Estudo de tempestades severas com a ocorrência de raios nas localidades de Santarém e Belém e suas consequências na sociedade (Lightning characteristics associated to severe storm cases which impacted the cities of Belem and Santarem, in Amazonia)

O objetivo desse estudo e analisar o comportamento das tempestades severas visando acompanhar os sistemas atmosfericos produtores de raios e sua relacao com a precipitacao para que seja possivel demostrar suas principais caracteristicas evolutivas, a pesquisa foi  desenvolvida atraves das analises de dois estudos de casos de tempestades severas nas cidades de Belem e Santarem .Foram utilizados dados de raios coletados pela rede STARNET e dados de chuva disponibilizados por varios sistemas de observacao de superficie. Atraves das analises desses dois estudos de caso observamos que os sistemas meteorologicos que atuam na mesma faixa de latitude nao seguem um padrao uniforme, e que as tempestades severas trouxeram grandes transtornos para as localidades. Dessa forma e preciso que haja cada vez mais politicas publicas de prevencao a essas comunidades, onde deve-se realizar um trabalho de antes, durante e depois dos episodios de tempestades severas. Esse tipo estudo mostra que pode ser possivel criar alertas de tempestades severas atraves das ocorrencias de raios, para se evitar perdas materiais como os desligamentos, transtornos para a sociedade e tambem perdas de vidas humanas e animais. A B S T R A C T The main objective of this study was to analyze the behavior of severe storms, by monitoring the evolution of atmospheric systems which produced intense rainfall and lightning occurrences, in order to draw their most significant characteristics. This research encompasses two case studies of severe storm events which impacted the cities of Belem and Santarem, in eastern Amazonia. Lightning data from the STARNET lightning detection network, as well as, rainfall data made available by several surface observation systems, were used. Even though the cases studied were restricted to similar latitudes, the storm generating systems did not present a common apparent behavior. Nevertheless, this study has demonstrated that, through the continuous monitoring of lightning occurrences, it is possible to develop severe storm warning protocols for these and other cities. This way one could avoid or diminish material and life losses, electric systems disruptions, and traffic jams caused by flooding. Keyword:. Amazon, severe storms, lightning, atmospheric systems


Introduction
The complex relationship between the forest and the climate in the Amazon Region, has been object of great scientific interest both in the global and local scales (Xavier et al 2000). During the past four decades, extreme events of flooding and droughts seem to be alternating with increased frequency in that Region. In 2005 an extreme drought which led to forest fires which disrupted the air traffic and an outbreak of respiratory diseases on the population in the state of Acre in southwest Amazonia (Marengo et al., 2008 a, b;Zeng et al., 2008, Cox et al., 2008. On the other hand, flooding has been recurrent in the Amazon Basin and in several of its cities in the same period. According to report from data collected by the Atmospheric Electricity Group of the Brazilian Space Research Institute, only in the eastern Amazon área there were 8 human deaths caused Pompeu by lightning (Almeida et al 2012;Dentel et al 2014).
In order to monitor and make a prognosis in large scale, and a forecast in small scale of those extreme events, one may use several observation tools to gather data. For instance, the similarity of the space distribution patterns of some parameters of lightning and rainfall from convective meteorological systems has been demonstrated by Roohr Vonder (1994) e Carey et al (2005) using satellite images. However, the details of the relationship between the lightning frequency of occurrences and rainfall over a given area of the Earth's surface, depend on its climate, topography, surface usage, electric conductivity of the soil and vegetation cover. Dissing and Verbyla (2003) showed that the lightning occurrence density distribution was affected by local convection and air circulations, resulting from the differential sensible heat exchanges at the top of two contrasting types of vegetation covers.
In the especial case of the Amazon Region these local contributions to the thunderstorm formation are mainly led to severe extremes when superimposed by at least one of the following large scale meteorological convection systems : The South Atlantic Convergence Zone (SACZ) \and The Intertropical Convergence Zone ( ITCZ) According to Anagnostou (2002) the presence of the SACZ over the southwestern part of the Amazon Region produces low level westerly trade winds and its absence results in easterly winds. During its annual southward drift over the eastern part of the Region, the ITCZ draws moisture from the Atlantic Ocean and produces intense convection over this sub region. Therefore, these two large scale systems control most of the seasonal behaviors of lightning and rainfall at the two cities where occurred the severe storm cases studied in this work.
In the past two decades a research group which includes the authors of this paper, developed several studies on lightning characteristics using data from two surface detection systems which cover the eastern Amazon Region (Ribeiro et al.,2014;Almeida et al.,2012;Dentel et al., 2014) and data from the Tropical Rainfall Measuring Mission (TRMM) satellite. Some environmental and local impacts associated to this atmospheric variable yielded the following results. Rocha et al. (1997a), showed the effect of the southward drift of the ITCZ front edge which produced a cloud-to-ground (CG) lightning occurrence peak in December in the city of Belem while in the city of Paragominas (~300 km south of Belem) the lightning incidence peaked in March. Regarding the surface distribution of CG lightning in this sub region, Rocha (2001) observed a lower density of events detected over coastal areas, as transition zones between ocean and land surfaces.
The relationship between lightning densities of occurrences and rainfall was also studied on a regional scale by Teixeira et al. (2008Teixeira et al. ( , 2009. Both used estimates of those two variables, from TRMM satellite data. Their spatial distribution superposition was confirmed, as well as, their association with typical regional meteorological convective systems. Naturally, the appearance of these systems present a space and time seasonal behavior, characteristic of the regional climatology (Ribeiro et al., 2011a). The lightning occurrence densities in eastern Amazonia present a large increment between December and May and with a delay of a couple of weeks one observes the regional rainy season period (Ribeiro et al., 2011b).
Some applications of the lightning versus rainfall relationship were already made by the author and co workers (Pompeu et al.,2016(Pompeu et al., ,2017, however the present work addresses the possibility to prognosticate or predict, at least a few hours in advance, the occurrence of a severe on a given locality of the eastern Amazon Region. In order to do so the authors used lightning flashes detection, rainfall, satellite images, METAR code and radio sondes ( Pompeu et al.,2011) data. The two last data sets came from meteorological stations located at the airports of Belem and Santarem. These combined data sources in different space and time scales were applied to analyze two severe storms which occurred on April 27, 2009, in Belem, and another observed on May 1 , 2009, in Santarem.
One expects that this work will contribute to the understanding of the context in which severe storms develop in this region, as well as, to indicate the best observation tools to alert the cities authorities and the local population to get ready to diminish damages and economic losses.

Methodology
Characteristics of the Areas of Study.
Considering the fact that the CG lightning density of occurrences is influenced by the type of surface beneath the cloud, an attempt was made to describe the main physiographic features surrounding the two cities were the severe storms were observed. These analyses were restricted to areas within 100 km circles centered in the cities of Belem and Santarem with geographic coordinates given below.

Characteristics of the area around Belem.
Belem is the capital of the state of Para and has its geographical center at latitude 1º27' 21" S and longitude 48º30' 15"W, on the average. its altitude is 4 m above sea level. This city is bordered by the Guama River and the Guajara Bay on the southern shore of the Amazon River, 120 km from the Atlantic Ocean ( Figure 1).
The annual rainfall totals recorded in Belem range between 2000 and 3000 mm (Figueroa and Nobre, 1990), and the rainy season is in general well defined between the months of December and May. This pluviometric regime is mainly modulated by the quasi stationary presence of the ITCZ over the city in the rainy period. The other large scale atmospheric circulation system, namely the SACZ, eventually contributes to enhance coonvection and the occurrence of severe storms (Souza, 2003).The local and regional precipitation is also significantly influenced by the interannual general circulation of the atmosphere. El Niño events in the Pacific Ocean diminish the rainfall totals and La Niña years produce positive anomalies of this variable over the area .
The circular area defined around Belem ( Figure 1) has the following surface type percentage composition: 45 % of agriculture and pasture, 20 % of low or marsh lands, 20 % of water surfaces and 15 % of native forests (Pompeu ,2012).

Characteristics of the area around Santarem
The city of Santarem has its geographic center at latitude 2º 24' 52''S and longitude 54º 42' 36''W. It is located on the left margin of the Tapajós River , at its confluence with the Amazon River. On a straight line it is 810 km westward from Belém ( Figure 2).
According to Sioufi (2005), mosto f the land área surrounding Santarem is covered by Dense Latifolia Equatorial Forest. The relief is moderate to lowlands with pasture áreas.
The rainy season in Santarem is delayed with regard to that in Belem by approximately one month. It usually starts in January and ends in May. The local dry period is between July and December. The annual rainfall totals around 2000 mm in Santarem. This value is significantly smaller; and the seasons are not so well defined as those of Belem (Nechet et al., 2006). The clouds formation around this city are less dependent of the inland reach of the ITCZ. Instead, they are largely influenced by local circulations of river and terrestrial (forest) breezes. (Fitzjarrald,. et al.,2008). The daily cycles of these breezes was described by Moura et al. (2004). Pompeu (2012) analyzed the surface types within the circular área shown in Figure 2 and determined the following cover percentages: 60% forests, 13% agriculture and pastures and 26% river surfaces.

Rainfall Data
Rainfall measurements were performed by automatic tipping bucket rain gauges pluviometers, located at each one of the area centers. In Belem the rainfall was measured at the 4 th Naval District meteorological station and in Santarem at the station belonging to the National Institute of Meteorology (INMET).
The severe storm selection criterion used in this work was drawn from the Brazilian Airforce -Aeronautics Meteorology Stormy days must include rainfall intensities which produce accumulations equal or higher than 25 mm in one hour or 40 mm in two hours. However, severe rainfall events are defined as those which produce rainfall equal or beyond 50 mm in one hour.

Lightning Flashes data
The lightning data analyzed in this work was collected by the Sferics Tracking and Ranging Network (STARNET), a detection system manufactured by the Resolution Display Inc (RDI) and operated in Brazil by Geophysics and Astronomy Institute of the University of São Paulo ( IAG/USP).
Daily register spreadsheets with lightning data obtained at 15 minute intervals were processed within a MATLAB environment, through a software called LEZEUS, developed by one of the authors of this paper. The studied areas surrounding each one of the two cities were defined within 100 km radii centered at the above mentioned pluviometer positions. In order to establish those limits the Arc View 3;2 software was used through its "Theme" tool , option "create buffers" . The procedure "Open theme table" exhibited one table with the number of lightning occurrences identified within the circle considered during the time period of observations.

Results and discussions
Case study 1 : Belem on April 27, 2009.
The severe storm observed on the abovementioned date, in Belem, was initiated by a sudden growth of the number of lightning occurrences at about 11:00 LT (Local Time). Figure 3 shows that in the two hourly intervals after that time, the STARNET detected 101 and 146 lightning events, respectively, within the area surrounding Belem. The rainfall measured at the center of the area, between 13:00 and 14:00 LT, was 39.4 mm, followed by 12.0 mm in the next hour. Therefore, the 51.4 mm rainfall total in two hours satisfies the severity criterion for this storm. Considering all hours of that day, the rainfall total was of 60.2 mm. Figure 4 shows infrared images obtained by the GOES 12 satellite on April 27, 2009. These images exhibit a pattern of evolution and enhancement of the nebulosity over the area of Belem (indicated by red circles) between 15:00 and 18: UTC. The meteorological systems configuration indicate the presence of a large scale band of deep convection over equatorial latitudes, northward from Belem, characterizing the action of the ITCZ on that day. The possible interaction of thr ITCZ with a Line of Instability (LI) visible southward may have led to the severe weather occurrence in Belem. The resulting lightning storm presented its highest activity in Belem, between 15:00 and 16:00 UTC or 12:00 and 13:00 LT.  By means of the ArcView software it was possible to define the positions of the lightning flashes within the area around Belem, during the day of this case study. The hourly evolution of the electric discharges incidence points were coded in colors and displayed in Figure 5. This Figure shows that the electric storm was more intense towards the northwest sector of the circle, over areas away from the urban part of Belem. Nevertheless the rainfall measured at its central position was impressive and caused flooding as shown in Figure 6 and all the other disruptions associated to such events. Besides the losses of vehicles, furniture in flooded houses, the risk of diseases by contaminated water, fallen trees and collapse of the electric energy distribution impacted the city for some hours. A detailed description of the damages and losses caused by this two hour storm was reported on the April 28, 2009 edition of the local newspaper "O LIBERAL".  On Figure 7, one may observe that the electric discharges activity was very intense on both April 29 and May 1 , 2009, with respectively 749 and 745 daily lightning flashes registered by the STARNET system within this target área. The persistence of the lightning occurrences throughout a three day period indicates that the area of study was influenced by a large scale deep convection atmospheric system. This Figure indicates a relatively small rainfall total on April 29 considering the high number of lightning occurrences on that date. One possibility to explain this fact is that the more intense rainfall may have happened away from the área center, where the rain guge was operating.
Although the electric activity was considerably diminished on April 30, the measured rainfall increased to reach its peak on the following day. As a matter of fact, April 30 is part of a three day severe storm which impacted Santarem. The moderate lightning occurrence number observed by the STARNET on this day , was mostly detected after 21:00 hours Local Time (Figure 8 A ) and reached 256 and 279 events at 22 and 23 HLT, respectively. The METAR Code data collected at the airport in Santarem was used (Figure 8 B)  of the corresponding hourly rainfalland other meteorological variables. The rainfall evolution was color coded in that Figure. It is interesting to point out that the hours of máxima thunder activity indicated by the METAR data coincided with the hours of higher lightnin numbers detected by the STARNET. However at about 22 and 23 HLT the METAR indicated only light rain occurrence. According to METAR,Intense rainfall only started at around 00:00 HLT of May 1, reaching a maximum at 00:40 HLT.At this particular time the METAR informed a SPCI situation which means an especialy heavy precipitation event. This severe rainstorm lasted until 05:00 of that day. This kind of time lag between lightning and rain storms is recurrently observed through similar methodology (Ribeiro et al, 2014) and may help on short time warnings of severe rainstorms. Next, the most intense storm in this case study will be analyzed. The thunderstorm activity on May 1, 2009 was significant between 00:00 and 02:00 UTC, as na extention of the evening storm observed on the previous day ( Figure 8B). The genesis of this storm may be found as a result of the interaction between a meso escale convective system centered northward from Santarem as indicated by the high density of lightning activity in Figure 9 B and the ITCZ latitudinal band evident on the satellite images of Figure 10. These images also show that the cloud tops were very cold within the circled área superimposed to our area of interest. This corresponds to a deep convective system which produced the lightning and rainfall severe storm observed. This stmospheric remained active at least up to 06:00 UTC or 03:00 HLT.
The type of meso scale atmospheric system which streghtened the ITCZ over Santarem during this period was not identified in this case. However, previous studies may help one to generalize the comon aspects of these two case studies. First the choice of two cities at similar latitudes propiciated that both could be influenced simultaneously by the presence of the ITCZ. There is reason to believe that the meso scale system which produced this sequence of events in Belem and subsequently in Santarem was a Instability (or Squall) Line which appeared near the Atlantic coast on April 27 and propagated towards the Continent and reached Santarem two days later. It may be interesting to point out that, the local characteristics of the daily convective pulses also played a role determining the hours of the most intense rainfall rates. In Belem these máxima occurred in the afternoon and in Santarem, after midnight and before dawn. These distinct characteristic between the two cities were described by Moura et al., (2004) and Fitzjarrald et al., (2008), noting that in Belem the convection follows the daily solar heating pulse, while in Santarem the Tapajos and Amazon River breezes delay the transport of moisture toward land to the evening hours.  The severe storms analyzed in this work happened near the end of the rainy season in eastern Amazonia, which starts in mid December and goes up to the end of May. 2009 was a particularly rainy year, and at the time of these storms the Tapajos River was almost 9 meters above its average level. After the storm the river overflew its banks and canoes were seen on its front streets. Port and commerce activities came to a halt and more than 130,000 people had losses or were displaced from their homes (Figure 11).

Conclusions
The analyzes of these two case studies may lead to the following conclusions: The two selected áreas of study were impacted by severe lightning and rainfall events in whuch the storm in Santarem followed the event in Belem after about three days. This seems to indicate that a meso scale atmospheric system, probably a Squall Line, drifted from east to west, coupling with the ITCZ which was more or less stationary over the latitude band common to the two targeted áreas. The events happened towards the final period of the rainy season in eastern Amazonia. The hours of the extreme intensity of the rainstorms were determined by local characteristics of the daily pulse of solar heating and water to land breezes usual circulations at each area. Besides these meteorological features driving the formation of deep convection systems in both cases, it wasobserved in general that na intense lightning storm preceded the rainfall peak intensities by one to three hours. This is in part due to the applied methodology of observing the flashes over a large área and registering rainfall at the center of the área. However this phenomenom persists even when the lightning detection is restricted to a small radius of observation. This may be an additional short term warning method of iminent severe rainstorm.
Although the conclusions of this study may not be generalized directly, its methodology which uses remote sensing of lightning discharges, satellite infrared images, raingauges and METAR coded data from surface meteorlogical stations, may be employed at low cost elsewhere, to generate useful warnings to the local Civil Defense and other public services, to take measures to protect the population interests. Of course to do so, it is necessary to have a qualifyed group of especialized people with real time access to the related information. CAPES for a Fellowship granted to the leading author and finally to the CNPq for a research grant to the REMAM Project.