Patterns leading to extreme events in Argentina: partial and generalized frosts

Transcription

1 INTERNATIONAL JOURNAL OF CLIMATOLOGY Int. J. Climatol. 7: 7 7 (007) Published online February 007 in Wiley InterScience ( DOI: 0.00/joc.7 Patterns leading to extreme events in Argentina: partial and generalized frosts Gabriela Viviana Müller* Instituto de Astronomia, Geofísica e Ciências Atmosféricas Universidade de São Paulo, Brasil Abstract: This paper studies the atmospheric circulation associated with frosts in the central region of Argentina with the objective of improving the prediction capacity for these events. Frosts are selected according to a spatial principle, which takes into consideration those events affecting an area above 5% of the weather stations located in the region known as Wet Pampas. Partial frosts (PF, less than 75% of the stations) and generalized frosts (GF, above 75% of the stations) constitute two different groups: one of them given by PF + GF and the other one only by the PFs. The analysis comprises the period between 9 and 990, where two extremes are selected: those years with a number of frosts one standard deviation above and below the mean (+σ and σ respectively). In each extreme group (i.e. PF + GF +σ and σ ;PF+σ and σ ), the average number of frosts varies according to the different periods of the year. Therefore, the following periods selected independently are studied: Annual of frosts (from April through October), seasonal (from May through September and from June through August), and each of the involved months individually. The mean circulation patterns analyzed from the composites of different meteorological variables explain the temperature anomalies observed in the region under study. In this way, the extreme occurrence of frosts is a result of the prevailing circulation during the different periods under consideration. Within each group (PF + GF and PF), the composites of the extremes +σ and σ show opposite anomaly patterns for the different periods. Results for +σ ( σ) are distinguished by negative (positive) temperature and humidity anomalies in the Wet Pampas for the composites of PF + GF and PF. At low levels, the wind anomalies show the prevalence of winds from the south (north) in the southern section of the continent. At upper levels, this pattern is accompanied by anomalies of the prevailing westerly (easterly) wind for PF + GF and PF. This means a subtropical jet over South America more (less) intense than normal is associated with a higher (lower) frequency of frost occurrence in the seasonal composites and the Annual. However, there are variations in wind anomalies at 50 hpa within the same extreme +σ ( σ) when comparing the groups PF with the previously studied group (formed only by the GF). Composites of group GF +σ showed a maximum anomaly of the westerly zonal wind located at the jet entrance region into South America, over the Andes. On the other hand, the composites for PF +σ show a component from the northwest located leeward the Andes, extending toward the southeast up to the Atlantic Ocean. The same anomaly for the extreme σ extends from the north of Argentina up to tropical latitudes, and it is more intense for groups PF + GF and PF than for group GF. Additionally, an analysis of the daily evolution of PF events during winters GF +σ shows that the PFs are caused by a cyclonic anomaly moving toward the east of the continent, causing cold air advection from the south. The same analysis for the GFs shows an anticyclonic anomaly dominating the southern area of South America, i.e. the anticyclogenesis that causes the GFs does not occur in the PFs. Copyright 007 Royal Meteorological Society KEY WORDS frost; partial frost; generalized frost; extreme event; circulation patterns; Wet Pampa Received May 00; Revised November 00; Accepted November 00 INTRODUCTION An extreme event is associated, by definition, with atypical sporadic features of the atmospheric circulation. Although temperatures below 0 C are not necessarily an extreme phenomenon at mid latitudes in winter, they become one when they appear recurrently or they persist for several days affecting vast areas. Similarly, * Correspondence to: Gabriela Viviana Müller, Departamento de Ciências Atmosféricas, Instituto de Astronomia, Geofísica e Ciências Atmosféricas, Rua do Matão, Universidade de São Paulo. E- mail: gabriela@model.iag.usp.br the absence of frosts or very low frost frequencies also constitutes extreme phenomena. This paper is aimed at analyzing the circulation associated with extreme phenomena given by maximum and minimum frequencies of frosts occurrence. From the synoptic perspective, some papers addressed these extreme winter situations, which caused temperature drops in the central-eastern region of Argentina (Escobar and Bischoff, 999; Barrucand and Rusticucci, 00; Escobar et al., 00). Other studies identified the situations giving rise to cold waves in the country (Rusticucci and Vargas, 995). Pezza and Ambrizzi (005) Copyright 007 Royal Meteorological Society

2 7 G. V. MÜLLER studied the dynamic conditions associated with different types of cold surges affecting subtropical South America and proposed a conceptual model to explain the propagation of these waves. Their results show a pattern, which is similar to the behavior expected to produce very cold temperatures over Central and North Argentina, although shifted toward the northeast and therefore indicating a later development of the wave train. It is interesting to note that extreme cold waves affecting Argentina do not necessarily propagate to Brazil, owing to a blocking pattern responsible for the inverse pattern of temperature anomalies to the south and to the north of 0 S (Pezza and Ambrizzi, 999). On the other hand, those waves that affect very low latitudes can also be generated by a very strong meridional displacement of mid-latitude air into the tropics in such a way that Argentina would remain unaffected during all stages. In particular, the synoptic types that cause frosts in the Wet Pampas were typified by Müller et al. (00). The Wet Pampas, which comprises an extensive region in the center-east of Argentina, has great economic importance, as it is the main agricultural production region in the country, with strong impact on the international markets. According to the synoptic-climatic classification, there are six patterns that explain 9% of the variance and therefore represent the most frequent synoptic conditions causing frosts in the area under study. The most important patterns (A, B, and C) are represented by the first three principal components (PCs) in decreasing order of the explained variance. In Pattern A, there is a high-pressure system north of 0 S, which causes radiative cooling due to its weak gradients that give rise to light winds or calm over the Wet Pampas. A postfrontal high-pressure system moving poleward onto the continent at 0 S is observed in Pattern B. The south to southwest wind generated causes a marked advection of very cold and dry polar air, favoring the occurrence of advection frosts. Pattern C represents a low-pressure system to the east of the continent and a high-pressure system entering from the Pacific, affecting the Wet Pampas from the west and producing a strong flow of cold and dry air from the south. The remaining patterns (D, E, and F) are connected to cold anticyclones that cause advective and/or radiative frosts and represent a possible, though rare, condition that is important due to its effect on the Wet Pampas. Previously, Müller et al. (000) grouped frosts in the Wet Pampas according to their spatial distribution for the period between 9 and 990, i.e. the quotient between the number of meteorological stations with frosts for each day, and the total number of meteorological stations with available data on that day, defining isolated frost, partial frost (PF), and generalized frost (GF). In this paper, and more recently in Müller (00), an important interannual variability in the frequency of frosts is shown. Therefore, the years showing extreme frequencies in a period of 0 years have been isolated for the analysis of the related circulation. This last criterion was applied in particular to GFs in the Wet Pampas by Müller et al., 005 (hereafter M5), when analyzing the mean circulation of the patterns leading to GF. From the comparison between the composite of the two extremes (maximum and minimum frequency of GF occurrence), it was possible to determine that the anomaly fields are opposed in practically all studied periods (annual, seasonal, and monthly periods). In particular, the analysis of atmospheric fields for the GF event composites, selected within the winters with maximum frequency of GF occurrence, indicates that these episodes are related to large-scale patterns characterized by well-defined wind and pressure fields. These are conditioned by dynamic factors resulting from the propagation of waves that reach South America during the extreme cold events. A double Rossby wave train is led to the continent by an intense zonal flow in subtropical and polar latitudes, respectively. This configuration favors the development of an intense southerly wind anomaly with a large meridional extension that intensifies the anticyclonic circulation over the entire southern cone of the continent for the most frequent GF events (M5) as well as for the most persistent GF events (Müller and Berri, 00). A more detailed analysis of frost events is undertaken in view of the results obtained for the GFs, taking now into account the frosts groups consisting of PF + GF and only PF. Following the same criteria of M5, both extremes, maximum and minimum frequency of frosts occurrence, are selected within each group (see section on Criteria, Data and Methodology). The purpose is to examine the characteristics of the mean annual, seasonal, and monthly atmospheric circulation associated with frosts during the different extreme periods. This analysis is aimed at identifying the physical mechanisms involved in the extreme periods of PF + GF (see section on Annual, Seasonal, and Monthly Composites for the Group PF + GF) and PF (see section on Annual, Seasonal, and Monthly Composites for the Group PF). In addition, it also studies the daily evolution patterns associated with the PF (see section on Daily Composites of PF) having place during winters of maximum frequency of GF occurrence, obtained by M5. In this case, the objective is to distinguish the processes that determine the occurrence of PFs from those involved in the GFs previously identified. Finally, the results obtained for both extreme groups PF + GF are compared with those of the extreme periods for GF (see section on Discussion and Conclusions). In this way, the patterns leading to the different frost events in the extreme periods of maximum and minimum frequency of frosts are established. This makes it possible to diagnose the different types of frosts in one of the most important regions for food production and of highest population density in Argentina, known as the Wet Pampas, and contributes toward improving the capacity to predict these events and minimizing the associated damage. CRITERIA, DATA, AND METHODOLOGY The Wet Pampas region lies between 7 0 S and 5 57 W, and the meteorological stations considered Copyright 007 Royal Meteorological Society Int. J. Climatol. 7: 7 7 (007) DOI: 0.00/joc

3 PATTERNS LEADING TO EXTREME EVENTS IN ARGENTINA 75 0N 0S 0S 90W 0W 0W Figure. The rectangular box centered at 5 S and 0 W depicts the Wet Pampa region. in the present paper are located within this area (Figure ). Frosts (defined as those days on which temperature is equal to or less than 0 C) are divided into three groups: isolated, PF, andgf. According to the criterion defined by Müller et al. (000), each day is classified as follows: isolated frost when the percentage of stations with frost in the Wet Pampas is equal to or less than 5% of the total; PF when it is between 5 and 75%; and GF when the percentage is equal to or more than 75% (see details in Müller, 00). On the other hand, the criterion defined by M5 for the GF selects those periods where the number of frosts is one standard deviation unit below ( σ ) and above (+σ ) the mean value for the period Following the same criterion, the extremes in the frequency of frosts occurrence are identified for the groups PF + GF and PF in the same 0-year period. Tables I and II indicate the years that conform the extremes of maximum (+σ ) and minimum ( σ ) frequency of frosts (for PF + GF and PF, respectively) in the several periods selected independently. These periods are as follows: annual (April to October), seasonal (May to September [MJJAS], and June to August [JJA]), and the months of the period annual, individually considered. For the monthly period, the number of years with a threshold of σ is high in the 0-year period and therefore, they are not listed in the tables. The months of April and October, in the groups PF + GF σ and PF σ,have this characteristic (Tables I and II, respectively). Thus, we obtained two groups and the corresponding composites of the different meteorological variables, which met the defined conditions. The anomalies of the following variables were considered: sea-level pressure, upper- and lower-level geopotential height, surface temperature, humidity at 50 hpa, wind vector at 50 hpa and 50 hpa, as well as zonal and meridional wind at 50 hpa. The anomalies were calculated with respect to the 9 99 climatology for each period. The National Centers for Environmental Prediction (NCEP) reanalysis data was used with a horizontal resolution of.5 latitude and.5 longitude (Kalnay et al., 99). In addition, the daily evolution of PF events during the winters GF +σ was analyzed. The selection criterion Table I. Years with extreme occurrence frequency of partial and generalized frosts (PF + GF) for the periods studied. Annual (Apr Oct) MJJAS JJA Apr May Jun Jul Aug Sep Oct +σ σ Table II. Idem Table I for partial frosts (PF). Annual (Apr Oct) MJJAS JJA Apr May Jun Jul Aug Sep Oct +σ σ Copyright 007 Royal Meteorological Society Int. J. Climatol. 7: 7 7 (007) DOI: 0.00/joc

4 7 G. V. MÜLLER for them follows the same principle defined by M5 for the selection of GF, with a minimum period of 7 days between the date of the selected PFs and the next one to be selected. If a GF occurred during this 7-day period, the 7 days are counted from that GF, as it is intended to differentiate the circulation associated with PFs from that associated with GFs (studied by M5). Using the same principle, the PFs to be selected must not be followed by a GF on the next day. Table III shows the dates complying with the established criterion for the winters GF +σ,i.e. 970, 97, and 9. Table III. Date of PF start (year/month/day) selected for the period JJA of GF +σ (see text for details). Partial Frost Events 970/0/05 970/0/ 970/07/9 970/07/7 970/0/0 970/0/ 97/0/0 97/0/7 97/07/5 9/0/0 9/07/0 9/07/0 9/0/0 9/0/09 RESULTS Annual, seasonal, and monthly composites for the group PF + GF When obtaining the composites of the years for each of the extremes +σ and σ for the group PF + GF (Table I), it appears that the variables exhibit an opposite behavior when both extremes are present in many of the analyzed periods. In this way, extreme frost occurrence is shown by the surface temperature anomalies for the composites of JJA (Figure ), with negative (positive) values for +σ ( σ). As expected, the highest anomaly values correspond to JJA, when compared with the composites obtained for the periods annual and MJJAS (not shown). There is a similar relationship for humidity anomalies, as shown in Figure (d). These characteristics are explained by an anomaly in the circulation from the north, noted at 50 hpa on the Wet Pampas for the composite of wind anomaly during the quarter JJA σ (Figure (f)). The extreme +σ wind anomalies are almost nil in the studied region, and from the south in Patagonia (Figure (e)). The periods corresponding to annual and MJJAS have very similar characteristics, as the respective cases of JJA due to the high similarity in the pertinent years (Table I). This is 0S 5S 5S 0. 5S 75W 70W 5W 0W 55W 50W 5W 0W 0S 5S S 5S 5S 5S 75W 70W 5W 0W 55W 50W 5W 0W 0S 5S (d) S S S 75W 70W 5W 0W 55W 50W 5W 0W 5S 75W 70W 5W 0W 55W 50W 5W 0W Figure. Anomalies of surface temperature ( C), specific humidity (g Kg ) at 50 hpa (d), vector wind at 50 hpa (e) (f), sea level pressure (hpa) (g) (h), for the composite of PF + GF for the period JJA +σ (left panel) and σ (right panel). Solid (dashed) lines indicate positive (negative) values. Copyright 007 Royal Meteorological Society Int. J. Climatol. 7: 7 7 (007) DOI: 0.00/joc

5 PATTERNS LEADING TO EXTREME EVENTS IN ARGENTINA 77 0S 0S (e) (f) 0S 0S 0W 90W 0W 0W.5 (g) 0S 0S W 0W 90W 0W 0W 0W 90W 0W 0W.5 (h) 0S 0S W 0W 90W 0W 0W Figure. (Continued). the reason why annual and MJJAS composites are not shown. Sea-level atmospheric pressure in +σ is characterized by a dipole in the Eastern Pacific, with a positive anomaly with its maximum at 5 S and a negative one at approximately 0 S, which show an amplitude and a slightly different extension in the case of JJA composites (Figure (g)) when compared with periods annual and MJJAS. On the other hand, for σ (Figure (h)), there are two weak anomalies, one negative, in the west of the continent, and another one positive in the east. This makes the difference between the extremes of frost occurrence in the Wet Pampas. Therefore, if high pressure prevails in the eastern Subtropical Pacific, as in +σ,the number of frosts increases due to the fact that the entrance of these systems is favored. The main synoptic types associated with frosts in the studied region correspond to anticyclone situations determining radiative and/or advective frosts. This was demonstrated in Müller et al. (00) on the basis of the climatic-synoptic classification of circulation patterns comprising the PF + GF days during the winters of the period In addition, the opposite extreme ( σ ) occurs when the positive anomaly is located over the Southern Atlantic in front of the Argentine coast, with an increase in the northerly wind (relatively warmer and more humid) in the Pampas region, which does not favor frost occurrence. For both +σ and σ, upper-level composites corresponding to the annual and the seasonal periods show similar configurations, with more intense anomalies for the quarter JJA. As shown in Figure (+σ ), a positive anomaly of the stream function in the southwest of the continent and a negative one toward the northeast determine an intensification of the zonal circulation. The maximum wind anomaly at 50 hpa is found in their interface, as clearly noticed in Figure, with a more important zonal wind anomaly in the composites of JJA. In winter σ (right panel), instead, the negative anomaly of the stream function over the southern cone of South America extending toward the Atlantic creates an anomalous anticyclone circulation in the seasonal (also in the annual composites). As a consequence, there is a deceleration in the subtropical jet intensity, which is higher at the eastern coast of the continent (Figure (d)). In the analysis of the mean monthly atmospheric circulation associated with the group PF + GF, there are some features in common with the description of the seasonal composites, although the years making up the monthly periods +σ and σ are not necessarily coincident (Table I). One of them refers to the opposite sign of the values of the anomalies between extremes. So, for the various months studied, the surface temperature and humidity fields show anomalies consistent with the studied extremes, i.e. negative (positive) in the years above (below) the corresponding mean. In the case of surface temperature, the analysis reveals that, in all months from April to October (not shown), the anomaly pattern is basically the same as in JJA (shown in Copyright 007 Royal Meteorological Society Int. J. Climatol. 7: 7 7 (007) DOI: 0.00/joc

6 7 G. V. MÜLLER 0S 0S 50W W 90W 0W 0W 0S 0S W 0W 90W 0W 0W 0S 0S 5 5 0S 0S (d) 0W 90W 0W 0W 0 0W 90W 0W 0W 0 Figure. Anomalies of stream function ( 0 m s ) in 50 hpa and vector wind and zonal component (m s ) at 50 hpa (d), for the composite of PF + GF for the period JJA +σ (left panel) and σ (right panel). Solid (dashed) lines indicate positive (negative) values. Figure ). In the month of June +σ ( σ),the entire region is affected by negative (positive) anomalies, with maximum values reaching.5 C (+. C). In the case of the humidity field, the month-per-month analysis does not show any difference with quarter JJA and therefore it is not shown. These results are consistent with flow anomalies at 50 hpa, which, for monthly composites σ (Figure, right panel), is from the northeast in June (Figure ) and from the north in July (Figure (d)). For +σ (left panel) instead, there is an atypical situation in July composite, with an anomaly from the north affecting the Wet Pampas (Figure ), while in June the flow anomaly is from the south in the same region (Figure ). This difference explains the higher negative values in the temperature anomalies in June with a maximum of.5 C as compared with anomalies in July, with a maximum of. C. August, in particular, looks very similar to the quarterly average JJA (not shown). Figure 5 shows sea-level pressure composites for the months of June, July, and August for the extreme σ (right panel). Although they exhibit differences, they are consistent with the dominant northerly wind anomaly described earlier. For +σ (left panel), a positive anomaly is located in the southwest of the continent over the Pacific Ocean in August (Figure 5(e)), while in June (Figure 5) and July (Figure 5) positive anomalies cover almost the entire Argentine territory. In particular, July shows a singularity compared to June, as the anticyclonic anomaly extends toward the east of the continent over the Atlantic Ocean. This pattern, associated with the northerly wind anomalies in the studied area (Figure ), suggests the prevalence of radiative frosts affecting the Wet Pampas during July. In contrast to the month of June, the southerly wind anomaly (Figure ) favors advective frosts. The monthly analysis of wind anomalies at 50 hpa is shown in Figure. The month of June +σ (Figure ) shows a zonal maximum leeward the Andes at the approximate latitude of the subtropical jet over South America. In July (Figure ), this anomaly appears displaced to the northeast and, in August (Figure ), it practically disappears. The remaining monthly composites +σ display weak (May and October) or moderate (September) anomalies, but always with a positive zonal component (not shown). On the other hand, the extreme σ shows frosts only for the months between May and September (Table I), and all these months (with the exception of September) have weak anomalies with a negative zonal component in the area of the subtropical Copyright 007 Royal Meteorological Society Int. J. Climatol. 7: 7 7 (007) DOI: 0.00/joc

7 PATTERNS LEADING TO EXTREME EVENTS IN ARGENTINA 79 0S 0S 0S 0S 0W 90W 0W 0W 0W 90W 0W 0W.5.5 (d) 0S 0S 0S 0S 0W 90W 0W 0W.5 0W 90W 0W 0W.5 Figure. Anomaly of vector wind at 50 hpa for the composite of PF + GF +σ (left panel) and σ (right panel), June and July (d). jet, i.e. a deceleration in the intensity of this jet. September is very similar to the extreme +σ. Annual, seasonal, and monthly composites for the group PF The composites of the years of each period for both extremes (+σ and σ ) in the groups PF + GF and PF show common characteristics for the same variable in the respective fields. This is mainly due to the fact that the years of the periods, annual, seasonal, and monthly in each group, are mostly the same for each extreme case (Tables I and II). However, the analysis of the composites of the different variables for group PF reveals some interesting differences, for example, the upper-level wind anomaly shows changes in the direction (Figures 7 ). In particular, in the region of the South American subtropical jet, the JJA composite for the group PF +σ shows an anomaly with a northwesterly component (Figure 7), while for PF + GF +σ it has a basically zonal configuration (Figure ). However, the most important difference appears at the 50 hpa level. Figure 7 shows wind anomaly for winter PF +σ, with advection from the north to the northern area of the Wet Pampas and from the south to the southern area of the same region. The mentioned northerly anomaly is absent in PF + GF +σ (Figure (e)); this last group only shows slight wind anomalies at 50 hpa all over the area under study. In spite of this, there is no important difference in the value of the negative surface temperature anomalies between both groups in the extreme +σ. This does not occur, either, with the positive temperature anomalies for the extreme σ, in spite of the fact that PF has greater northerly wind anomalies (Figure 7(d)) than those observed in PF + GF (Figure (f)). What has been said about the seasonal composites extends to monthly composites where, for instance, for the month of August, in group PF +σ, there is a pattern slightly different from that of the group PF + GF +σ.in the latter, the wind anomaly at 50 hpa is small or null, particularly in the region of the Wet Pampas (not shown). However, for PF, the anomalous flow from the north extends up to Patagonia, where another anomaly, this Copyright 007 Royal Meteorological Society Int. J. Climatol. 7: 7 7 (007) DOI: 0.00/joc

8 G. V. MÜLLER 0S 0S 0S W 0W 90W 0W 0W 0S W 0W 90W 0W 0W (d) 0S 0S W 0W 90W 0W 0W 0S 0S W 0W 90W 0W 0W (e) (f) 0S 0S 0S W 0W 90W 0W 0W 0S W 0W 90W 0W 0W Figure 5. Anomaly of sea level pressure (hpa) for the composite of PF + GF +σ (left panel) and σ (right panel), June, July (d), and August (e) (f). Solid (dashed) lines indicate positive (negative) values. time a cyclonic one, appears in the southeast of Argentina (Figure ). In this group, the surface pressure field shows a positive anomaly affecting the east of the Wet Pampas, not appearing in the composite for PF + GF +σ. On the other hand, for August PF σ, Figure shows an important anomaly from the north, affecting the entire Patagonia, which has a greater magnitude than for the group PF + GF σ, with, in addition, an anomalous advection from the north in the Pampas region (not shown). The layout of these anomalies determines the differences observed in the temperature and humidity fields, as well as the value of the anomalies that sustain a negative (positive) sign for +σ ( σ) in both studied groups. Daily composites of partial frosts The evolution of the days previous to the PF events is analyzed in order to identify the circulation patterns associated with the PF and to distinguish them from those that were previously obtained for the GF (M5). The studied period corresponds to winters with maximum frequencyof GFoccurrence (GF+σ ), taking into account a selection of PF having place during such period (according to the criterion discussed in the section Criteria, Data, and Methodology). The composites of the PF selected for the three days preceding the PF up to one day after are shown. Starting the analysis with the mass field, Figure 9 shows the average geopotential height contours at 50 and 50 hpa (left and right panels, respectively). It is shown that the southern cone of South America is under the influence of a trough during the days prior to the event (Figure 9, (f) (h)) and it moves to the east, being over the Atlantic Ocean on day 0 (Figure 9(d), (i)). The associated upper-level geopotential anomalies show a negative perturbation windward and a positive one leeward the Andes on day (Figure 0), moving to the east-northeast, as seen in the meridional wind anomalies at 50 hpa (Figure (e)). On the other Copyright 007 Royal Meteorological Society Int. J. Climatol. 7: 7 7 (007) DOI: 0.00/joc

9 PATTERNS LEADING TO EXTREME EVENTS IN ARGENTINA 0S 0S S 0S W 90W 0W 0W 0W 90W 0W 0W 0 0 0S 0S 0W 90W 0W 0W Figure. Anomalies of wind vector and zonal component (m s ) at 50 hpa for the composite of PF + GF +σ, June, July, and August. Solid (dashed) lines indicate positive (negative) values. 0 hand, at low levels, there is not a too intense cyclonic anomaly affecting the Wet Pampas region, in particular, until the day before the frost event (Figure 0(g) (h)). In its movement to the east, this anomaly advects air from the south, as noted in the meridional wind anomalies at 50 hpa (Figure (f) (j)). This flow appears as a statistically significant positive anomaly (99% level) that affects the center and north of Argentina during day (Figure (h)) and part of the Wet Pampas on day 0 (Figure (i)). On the following day, a weak component from the north, represented by the meridional wind anomaly in Figure (j), inhibits the occurrence of frosts. In brief, during the extreme winters under consideration (GF +σ ), the PFs originated mainly by cold air advection, similar to GF, but both cases differ in the origin of the system causing them. In the case of the GFs, an anticyclonic anomaly develops prior to the frost event, irrupting into the southern cone of South America and then moving toward the east-northeast causing the cold air advection as it passes (M5). Conversely, on the days before the PF, a cyclone anomaly develops over the continent (e.g. Figure ), moving to the east, being over the Atlantic Ocean on the day of the frost (e.g. Figure ), generating the irruption of air from the south with a maritime path and affecting the east of the Wet Pampas (e.g. Figure ). This pattern may be associated with the third PC (Pattern C) for the group of the La Niña years studied in the synoptic-climatic classificationbymüller et al. (00). It is interesting to note that the studied winters (i.e. GF +σ ), from where the PF have been selected, have the characteristics of La Niña years. There is a cold sea surface temperature (SST) anomaly in the central-eastern Pacific Ocean for the composites during the winters GF +σ (M5). DISCUSSION AND CONCLUSIONS On the basis of the spatial selection of frosts in the Wet Pampas during the period 9 990, we studied the Copyright 007 Royal Meteorological Society Int. J. Climatol. 7: 7 7 (007) DOI: 0.00/joc

10 G. V. MÜLLER 0S 0S 5 5 0S 0S 0W 90W 0W 0W 0W 90W 0W 0W 0 0 (d) 0S 0S 0S 0S 0W 90W 0W 0W 0W 90W 0W 0W.5.5 Figure 7. Anomalies of vector wind and zonal component (m s ) in 50 hpa and of the wind vector at 50 hpa (d), for the composite of PF for the period JJA +σ (left panel) and σ (right panel). Solid (dashed) lines indicate positive (negative) values. 0S 0S 0S 0S 0W 90W 0W 0W 0W 90W 0W 0W.5.5 Figure. Anomalies of vector wind at 50 hpa for the composite of PF, period August, +σ (left panel, ) and σ (right panel, ). Copyright 007 Royal Meteorological Society Int. J. Climatol. 7: 7 7 (007) DOI: 0.00/joc

11 PATTERNS LEADING TO EXTREME EVENTS IN ARGENTINA 0S 0S 0S 0S 0S 0S E 0E 90E 0E 50E 50W 0W 90W 0W 0W S 0S E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 0S 0S 0 0E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 0 0E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 (d) (e) E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 (f) 0S 0S E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 (g) 0S 0S E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 (h) 0S 0S E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 (i) 0S 0S E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 (j) 0S 0S E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 0 Figure 9. Mean value of geopotential height composites from day to day at 50 hpa (e) and 50 hpa (f) (j), corresponding to PF. Contours are every 00 gpm at 50 hpa and every 50 gpm at 50 hpa. characteristics of the mean atmospheric circulation for the years of extreme frequency of frosts occurrence, i.e. maximum (+σ ) and minimum ( σ ). In both studied groups, PF + GF and PF, there is a certain similarity in the years of the different periods +σ and σ, respectively. So, in a same group, the period annual consists of the same years as the period MJJAS in the respective extremes +σ and σ. At the same time, both seasonal periods (MJJAS and JJA) consist of the same years, with the exception of the year 97 in PF + GF and 9 in PF, which were added to the composites in the extreme +σ ; while in the extreme σ only group PF shows differences between some years of the seasonal periods. The above implies that, in a same group, we obtain basically similar space patterns for the variables considered (with only some differences in the magnitude of the anomalies) when comparing the composites obtained for the periods annual and seasonal. Opposite anomaly patterns appear in the comparative analysis of the composites for the different periods +σ and σ of both groups (PF + GF and PF). For both PF + GF +σ and PF +σ in low levels, there is a pressure anomaly dipole, one anticyclonic center in the north and another cyclonic in the south, Copyright 007 Royal Meteorological Society Int. J. Climatol. 7: 7 7 (007) DOI: 0.00/joc

12 G. V. MÜLLER 0S 0S E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 0S 0S 0S 0S 0S 0S 0S 0S E 0E 90E 0E 50E 50W 0W 90W 0W 0W E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 (d) 0 0 0E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 (e) 0 0E 0E 90E 0E 50E 50W 0W 90W 0W 0W S 0S 0S 0S (f) E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 (g) 0 (j) S 0S E 0E 90E 0E 50E 50W 0W 90W 0W 0W E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 (h) 0S 0S E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 (i) 0 0S 0S E 0E 90E 0E 50E 50W 0W 90W 0W 0W Figure 0. Anomaly of geopotential height composites from day to day at 50 hpa (e) and 50 hpa (f) (j), corresponding to PF. Positive (negative) contours in solid (dashed) line every 0 gpm. to the southwest of the continent, over the Pacific Ocean (this location sustains slight modifications in the monthly periods). Conversely, in group GF +σ studied by M5, an important anticyclonic surface anomaly appears, at the above-mentioned location. According to this, the configuration in GF +σ favors the entrance to the continent of anticyclones from the southwest, i.e. where the Andes are lower. This explains the low-level wind anomalies in the JJA composites for GF +σ noted by M5, where an anomaly in the north dominates the centralnortheast region of Argentina and another one in the south in the south of the country. On the other hand, for the periods with minimum frosts frequency ( σ ), the composites of PF + GF and PF show a positive sea-level pressure anomaly in the east of the continent (with the exception of some monthly periods). This favors the flow of warmer and more humid air all over the region of the Wet Pampas, which is corroborated by the northerly low-level wind anomaly. This pattern is accompanied by easterly wind anomalies at upper levels, which extend from the north of Argentina up to tropical latitudes, with a greater intensity in the groups PF + GF and PF than in GF. This means that a subtropical jet less intense than normal is associated with fewer frosts in the Wet Pampas. The upper-level wind field for composites +σ reveals a positive anomaly around 0 S, which constitutes an intensification of the subtropical jet in PF + GF and PF; i.e. the jet over South America more intense than normal is associated with a higher frost occurrence frequency in Copyright 007 Royal Meteorological Society Int. J. Climatol. 7: 7 7 (007) DOI: 0.00/joc

13 PATTERNS LEADING TO EXTREME EVENTS IN ARGENTINA 5 0S 0S 0S 0S 0S 0S 0S 0S 0S 0S 0 0E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 (d) 0 0 0E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 0 0E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 (e) 0 0E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 0 0E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 0S 0S 0S 0S 0S 0S 0S 0S 0S 0S (h) 0 0E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 (i) 0 0 0E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 (f) (j) 0 0E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 (g) 0 0E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 0 0E 0E 90E 0E 50E 50W 0W 90W 0W 0W 0 Figure. Anomaly of meridional wind composites from day to day at 50 hpa (e) and 50 hpa (f) (j), corresponding to PF. Positive (negative) contours in solid (dashed) line every m s. The shaded areas indicate regions with 99% statistical significance according to Student s t-test. the Wet Pampas in the seasonal and annual composites. However, there are also nuances in the wind anomalies at 50 hpa within the same extreme +σ, particularly when the groups PF and GF are compared. In the latter, the maximum anomaly is zonal and it is located over the Andes in the jet entrance region (M5). In PF, it has a northwest component that is located leeward the Andes, extending to the southeast up to the Atlantic Ocean. On the other hand, monthly composites show the same relationship as the seasonal in GF, i.e. a jet more (less) intensely associated with higher (lower) frosts frequency. Monthly composites for PF show a greater variation of intensity and in the position of the upper-level maximum zonal wind anomaly. Additionally, PF events are selected in the winters GF +σ and a comparative analysis is performed between daily composites of PF with those of GF (the latter discussed in M5). Results indicate significant differences between both groups. In the days before the occurrence of GF, there is a clear wave train propagating within the subtropical and polar waves guide (M5). Particularly on the day before the event occurrence, these waves arrive in phase to the west of the South American continent (Müller and Ambrizzi, 00). This characteristic does not Copyright 007 Royal Meteorological Society Int. J. Climatol. 7: 7 7 (007) DOI: 0.00/joc

14 G. V. MÜLLER 5/7/70 //79 0S 0S W 90W 0W 0W 0 0S 0S W 90W 0W 0W 0 // 0S 0S W 90W 0W 0W 0 Figure. Anomalies of geopotential height for day of PF and day 0 of PF and meridional wind for day 0 of PF at 50 hpa. Positive (negative) contours in solid (dashed) line every 0 gpm and m s. PF have been selected from Table III. appear in the daily composites of PF, where there is a wave train along the subtropical jet, less intense than the one shown for the GF. For this last case, when the waves enter the continent, they stretch meridionally and bend with a NW SW orientation, progressively moving to the northeast leeward the Andes, and following the evolution of the anticyclone that occupies a large section of the country. The anticyclogenesis causing the GF is generally associated with PFs that occur immediately before and/or after the event. However, the PF situations studied here are not associated with GF occurrences (according to the selection criterion) and they are generated by a cyclonic anomaly, which moves to the east of the continent, causing cold air advection from the south. Summarizing, the comparative analysis of the different frost events during the periods of extreme occurrence frequency gives diagnostic elements of circulation in each case. This is the base to approach a frosts forecast in one of the most productive regions in South America, the Wet Pampas, aiming at reducing its economic and social impacts. ACKNOWLEDGEMENTS The author wishes to thank CNPq for financial support (grant no. 5 09/00-0). Also acknowledged are the comments of the anonymous reviewers, which contributed in improving this work, and the revision of the English text by Dr Gustavo Carrio. This paper is part of the author s PhD thesis. REFERENCES Barrucand MG, Rusticucci MM. 00. Climatología de temperaturas extremas en la Argentina. Variablidad temporal y regional. Meteorológica : 5 0. Escobar GCJ, Bischoff SA Meteorological situations associated with significant temperature descents in Buenos Aires: an application to the daily consumption of residential natural gas. Meteorological Applications : 5 0. Escobar GCJ, Compagnucci RH, Bischoff SA. 00. Sequense patterns of 000 hpa and 500 hpa geopotential height fields associated with cold surges in Buenos Aires. Atmósfera :. Kalnay E, Kanamitsu M, Kistler R, Collins W, Deaven D, Gandin L, Iredell M, Saha S, White G, Woollen J, Zhu Y, Celliah M, Ebisuzaki W, Higgins W, Janowiak J, Mo KC, Ropelewski C, Wang J, Leetmaa A, Reynolds R, Jenne RE, Joseph D. 99. The Ncep/Ncar 0-year reanalysis project. Bulletin of the American Meteorological Society 77: 7 7. Müller GV. 00. Variabilidad Interanual de las Heladas en la Pampa Húmeda. Revista Brasilera de Meteorología : 5. Müller GV, Ambrizzi T. 00. Teleconnection patterns associated with extreme frequency of Generalized Frosts. Part I: Rossby waves propagation regions in the Austral Hemisphere. In th International Conference on Southern Hemisphere Meteorology and Oceanography. Foz do Iguazu, Brazil. Müller GV, Berri GJ. 00. Atmospheric circulation associated with persistent generalized frosts in Central-Southern South America. Monthly Weather Review (in press). Müller GV, Nuñez MN, Seluchi ME Relationship between ENSO cycles and frosts events within the Pampa Húmeda region. International Journal of Climatology 0: 9 7. Müller GV, Ambrizzi T, Núñez MN Mean atmospheric circulation leading to generalized frosts in Central Southern South America. Theoretical and Applied Climatology : 95. Copyright 007 Royal Meteorological Society Int. J. Climatol. 7: 7 7 (007) DOI: 0.00/joc

15 PATTERNS LEADING TO EXTREME EVENTS IN ARGENTINA 7 Müller GV, Compagnucci RH, Núñez MN, Salles MA. 00. Spatial analysis of Surface Circulation associated to frosts in the wet Pampas. International Journal of Climatology : 9 9. Pezza AB, Ambrizzi T Temperature Fluctuations during wintertime in South America, correlating Patagonia region with Southern Brazil (available in Portuguese). Revista Brasilera de Meteorología :. Pezza AB, Ambrizzi T Dynamical conditions and synoptic tracks associated with different types of cold surges over tropical South America. International Journal of Climatology 5: 5. Rusticucci M, Vargas W Synoptic situations related to spells of extreme temperatures over Argentina. Meteorological Applications : Copyright 007 Royal Meteorological Society Int. J. Climatol. 7: 7 7 (007) DOI: 0.00/joc