Severe Weather Events over Southeastern Brazil during the 2016 Dry Season

Amanda Rehbein, Lı́via Márcia Mosso Dutra, Tercio Ambrizzi , Rosmeri Porfı́rio da Rocha, Michelle Simões Reboita, Gyrlene Aparecida Mendes da Silva , Luiz Felippe Gozzo, Ana Carolina Nóbile Tomaziello, José Leandro Pereira Silveira Campos, Victor Raul Chavez Mayta, Natália Machado Crespo, Paola Gimenes Bueno , Vannia Jaqueline Aliaga Nestares, Laı́s Tabosa Machado, Eduardo Marcos De Jesus, Luana Albertani Pampuch, Maria de Souza Custódio, and Camila Bertoletti Carpenedo


Introduction
Climate and weather components that affect directly the population and economy of Southeastern Brazil [1,2] have been widely studied in recent years, as well as the large-scale forcings from tropical and extratropical origins.It is well known that the climate of Southeastern Brazil is influenced by the South America monsoon system, where during the summer (Dec-Jan-Feb), there is a predominance of intense convective precipitation due to the availability of plentiful heat and moisture over the tropical region [3]. is intense convective precipitation delineates a cloud corridor known as the South Atlantic Convergence Zone (SACZ), which extends from the southwest Amazon Basin and through Southeast Brazil, reaching the Atlantic Ocean [4].From the beginning of autumn until midspring, the frequency of SACZ episodes decreases, initiating the dry period over Southeastern Brazil [5].Moreover, it is during this period that the South Atlantic Subtropical Anticyclone (SASA) reaches its most westerly position, extending over Southeastern Brazil, which impedes the passage of frontal systems [6].
erefore, during this period, the precipitation events are normally quick, isolated, and not intense.
During the dry season of 2016, there were 9 consecutive atypical days (May 30 to June 07, 2016) with thunderstorms, tornadoes, and at least one microburst over Southeastern Brazil.
ese phenomena caused oods, smashed houses, personal injuries, and two documented deaths (http:// g1.globo.com/sao-paulo/sorocaba-jundiai/noticia/2016/06/meteorologistas-analisam-se-tornado-causou-destruicaoem-jarinu.html and http://www.saoroquenoticias.com.br/noticia.asp?idnoticia 16053).ese atypical weather events a ected mainly the southern part of Southeastern Brazil, with the most severe conditions occurring over the cities of Campinas, Jarinu, and São Roque, which are close to city of São Paulo in São Paulo State.In Campinas, a probable microburst occurred on June 05 between 00:00 and 00:30 Local Time (LT; Rachel Ifanger Albrecht, personal communication, 2016).At Jarinu on June 05 at about 21 LT and at São Roque on June 06 in the late afternoon, the occurrences of tornadoes were con rmed by analysis of damage by local meteorological institutes and civil defense, besides being observed in the meteorological radar data (Rachel Ifanger Albrecht, personal communication, 2018).Precipitation anomalies from May 30 to June 07 reached values of around 200 mm in Southeastern Brazil, in a region comprising São Paulo State and parts of other surrounding states (Figure 1).For instance, at the meteorological station of the Institute of Astronomy, Geophysics and Atmospheric Science of the University of Sao Paulo (IAG/USP), located in the southern part of the city of São Paulo, the climatological precipitation for the period of 1981 to 2010 is 55.5 mm for the entire month of June.In the rst 7 days of June 2016, at this station, the total precipitation was 175.4 mm (316% of the climatological value for the entire month).
e aims of the present study are (a) to investigate the dynamic forcings associated with those severe weather and extreme rainfall events over Southeastern Brazil in the dry season of 2016 and (b) to verify whether or not the mostused forecast models in Brazil predicted this period of intense precipitation.e datasets and methodology used are described in Section 2; Section 3 presents the synoptic discussion, low-frequency analysis, and the model forecasts results; and the concluding remarks are given in Section 4.

Data and Methods
e period of analysis is May 30 to June 07, 2016, and the region of interest covers the area between latitudes 25 °S to 19 °S and longitudes 53 °W to 42 °W (red box in Figure 1), where the most intense precipitation and severe weather events were registered.

Synoptic and ermodynamic
Analysis.e synoptic elds were constructed using data from the ERA-Interim reanalysis [7] from the European Centre for Medium-Range Weather Forecasts (ECMWF).ese data are available every six hours (0000, 0006, 1200, and 1800 UTC) with spatial resolution of 0.75 °, for various pressure levels [7].We analyzed the synoptic elds at low, middle, and upper levels at each available time; however, for brevity, only the 1200 UTC elds are presented here.Infrared satellite images (about 10.7 µm) with 4 km and 30 minutes of spatial and temporal resolution, respectively, are from the Geostationary Operational Environmental Satellite (GOES-13; Janowiak et al. [8]) and were made available by the CPC/NCEP/NWS (Climate Prediction Center/National Centers for Environmental Prediction/National Weather Service) via ftp://ftp.cpc.ncep.noaa.gov/precip/global_full_res_IR/.
Five thermodynamic indices (Convective Available Potential Energy-CAPE, Convective Inhibition-CIN, K index, Total Totals-TT, and Showalter) are used to characterize the environmental instability.Moreover, two kinematic indices (sweat and vertical shear of horizontal wind-here termed "wind shear") are also presented because when their values are strong, the environment is favorable to severe weather events [9,10] and to the formation of stronger convective supercells [11].e instability and kinematic indices were obtained for a point (23   2 Advances in Meteorology sites-Campinas, Jarinu, and São Roque. is point is located less than 60 km from these sites.In the literature, a distance of up to 180 km is used for the representativeness of such surveys [12,13].e Convective Available Potential Energy (CAPE) and Convective Inhibition (CIN) were obtained from the Global Forecast System (GFS) model analysis with spatial resolution of 0.5 °and available for 0000, 0600, 1200, and 1800 UTC. e K index [14], Total Totals (TT; Miller [15]), Showalter [16], wind shear, and Sweat index were calculated using the GFS analysis data.e Sweat index is adapted from Miller [15] to Southern Hemisphere wind conditions following Nascimento [10].

Climate Analysis.
e weather and climate in South America are influenced by relatively well-known teleconnection patterns of tropical and extratropical origins that we can observe and measure through indices and statistical analysis.In this study, we investigated the influence of the most important atmospheric and oceanic phenomena that can affect the weather over the Southeastern Brazil: Madden Julian Oscillation [17], Pacific South America pattern, first and second modes [18,19]; Indian Ocean Dipole (IOD; [20][21][22]); Southern Annular Mode [23]; and blocking events [24,25].
e MJO is triggered in the Indian and Pacific Oceans and propagates eastward over the tropical region with a cycle of about 30 to 60 days [26].During its propagation, it comprises regions with enhanced and suppressed convection.In São Paulo, the most favorable conditions for convection occur with suppression of convection over Indonesia, when the MJO is in its phases 8 and 1, as shown by Jones and Carvalho [17].Here, to better understand the influence on the extreme rainfall variability over Southeastern Brazil by the eastward-propagating MJO-related large-scale convective and circulation envelope, we have constructed lagged/lead composites for the 0.21 sigma-level (approximately 200 hPa) velocity potential and outgoing longwave radiation (OLR) anomalies.
e velocity potential was obtained from National Centers for Environmental Prediction/National Center for Atmospheric Research (NCEP/NCAR; Kalnay et al. [27]) and the OLR field from the High Resolution Infrared Radiation Sounder (HIRS; Lee et al. [28]).Daily anomalies of OLR and velocity potential were calculated at every grid point by subtracting the long-term average  in order to remove the seasonal cycle.e intraseasonal signals are isolated from the OLR daily anomalies by applying Lanczos bandpass filter [29] using cutoff frequencies at 20 and 96 days.To assemble the composites, we considered the Wheeler and Hendon [30] real-time multivariate MJO (RMM) index for our period of analysis.
is index is available at the Centre for Australian Weather and Climate Research website (see: http://www.bom.gov.au/climate/mjo/) and is based on a pair of empirical orthogonal functions (EOFs) of the combined fields of near-equator averaged 850 hPa zonal wind, 200 hPa zonal wind, and satelliteobserved outgoing longwave radiation (OLR) data [30].
e evolution of these anomalies from "day −12" to "day +9", where "day 0" represents the active phase (enhanced convection) over tropical South America, is shown in Figure 8. RMM amplitude in phase 8 reaches its maximum value at "day 0," which means that the association between rainfall anomalies and MJO passage over Southeastern Brazil was strong.
e PSA modes are teleconnection patterns extending poleward and eastward over the Pacific Ocean [31], modulating the circulation and precipitation anomalies over South America [32].PSA teleconnection patterns consist of two distinct modes: PSA 1, related to the El Niño Southern Oscillation (ENSO; Karoly [31]) and PSA 2, associated with the MJO during the winter [18].Both of them have impacts on the climate of South America, and consequently on the rainfall intensity and distribution over São Paulo state.e PSA modes are defined as the first and second leading rotated principal component modes of the 200 hPa stream function anomaly, respectively [18,32]; these patterns are also presented in other time scales such as pentads and annual [32][33][34].In this study, both PSA modes were computed using ERA-Interim reanalysis for the 200 hPa pentad stream function anomaly data.e covariance matrix was obtained through the extraction of the annual cycle computed with the climatology of 1981-2010 as a basis period.
Saji et al. [35] showed that the anomalous warming of the tropical Indian Ocean due to low level evaporation can lead to divergence in the upper troposphere, sourcing Rossby wave trains propagating from the Indo-Pacific region towards the South Atlantic Ocean in an arch-like trajectory.Taschetto and Ambrizzi [22] showed that anomalous warming throughout the Indian Ocean Basin can excite Rossby wave trains moving towards the South Atlantic, and also amplifying El Niño patterns in the precipitation over the South American continent, for the austral autumn season (March-May).In order to explore the effects of the Indian Ocean on South American precipitation, the Indian Ocean Dipole (IOD; Saji et al. [36]; Webster et al. [37]), that is, the difference between the Eastern and Western Basin sea surface temperature anomaly (SSTa), is computed through the extraction of the annual cycle based on the 1981-2010 climatology, for 36 years (1980-2016) of ERA-Interim data.
e SAM, also known as Antarctic Oscillation (AAO), is the main mode of extratropical circulation variability in the Southern Hemisphere.It consists of zonally symmetric structures, with geopotential height perturbations of opposing signs in Antarctica and in the surrounding zonal ring centered near 45 °latitude [38].Reboita et al. [23] observed that during negative SAM phases, the cyclone trajectories are northward of their positions during the positive phase, and in the South America and South Atlantic sectors, there is intense frontogenetic activity and a positive precipitation anomaly over southeastern South America, which influences the weather in São Paulo.To monitor SAM, we used the daily AAO index available on the Climate Prediction Center/National Oceanic and Atmospheric Administration (CPC/NOAA) website (http://www.cpc.ncep.noaa.gov/products/precip/CWlink/daily_ao_index/aao/aao.shtml).
is index is constructed using 700 hPa geopotential height anomalies projected onto the leading EOF mode [39].To define the phase of the SAM, we use a methodology similar Atmospheric blocking episodes are due to quasistationary planetary waves of large amplitude [40], persisting from days to a few weeks, leading to episodes of prolonged extreme weather conditions over some areas.Over the Southeastern Paci c, Southern Atlantic and Oceania, the lowpressure anomalies occurring on the equatorial ank of the blocking pattern favor the development of transient systems that may cause precipitation as they move eastward (Mendes et al. [41]).e resulting impacts on temperature and precipitation are most frequently observed over Southern Brazil, but they can also in uence our region of interest (Southeastern Brazil; Mendes et al. [41]).In the latter case, Mendes et al. [41] observed that southeastern Paci c blocking has higher impact on precipitation in austral summer and spring (wet season), while the Atlantic blocking a ects precipitation in austral autumn and winter (dry season).
For the identi cation of blocking events over the Southern Hemisphere, we used the objective method of Tibaldi et al. [42], modi ed from Lejeñas [43]. is method was adapted to a smaller horizontal spacing of the ERA-Interim reanalysis (1.5 °× 1.5 °of horizontal resolution) instead of 3.75 °× 3.75 °used by Tibaldi et al. [42] and strati ed into ve bands of latitudes, according to Oliveira et al. [25].For an episode to be characterized as a blocking event, it must persist for at least 3 days [25,44,45].
e ENSO signal was not evaluated in this work because the São Paulo (SP) region is located in between the two sectors of South America in El Niño (EN), and La Niña (LN) episodes usually a ect the observed precipitation with opposing contributions.For instance, during EN conditions, there is increased precipitation over the southeastern sector of South America (including Southern Brazil) and reduced precipitation over the northern/northeastern sector of South America (including northern/northeastern Brazil; Grimm and Ambrizzi [46]; da Rocha et al. [47]).Given its location, the SP region is considered to be a transition region where the e ect of ENSO could be either to increase or reduce precipitation [48].

Synoptic Analysis.
A rainfall anomaly averaging 47 mm occurred over the SP region (red box in Figure 1) during the period May 30 to June 07, 2016.Figure 1 shows that in speci c regions, rainfall anomalies reached more than 100 mm over these 9 days.e satellite images show convective systems forming in the western SP region and moving eastward throughout their life cycle (see, e.g., Figure 2).In addition, some convective systems were generated northwest of the SP region propagating along the low-level mean ow and growing as they moved into the region.Each system had its own lifetime, starting, and developing preferentially during early afternoon (1200 to 1500 LT). Figure 1 also shows that the predominant wind at 850 hPa owed from the southern Amazon Basin into the SP region.is is the same as the con guration observed by Morales et al. [49] on days with thunderstorms in the city of São Paulo.
Figures 3(a)-3(d) show vertical pro les of horizontal temperature advection, divergence of the horizontal wind, pressure vertical velocity (omega), and moisture convergence averaged over the SP region, from May 30 to June 07, 2016, every 6 hours.Overall, warm-air advection occurred at all levels of the troposphere, peaking on June 05 (Figure 3(a)) when two of the most severe events were reported (microburst at Campinas, SP, and tornado at Jarinu, SP). is warm advection contributes to increased instability over the SP region and is associated with a northerly mean ow, as further discussed in this section.On June 07, the last day of the observed anomalous precipitation over the SP region, intense cold advection occurred in the lower troposphere (Figure 3 e lower-level moisture convergence and warm-air advection over the SP region occurred in association with the South American low-level jet (SALLJ) (east of the Andes), which comes from tropical latitudes over the south Amazon Basin towards the subtropics, exiting over the SP region.is SALLJ developed around 0600 UTC on May 30 (not shown) and was sustained until 0000 UTC on 07 June.It can be seen as a northwesterly band of maximum wind intensity at 850 hPa in association with a poleward transport of warm and moist air (Figure 4).Commonly, days with thunderstorms over the city of São Paulo (about 100 km away from where the severe events occurred) are accompanied by strong northerly winds [49].
e SALLJs are observed throughout the year but are more frequent and intense during the warm season (NDJF) when the northeast trade winds in the equatorial western Atlantic ow towards the Amazon Basin [50].
e trade  Advances in Meteorology winds are deflected toward the southeast as they approach the mountain barriers and then converge with the flow from the western branch of the South Atlantic Subtropical High (SASH), producing strong wind speeds at low levels and convective development over the exit region of the jet (see, conceptual model presented in Marengo et al. [51] and their Figure 1). is exit region is located typically over Southern Brazil-Northern Argentina as described in Marengo et al. [51].During May 30 to June 06, 2016, however, the SALLJ was active northeastward of its typical position, such that its exit was located over the SP region instead of Southern Brazil (Figure 4). is displacement in the jet direction was likely driven by the intense extratropical cyclonic activity over the Atlantic Ocean adjacent to the South American coast during the days analyzed (Figure 5). is activity consisted of two main extratropical cyclones plus three secondary cyclones forming and acting along the southeastern coast of South America.e cyclonic (clockwise) circulation predominant over the southwestern Atlantic Oceanfavored the blocking-by the northwesterly SALLJ-of the advection of cold and dry air from higher latitudes into eastern Argentina and Southern Brazil (Figure 4).At 0000 UTC on June 07, a cold front finally penetrates the continent (figure not shown), reaching the southern Amazon region and weakening the SALLJ, by means of advection from a colder and dryer air mass into the rear of the frontal zone (Figure 4(i)).
ese persistent low-level patterns (Figures 4 and 5) all seem to be unrelated to a persistent subtropical jet at high-levels (Figure 6).
During the period of interest, atmospheric instability was calculated in terms of various thermodynamic (Figures 7(a)-7(e)) and kinematic (Figure 7(f) and 7(g)) indices.e K index indicates high probability (above 80%) of storm occurrence from 1200 UTC on May 31 to 0000 UTC on June 07 (except at 1200 UTC on June 02 and 03; Figure 7(a)).After this period, the K index decreased substantially.
e Total Totals index was high (above 46 °C) during most of our period of interest (Figure 7(b)), indicating some scattered storms.From 0000 UTC on June 04 to 1200 UTC on June 06, the warm air advection at middle levels was greater than on other days (Figure 3(a)), decreasing the values of the Total Totals during this period.e Showalter index indicated the possibility of storms, remaining most of the time below 1 °C (Figure 7(c)).However, it did not match the thresholds for tornados (below −6 °C) in the days in which this phenomenon happened (June 05 and 06).
is is possibly due to the strong midlevel warm advection over the SP region during these days (Figure 3(a)), which contributed to raising the temperature at 500 hPa.It is interesting to note that the CAPE was relatively low (Figure 7(d)), CIN was relatively intense (Figure 7(e)), and the wind shear was above the traditional threshold to favor rotating supercells (15 m•s −1 ; Figure 7(f) [10]).e severe weather observed during these days is the characteristic of the cold season, where there is usually low thermodynamic convective potential but strong wind shear (an example of such a type of event in the United States is presented in Markowski and Straka [52]).e wind shear was high even after the period of interest because of the influence of the jet streams on those days (Figure (6)).e Sweat index increased from May 30 to June 6; however, it did not reach the thresholds for the development of severe storms in the United States (above 300; Figure 7(g)).Other studies in Brazil have also found that some thermodynamic and kinematic indices are indicative of the instability of a given region; however, they may not reach established thresholds of severity even when severe weather occurs [53,54].

Low-Frequency Analysis.
e following analysis and discussions cover the contribution of climate indices (Table 1) and planetary-scale influence on the severe rain event studied in the present paper.

3.2.1.
e Influence of the Madden-Julian Oscillation.Anomalous upper-level velocity divergence (represented by negative velocity potential anomalies-dashed lines in Figure 8-had been established over tropical South America by "day −9" (May 28) and persisted to "day +3" (June 09), favoring upward motion mainly between "day −3" (June 03) and "day 0" (June 06), the period of the most intense rainfall events over the state of Sao Paulo.us, the OLR pattern showed enhanced convection over southeastern South America and the adjacent Atlantic Ocean toward the southern Amazon basin, resembling the austral winter situation with the active MJO phase over South America (e.g., [55]).In addition, "day 0" was when the RMM reached phase 8 and its amplitude began to increase.As pointed out in previous studies (e.g., Jones and Carvalho 2012 [17,56,57]), phase 8 and phase 1 favor convection over tropical South America.Even though phase 1 favors convection over tropical South America and the following days were in that phase, it was not sufficient to favor convection over the SP region because, as shown by the synoptic analysis, and on June 07, the atmospheric environment over this region began to stabilize, after the cold front passage, with cold and dry air being advected from the south.
Upper-level tropical convergence (represented by positive velocity potential anomalies (Figure 8)) progresses eastward from the western tropical Pacific from "day −12" (May 25) and arrives in tropical South America by "day +9" (June 15), indicating the end of the active MJO period over tropical South America.It is noteworthy that the evolution of the MJO influence (onset-peak-demise) has a period of approximately 10 days (from "day −6" to "day +3"), average duration of the MJO passage over South America [58].

Pacific South America Mode.
e dominant mode of low-frequency climate variability for the last pentad of May and first pentad of June 2016 was the PSA 2 mode, presenting negative values below 2 standard deviations (Table 1); this possibly results from a combination of ENSO influence in the interannual band [31] and tropical convection associated with the MJO in the intraseasonal band [18].
is strongly negative value of PSA 2 during late May/early June 2016 likely contributed to the persistent cyclogenetic activity over the Atlantic near the South Advances in Meteorology    10 Advances in Meteorology American coast that in uenced the weather in the SP region during the period analyzed.e PSA 1 mode in June 2016 was not signi cant, staying between ±1 standard deviation (Table 1).

Indian Ocean Dipole and Wave Source Analysis.
For May/June 2016, the IOD presents negative values, con guring a negative dipole event (western basin colder and eastern basin warmer).rough the analysis of the wave activity ux divergence (Figure 9) for the May-June 2016 basic state [59], a wave like pattern coming from the Indian Ocean eastern basin (Indo-Paci c region) is not found; thus, this region is probably not a Rossby wave train source in uencing the South America-even though it is found to be an upper-level divergence source (see the red shaded area over the eastern basin).

Southern Annular Mode.
e AAO index, which measures the phase of the SAM, was positive during the period of study, with an average of +1.759 (Table 1), indicating the predominance of negative anomalies of geopotential height at southern high latitudes and positive anomalies in the middle latitudes [38,60,61].Reboita et al. [23] showed that the low-pressure circumpolar belt is shifted south during the positive phase of the AAO in relation to that in the negative phase, which is unfavorable for the propagation of cyclonic systems to the north that could propagate to Southeastern Brazil.erefore, the AAO did not interfere in the analyzed extreme event.

Blocking Events.
No blocking events a ecting the weather over South America were found during the period evaluated ( gure not shown).

Conclusions
A sequence of successive convective systems favoring atypical precipitation events, with high volumes of rain, occurrences of tornadoes, and a microburst over Southeastern Brazil took place during the dry season of 2016, more specifically, from May 30 to June 07, 2016 (9 days).ese anomalous events caused flooding, damages to houses and buildings, and shortages of electricity and water in several places, with many injuries and two deaths documented.
ese severe weather events were associated with a daily sequence of convective systems that formed preferentially during early afternoon in the western part of SP state (red box in Figure 1) and also northwest of the SP region, propagating along the low-level mean flow and growing as they moved into the region.e convective systems were embedded in an instable environment (high K index, high Total Totals, low Showalter, and high wind shear) with an intense and persistent South American low-level jet (SALLJ) (east of the Andes) advecting heat and moisture from the Amazon Basin into Southeastern Brazil.e exit region of the SALLJ was located over the SP region instead of Southern Brazil-northern Argentina (which is its typical exit location according to climatological studies).
is displacement along the direction of the jet was likely driven by a sequence of extratropical cyclones that formed to the south of the region of interest over the Southwest Atlantic Ocean during the 9-day rainy period.e SALLJ was identified as the main mechanism contributing to the observed precipitation; when this jet weakened (in association with a colder and drier air mass advected over its exit region), the convective activity decreased over the SP region, characterizing the end of the atypical precipitation events.e effects of the local dynamic and thermodynamic processes mentioned in the previous paragraph were intensified by the MJO in its phase 8 and by the intense negative PSA 2 mode.
ere was no evidence of a constructive or destructive contribution from the PSA 1, from the IOD or from atmospheric blocking.Moreover, the AAO index was positive (+1.759) during the 9-day rainy period, which would disfavor cyclone propagation to the north; thus, the AAO was not a constructive factor in the analyzed extreme event.
Given the above analysis, one might ask if numerical models were able to predict this anomalous precipitation event.A preliminary analysis considering four different numerical models indicate that they all failed to forecast the beginning of the precipitation period.However, they successfully forecasted its end (Figure 10).Overall, the Global Forecasting System (GFS) was the best model over the 9-day rainy period, followed by the Brazilian Regional Atmospheric Modeling System (BRAMS), the Eta Model, and the Weather Research and Forecasting Model (WRF).e 24-hour forecasts of the WRF were the worst among all forecasts, underestimating both the area and intensity (not shown) of precipitation.Among the possible reasons why these models did not present good performance is that they might have not represented the main dynamic forcings (e.g., SALLJ), leading to this anomalous rainy event.Results of a detailed verification will be presented and discussed in a future paper.
Atypical precipitation events like the one analyzed in the present paper can happen again in the future, causing further significant impacts for society and the economy.Studies to detect and evaluate the mechanisms contributing to these anomalous events are important for the improvement of the forecasts and mitigation of the associated consequences.Suggestions for future studies include investigating why the models did not predict the beginning of the precipitation period and whether this type of severe weather will be frequent in the coming warmer climate.

Figure 1 :
Figure 1: Precipitation anomalies (mm) and 850 hPa wind composite (m•s −1 ) for May 30 to June 07, 2016.e blue (brown) colors indicate the positive (negative) rain anomalies, and the red box indicates the area of study.e anomalies were calculated using the 30-year period, 1981 to 2010.
Advances in Meteorologyto Reboita et al. [23], in which values above (below) one standard deviation indicate the positive (negative) phase.e standard deviation value of the daily SAM time series from 1979 to 2015 is equal to 1.4, and thus values between ±1.4 indicate the neutral phase.

Figure 9 :
Figure 9: Wave activity ux (×10 6 ) for May-Jun 2016.e arrows represent the wave ux W x and W y , and the shading represents the wave ux divergence, wave ux divergence indicates Rossby wave sources and convergence indicates Rossby wave sinks.Following Takaya and Nakamura [59], the wave ux is parallel to the phase velocity of the Rossby wave packages.

Figure 10 :
Figure 10: Temporal evolution of the percentage of the SP region area with precipitation above the threshold of 5 mm•day −1 (%) for CHIRPS and for di erent forecast times of the models (a) BRAMS, (b) WRF, (c) Eta, and (d) GFS, from 29 May to 9 June 2016.Dotted lines indicate the start (30 May 2016) and end (07 June 2016) of the period of interest.

Table 1 :
Values of the climate indices analyzed.σ is the standard deviation.