability associated with ENSO is relevant for the regional economy and its forecast may help to diminish huge damages caused by floods which accompany EN events. [3] A number of authors have discussed the ENSO signal in the precipitation in regional or global studies which include SESA. Pittock [1980] and Aceituno [1988] showed a significant negative correlation between the Southern Oscillation index (SOI) and precipitation in eastern Argentina during austral spring. Ropelewski and Halpert [1987, 1989] found positive precipitation anomalies in SESA during spring and summer of EN years and negative anomalies in the period July – December of LN years. Kiladis and Diaz [1989] found similar results through the analysis of EN minus LN composites. Grimm et al. [2000] showed that the strongest signal of EN and LN events is produced in spring of the onset year. There are more localized descriptions based on analysis of different areas of SESA, [e.g., Rao and Hada, 1990; Pisciottano et al., 1994; Grimm et al., 1998], which are consistent to the results of global and continental studies. [4] In the articles mentioned in the previous paragraph were used one or more of the following methodologies to Copyright 2005 by the American Geophysical Union. 0094-8276/05/2004GL021749

analyze the signals of EN and LN events on the precipitation fields. i) Linear correlations between precipitation or some index of it and the SOI or sea surface temperature (SST) in EN regions, ii) Anomalies of precipitation during EN and LN events with respect to the climate average and iii) Precipitation differences between EN and LN cases. The use of these methods implicitly assumes the existence of a lineal relationship between the equatorial Pacific and the precipitation. But this relationship may not be entirely of this type, as it was shown by Silvestri [2004] for austral summer, and therefore it is possible that these tools are not to be the most appropriate to explore the ENSO signal in SESA. Then, the differences between the precipitation during EN, LN and neutral (NEU) phases are analyzed to know some aspects of this signal during austral winter. The study is focused in June –July, because in section 3 will be shown that in this period the differences of precipitation between the three ENSO phases are more significant in SESA during the South Hemisphere (SH) cold season.

2. Data [5] Monthly precipitation data were obtained from the University of Delaware (UDel) dataset (http://climate.geog. udel.edu/climate/) and values of SST of the regions EN3.4 and EN1+2 were taken from Climate Diagnostics Center (http://www.cdc.noaa.gov/ClimateIndices/). This study covers the period 1953 – 99 and ENSO events were identified according to Trenberth [1997]. [6] The use of UDel dataset allows to have information of precipitation of all SESA in the period 1953 – 99. It is important to emphasize that the characteristics that will be described using this data (UDel) are consistent with those from a network of 140 rain gauge stations. Rain gauge time series for 1953 – 1999 were obtained from the National Meteorological Service of Argentina, National Direction of Meteorology of Paraguay. Series of Bolivia, Brazil and Uruguay were obtained from the National Center for Atmospheric Research (NCAR) Monthly Climate Data for the World.

3. Austral Winter Precipitation in SESA During Each ENSO Phase [7] Figure 1 shows the differences of precipitation in SESA between EN, LN and NEU cases in the period June– July (Jun – Jul). Significance is calculated using the Wilcoxon-Mann-Whitney nonparametric test [Wilks, 1995]. [8] It is observed that a wide area in the north of SESA has values of the same sign (positive) in EN minus NEU and LN minus NEU composites (Figures 1a and 1b). In fact, in EN minus NEU, positive significant values extend from

L05709

1 of 4

L05709

SILVESTRI: NONLINEAR ENSO SIGNAL IN SOUTH AMERICA

L05709

Figure 1. a) Composite difference of precipitation EN minus NEU in June – July; b) As in a) but for LN minus NEU; c) As in a) but for EN minus LN. Areas with statistically significant values at 85, 90 and 95% are shaded in dark, intermediate and light gray, respectively. The contour interval is 5 mm. Negative contours are dashed and the zero contour is omitted. the east of Bolivia to the south of Brazil, whereas in LN minus NEU composite cover Paraguay, the northeast of Argentina and the south of Brazil. On the other hand, in the south of SESA, there are significant positive differences between EN and NEU cases but nonsignificant values in LN minus NEU composite, except for a little sector in the east of Uruguay. The difference EN minus LN shows significant values in the southeast of SESA (Figure 1c). [9] The differences observed in Figure 1 are due to a zonal displacement of the isohyets. In fact, Figure 2 shows that there is a southeastward displacement of the isohyets in the north part of SESA in NEU cases regarding its position during EN and LN events; whereas, in the south of the region, there is an eastward displacement in LN and NEU cases regarding its position during EN events. [10] The results of Figure 1 suggest that in austral winter there is not a relationship of warm-wet/cold-dry type between the SST of the equatorial Pacific and the precipitation in SESA, as it occurs in spring and fall [e.g., Grimm et al., 2000; Montecinos et al., 2000]. [11] A similar analysis was done for the periods April – May and August – September. In both cases there are no vast areas with significant values over SESA except for the difference EN minus NEU of April – May in the north of Paraguay (positive magnitudes). Therefore, there are significant differences during Jun – Jul between the ENSO phases in the precipitation over SESA, which are not observed in the rest of the SH cold season (semester April – September).

[12] Trenberth [1997] used the SST in the region EN3.4 to identify the beginning and the end of each ENSO phase. Here, this index is used to explore in detail the relation between the equatorial Pacific and the precipitation over SESA in Jun – Jul. Figure 1 shows that during this period EN and LN events are significant rainer than NEU cases in the area between 20°S – 30°S and 50°W –60°W, approximately. This area, hereinafter referred as the north region (N), covers the north part of Cuenca del Plata [e.g., Berbery and Collini, 2000]. EN events are also significant rainier than NEU cases in the sector limited by 34°S – 39°S and 57°W – 63°W, which is a region of very important agricultural and cattle raising production [e.g., Labraga et al., 2002], hereinafter referred as south region (S). [13] The scatter diagram between the precipitation in N and the SST in EN3.4 shows that there are LN events rainier than EN events (Figure 3a). Moreover, precipitation during EN cases has a wide variability which does not seem to be associated with the conditions in the EN3.4 region. In fact, precipitation in the events of years 1957, 1983 and 1994 was more than M (climatic average) plus SD (standard deviation) and in years 1953, 1963, 1969, 1992 and 1993 was less than M but in all these cases the SST in EN3.4 was almost the same. It suggests that the magnitude of the precipitation over N during EN events does not depend on the intensity of the warm anomaly in the equatorial Pacific. It is also observed a wide variability independent of the conditions of the region EN3.4 in the precipitation during

Figure 2. a) Mean precipitation during EN (solid line) and NEU cases (dotted line). Only the isohyets of 30 and 60 mm are indicated; b) As in a) but for LN and NEU cases. 2 of 4

L05709


L05709

[16] Figure 4 shows the relation between the precipitation in S and the SST in regions EN3.4 and EN1+2. The climatic anomalies were positive in 86% of EN cases (twelve out of fourteen EN events) and in 44% of LN cases (four out of nine LN events). It indicates a high probability that an EN event will give a positive climatic anomaly but the situation is ambiguous in LN cases. The lineal correlation is significant between the precipitation and the SST in EN3.4 but, similar to Figure 3b, more (less) precipitation is not necessarily associated with a warmest (coldest) EN3.4 region.

4. Conclusions

Figure 3. a) Scatter diagram between SST in EN3.4 and precipitation in N; b) As in a) but for SST in EN1+2. Values of M (solid line) and M ± SD (dotted lines) are indicated. Correlation between both variables is showed in each figure (correlations significant at 95% are individualized with asterisks).

[17] EN and LN events are significantly rainier than NEU cases in a large part in the north of SESA during Jun – Jul. Moreover, there are no significant values in the differences EN minus LN and the precipitation in some LN events is more than the rainfall in some EN cases and vice-versa. Therefore, the response of the rain in this region to ENSO events is clearly nonlineal. In the southeast of SESA the precipitation during EN phase is significantly greater than in LN and NEU cases, but there are not significant values in the difference LN minus NEU. [18] This analysis shows that during austral winter there is not a relation of warm-wet/cold-dry type between the SST of the equatorial Pacific and the precipitation over most of SESA, as it is observed in spring and fall. Also, rainier EN or LN events are not associated with a warmer equatorial Pacific.

LN events. In effect, if the year 1988 is omitted, the SST in EN3.4 is not associated with the differences of precipitation among LN events. These characteristics are similar to the ones described by Barros and Silvestri [2002] for austral spring. On the other hand, Silvestri [2004] found that in austral summer most part of SESA is rainier if the equatorial central Pacific is warmer. [14] The characteristics of the relation between SST in EN3.4 and precipitation in N mentioned in the previous paragraph are also observed if SST in any other EN region is considered. For example, Figure 3b shows the relation with the SST in EN1+2 (the easternmost EN region). [15] The scatter diagrams of Figure 3 show that in only 64% of EN cases (nine out of fourteen EN events) the precipitation was more than the climatic mean in the N region. Said percentage does not let to say that the odds are that an EN event will give a positive climatic anomaly. It can be conjectured in LN events due to the fact that in 78% of the cases (seven out of nine LN events) the precipitation was more than the climatic mean. Precipitation in this region has significant correlation with SST in EN1+2 (Figure 3b). But, even though the lineal correlation coefficient is significant positive, the scatter diagram reveals that the relation between these variables is not of the warm-wet/ cold-dry type as in spring and fall. Said result is known from the analysis of Figure 1. 3 of 4

Figure 4. As Figure 3 but for precipitation in S.

L05709


[19] It is important to mention that the analysis presented in this paper was done with fourteen (nine) EN (LN) cases. This quantity may not be enough to assure that the relations which were found describe irrefutably the characteristics of the link between the precipitation in SESA and the SST when there is an abnormal heating (cooling) in the equatorial Pacific during austral winter. Unfortunately, the available information of the precipitation in SESA and of SST is not suitable to do an analysis of the events which took place before ’50s. Therefore, the description in this article is not conclusive, but it illustrates the characteristics observed in the events of the last fifty years. [20] Acknowledgments. The author is thankful to Dr. Vicente Barros for the revision of a draft of this manuscript. He is also grateful to the anonymous reviewers for their constructive comments and suggestions. This work was funded by ANPCyT (PICT 07-09950) and CONICET (PIP 02339).

References Aceituno, P. (1988), On the functioning of the Southern Oscillation in the South American sector. Part I: Surface climate, Mon. Weather Rev., 116, 505 – 524. Barros, V. R., and G. E. Silvestri (2002), The relation between sea surface temperature at the subtropical south-central Pacific and precipitation in southeastern South America, J. Clim., 15, 251 – 267. Berbery, E. H., and E. A. Collini (2000), Springtime precipitation and water vapor flux over southeastern South America, Mon. Weather Rev., 128, 1328 – 1346. Grimm, A. M., S. E. T. Ferraz, and J. Gomes (1998), Precipitation anomalies in southern Brazil associated with El Ninõ and La Ninã events, J. Clim., 11, 2863 – 2880.

L05709

Grimm, A. M., V. R. Barros, and M. E. Doyle (2000), Climate variability in southern South America associated with El Ninõ and La Ninã events, J. Clim., 13, 35 – 58. Kiladis, G. N., and H. F. Diaz (1989), Global climatic anomalies associated with extremes in the Southern Oscillation, J. Clim., 2, 1069 – 1090. Labraga, J. C., B. Scian, and O. Frumento (2002), Anomalies in the atmospheric circulation associated with the rainfall excess or deficit in the Pampa Region in Argentina, J. Geophys. Res., 107(D23), 4666, doi:10.1029/2002JD002113. Montecinos, A., A. Diaz, and P. Aceituno (2000), Seasonal diagnostic and predictability of rainfall in subtropical South America based on tropical Pacific SST, J. Clim., 13, 746 – 758. Pisciottano, G., A. Diaz, G. Cazes, and C. R. Mechoso (1994), El Ninõ – Southern Oscillation impact on rainfall in Uruguay, J. Clim., 7, 1286 – 1302. Pittock, A. B. (1980), Patterns of climatic variation in Argentina and Chile. Part I: Precipitation, 931 – 960, Mon. Weather Rev., 108, 1347 – 1361. Rao, V. B., and K. Hada (1990), Characteristics of rainfall over Brazil: Annual variations and connections with the Southern Oscillation, Theor. Appl. Climatol., 42, 81 – 90. Ropelewski, C. H., and S. Halpert (1987), Global and regional scale precipitation patterns associated with the El Ninõ/Southern Oscillation, Mon. Weather Rev., 115, 1606 – 1626. Ropelewski, C. H., and S. Halpert (1989), Precipitation patterns associated with the high index phase of the Southern Oscillation, J. Clim., 2, 268 – 284. Silvestri, G. E. (2004), El Ninõ signal variability in the precipitation over southeastern South America during austral summer, Geophys. Res. Lett., 31, L18206, doi:10.1029/2004GL020590. Trenberth, K. E. (1997), The definition of El Ninõ, Bull. Am. Meteorol. Soc., 78, 2771 – 2777. Wilks, D. S. (1995), Statistical Methods in the Atmospheric Sciences, 467 pp., Elsevier, New York.

G. E. Silvestri, Consejo Nacional de Investigaciones Cientı´ficas y Tećnicas (CONICET), Buenos Aires, Argentina. ([email protected])

4 of 4