7.05 SEVERE WEATHER ANALYSIS AND FORECASTING WITH THE INTEGRATION OF LIGHTNING, RADAR AND SATELLITE INFORMATION IN OPERATIONAL CENTER IN BRAZIL

Cesar Beneti¹* Leonardo Calvetti¹ Marco Jusevicius¹ Augusto Pereira Filho² Rosangela B. Gin³ ¹SIMEPAR, Curitiba, PR, Brazil ²USP/IAG, Sao Paulo, SP, Brazil ³UNIFEI, Sao Paulo, SP, Brazil 1. INTRODUCTION Recent studies of total (intracloud and cloud-to-ground) lightning observations have shown that although equatorial Africa stands out as the most active lightning region on Earth, the most violent and extreme lightning storms are more concentrated in the subtropic to the extratropics (Goodman 2003). In South America, southern parts of Brazil and northeastern Argentina are regions particularly prone to severe weather events (strong winds, intense precipitation, tornadoes, hail, flash flood, lightning), as Gin et al (1998), Gin and Beneti (2003) and Pinto et al (1992), among others, have shown. However, up to now, there is a limited knowledge in relation to the frequency, type, morphology and intensity of cloud-to-ground lightning related to these storms, as well as the environment in which they occur. The operational suite in weather surveillance and forecasting offices has changed a lot in the past few years. Besides conventional meteorological infor-mation, remote systems such as satellite, weather radars, and lightning sensors supply real-time vital information. The primary tools for detecting thunder-storms are weather radar, lightning detectors and satellite imagery. Lightning detection systems and radars are capable of storms surveillance with shorter time intervals between observations and the integration with meteorological satellites provides meteorological surveillance for a larger area in real time. Although lightning detection systems allow for the surveillance of electrically active storms, it is not yet possible to evaluate spatial and temporal evolution of the storms quantitatively. With a Doppler weather radar, hazard storm potential (e.g. flash flood, hail, strong wind, tornado), precipitation estimation and forecasting (in conjunction with numerical weather prediction models) are already possible. Severe storms with lightning activity affect constantly Parana State, in the south of Brazil, where the experience with operational systems integration for analysis and forecasting of severe weather is presented. With this work, we expect to have a better understanding and improvement of our forecasting abilities related to those storms with strong precipitation and lightning. The integration of these systems in an operational environment such as available at SIMEPAR are the initial steps to achieve these goals. This paper presents a description of the experience with integration of lightning data and other information for severe weather monitoring and forecasting in a meteorological operational center in the south of Brazil and also analysis of some severe events registered in the area. *Corresponding author address: Cesar Beneti, SIMEPAR, Caixa Postal 19100, Curitiba, PR, Brazil, CEP 81590-110. E-mail: beneti@simepar.br

2. HYDROMETEOROLOGICAL MONITORING SYSTEMS AVAILABLE AT SIMEPAR For the analysis and forecasting of severe storms occurrence in the south of Brazi, a network of hydrometeorological observation systems, operated by SIMEPAR since 1996, is composed of a network of hydrometeorological automatic stations; a S-band Doppler weather radar; satellite reception and processing; lightning detection sensors network, computer facilities. Figure 1 presents the location of the hydrometeorological monitoring system described in this section.

Figure 1 – SIMEPAR Hydrometeorological Monitoring System in the south of Brazil operated by SIMEPAR.

2.1 Automatic Weather Stations A network of 35 automatic hydrological stations, measuring precipitation and river level, and 35 meteorological stations provides hourly observations of precipitation, pressure, temperature, humidity, wind direction, velocity and solar net radiation throughout Parana state, which are operated by SIMEPAR. A network of automatic raingauge stations from other companies are also used, with a total of more than 100 points of measurements of hydrometeorological variables. Every 3 hours, in general, data from these stations are transmitted by satellite to SIMEPAR where they are processed, archived, used for assimilation in numerical weather prediction model and integrated with other informations for weather forecasting and monitoring as well as research applications. 2.2 S-Band Doppler Weather Radar SIMEPAR operates a weather radar installed in the central region of Parana with a Doppler coverage (radius of 200km, with wind and precipitation measurements) including Curitiba metropolitan area and central and east parts of Parana State. This is a S-Band Doppler radar, EEC DWSR-95S model, with a 0.95 antenna beam width, configured to provide high spatial (4km2 area) and temporal (5 to 10 minutes) resolution information. A selection of radar products (e.g. rainfall accumulation, wind profile, nowcasting algorithms)

and also raw volumetric radar data are automatically transmitted in real-time from the radar site to the Weather Forecasting Center at SIMEPAR where they area analyzed by the meteorology staff and integrated with other observations with applications developed in-house. Some of these applications are presented in this paper. In the near future, volume data will also be incorporated in the data assimilation system for improvement of numerical weather prediction models. Radar data is already integrated with raingauge and satellite information for quantitative precipitation estimation and forecasting. 2.3 Meteorological Satellite Information A system for the reception and processing of real-time high-resolution data from meteorological satellites provides products and algorithms for several applications. Among those are fog and fire detection, precipitation estimation, and vertical profiles of temperature (TOVS), besides visible, infra-red and water vapor images which are used in weather monitoring and forecasting. 2.4 Lightning Detection System In 1998, SIMEPAR (a meteorological center), CEMIG and FURNAS (power companies) started a joint cooperation agreement which created the Brazilian Lightning Detection Network (BLDN) covering the south, southeast and center-west of Brazil handling real time lightning information (Beneti and Sato 2000) as presented in Figure 2. With this integration, data from 24 sensors are shared with three central analyzers (Curitiba-PR, Rio de Janeiro-RJ and Belo Horizonte-MG), with an increase in area coverage, lightning detection efficiency and location accuracy. Figure 2 presents the location of BLDN sensors. Although this network initially started with only three companies (SIMEPAR, CEMIG and FURNAS), there are other institutions and power companies integrating this with the perspective of more 26 sensors by the end of 2005. The lightning detection system employs a time-of-arrival (TOA) and magnetic-direction-finding (MDF) technology to detect, locate and measure CG lightning strokes. A network central analyzer ingests data from the sensors and run algorithms for lightning detection and location (Cummins et al 1998). The results computed by the central analyzer are immediately made available to the meteorologists at SIMEPAR Weather Forecasting Center through dedicated purpose computers and also archived in the system’s database. For each lightning stroke detected, its location, time of occurrence, and other characteristics area available in the system’s database. After lightning data has been archived, several products are generated for applications on storm forecasting and monitoring at SIMEPAR, real-time monitoring of transmission and distribution systems for Parana Power Company (COPEL) and other users. For the dissemination of real-time lightning data, historical analysis, lightning related severe weather alerts and integration with radar and satellite information, applications were developed at SIMEPAR and presented in Figure 3 (Vasconcellos et al 2003). 2.5 Scientific Visualization and Processing System Computational facilities at SIMEPAR include two high-performance parallel processing computers with

16 nodes each running operational numerical weather prediction models and other research applications, another high-performance computer running data base applications, and several workstations and personal computers for software development, weather monitoring, and dissemination of products and services. In order to visualize model output and other meteorological observations (satellite, radar, lightning and station data) an integrated tool was developed (Zandona et al 2005). Figure 4 presents examples of model and satellite information available for analysis.

Figure 2 – Location of the Brazilian Lightning Detection Network sensors in 2005.

Figure 3 – Lightning data and power line (above) and integration with infrared satellite image (below) using application developed at SIMEPAR for internet dissemination.

(a)

(b)

Figure 4 – Examples of numerical model output visualization (a) sea level pressure and temperature; (b) surface wind and infrared satellite information. 2.6 Weather Forecast Center Activities The wealth of meteorological data available for weather forecasters, the differing characteristics of the various observing systems, and the ability to integrate these data into a comprehensive analysis of the atmospheric state are vital points to be considered in an operational environment. A team of meteorologists in a 24/7 shift routine elaborates medium range forecast for south and southeast of Brazil, as well as short-range and very-short-range forecast warnings for several users, mainly Parana State Civil Defense, power companies, petrol refineries, civil aviation and others. In order to disseminate all the information and forecasts available in real time or as needed, we have developed some applications for the Operational Weather Forecast Center at SIMEPAR. An interactive system allows the forecaster to inherit a forecast database from the previous shift in which he/she can update or modify the bulletins. Using a database environment containing various weather elements previously identified and selected, the forecaster no longer needs to type in text for routinely scheduled forecast products (Figure 5a). A quality control system is used to constantly evaluate forecasting skills, specially for the very-short-range forecasts, due to the necessity to qualify the severe weather alerts within the users needs. All warnings and contacts are registered in this database for evaluation, monitoring and distribution through the internet, as in Figure 5b.

(a)

(b)

Figure 5 – Database interactive system for weather forecasting bulletin dissemination and evaluation. (a) Medium range forecast bulletin environment; (b) Nowcasting and very-short-range environment.

3 SEVERE STORM EVENTS Four severe mesoscale convective systems were chosen for this study: two storm events with large amount of hail, strong wind and flash flood in Curitiba metropolitan area, a pre-frontal squall line with strong wind and intensive precipitation along its displacement throughout Parana state, and another supercell storm which affected the west of Parana with severe damage for power transmission lines. For these events, data from SIMEPAR hydrometeo-rological observation systems, described in the previous section, will be presented and discussed. 3.1 Supercell Event on 06 July 2003 A pre-frontal environment in the previous day and the fast passage of the cold front during 06 July 2003 generated conditions for severe weather occurrence in Parana State. This storm caused severe damage in the Curitiba Metropolitan Area (henceforth CMA). Several car accidents in highways with one death, more than 60 thousand people in the area had power cuts for more than 6 hours. Large amount of hail in less than 30 minutes fell in Curitiba. With the hail event, flooding followed. For about 3 hours before it reach CMA, meteorologists at SIMEPAR were monitoring this storm, sending alerts and reports to

Civil Defense and power companies, regarding its strength and possible damages. Figure 6 presents lightning information from 1600UTC to 2100UTC. A total of 1776 strokes, with 96% negative polarity, were selected to be related to this storm event. From lightning information, storm movement, from west to east, was evaluated to be about 50km/h. Wind observations from the automatic weather stations network recorded wind gust in excess of 60km/h. Air temperature drop around 7ºC in Curitiba when the storm reach its boundary. Raingauge network recorded values of more than 30mm in one hour (not shown here). Radar observations from this event showed a stationary behavior during most of the storm life cycle, with a tilt in the direction of movement related to atmosphere steering level. Echo top (15dBZ threshold) above 10km was observed most of the time. Reflectivity vertical cross sections (not shown here) identified hail around 1900 and 2000UTC reaching surface. Variations of cloud top and vertical reflectivity structure were observed, however, after hail occurrence (around 1930UTC). A comparison of storm relative vertically integrated liquid (VILMAX) and lightning rate activity is shown in Figure 7. High VILMAX values are an indication of hail in the storm, which was confirmed later in the day. During this storm life cycle two peaks for VILMAX were observed. However, only 5 minutes prior to this increase in VILMAX, an increase in lightning activity was observed. Surface equivalent potential temperature obtained from the meteorological network indicated that the storm path was along the area of strong gradient region with an increase in lightning activity in its peak, as observed in other studies (Smith et al. 2000, Gilmore and Wicker 2002, Carey and Buffalo 2005, among others). The strong relation between electrical activity, storm updrafts and cloud ice water content implies that an increase in electrical activity should follow severe weather, such as hail and strong wind gust, as observed in this storm. Although not shown, vertical reflectivity profile showed that this storm was very efficient in producing and maintaining itself during its displacement in Parana state. 3.2 Squall Line Event on 09 December 2003 Although hail was not observed during this squall line event of 09 December 2003, it has interesting electrical characteristics and related weather. In Figure 9, a satellite image with lightning information related to a frontal system in the south of Brazil, around Parana region is presented. High lightning activity was observed around the continental cold front border, as one should expect. Since its relative broad area of activity and strong indication of severity, meteorologists were able to evaluate its mean velocity displacement and warn the area, basically the whole Parana State, with possible strong winds, hail, power interruption, as it was later confirmed. Radar observations, Figure 10, showed a leading line trailing stratiform mesoscale convective system with cloud echo top height (15dBZ threshold) well above 8km throughout its life cycle. Reflectivity above 45dBZ was observed in the convective region of the squall line during its movement along Parana region, with more than 8 hours of observation. This situation

is indicative of severe weather, as observe: gust wind above 50km/h with some peaks above 80km/h, precipitation rate of more than 30mm/h, strong lightning activity (Figure 11). Civil Defense reports throughout the state were issued with more than 5 thousand families with some destruction in their homes (mostly rooftops). One interesting observation of this severe storm event was the polarity reversion during part of the storm life cycle, as presented in Figure 12. Lightning activity was mostly concentrated on the convective region of the squall line, with a total of 14321 CG lightning strokes detected with the SIMEPAR LPATS. For this event, 64% of the total lightning observation was positive and 46% was negative CG strokes, although future analysis will be necessary to confirm and identify the main storms related with predominantly positive polarity cloud-to-ground lightning.

Figure 6 – Lightning information from 1600 to 2100UTC, for 06 July2003. This supercell storm reached Curitiba around 1900UTC.

Figure 7 – Temporal evolution of VILMAX and stroke rate for 06 July 2003 storm. Vertical arrow indicates the time of hail storm observation in Curitiba.

Figure 8 – Surface equivalent potential temperature and lightning strokes for 06 July 2003, indicating storm path.

Figure 9 – IR satellite and lightning information integrated for the squall line of 09 December 2003.

Figure 10 - Radar reflectivity vertical cross-section of squall line event on 09 December 2003, 0051UTC.

Figure 11 – Lightning information integrated from 2200 to 0600UTC during passage of squall line on 09 December 2003.

Figure 12 – Temporal evolution of CG stroke rate for the squall line event on 09 December 2003.

3.3 Supercell Event on 21 April 2005 Initially embedded in an environment with post-frontal (Figure 13) high humidity and large thermal instability, this storm presented a strong convective

gradient of reflectivity, with values of 45dBZ and above throughout its displacement in Curitiba Metropolitan Area (CMA). Figure 14 presents a low elevation (0.7deg) PPI for this storm shortly after its passage in Curitiba. During a time of its development there was indication of a hook echo in this storm as shown in this example. Reflectivity profiles indicated values above 45dBZ up to 5km during the most intense hours. Due to its lightning activity and radar characteristics, the forecasters were able to identify this storm with 2 hours before it reached CMA, sending out warnings to Civil Defense and Parana Power Company. Strong lightning activity followed this storm during nearly 5 hours, with more than 1500 strokes with 96% of negative polarity as in Figures 15 and 16. The stroke rate and polarity were similar to the observed in the storm described in section 3.1 of this paper. Storm relative vertically integrated liquid (VILMAX), obtained from volumetric radar obser-vations indicate the presence of hail in the storm, which was later confirmed when it moved along Curitiba area, causing great impact in the city, with flooding in main roads, trees falling, powerline disruptions and rooftops blown away. Again, tornadoes were not observed in this case.

Figure 13 – Infrared satellite and lightning information at 2045UTC, 21April 2005 event.

Figure 14 – Reflectivity data from SIMEPAR S-Band Doppler weather Radar and lightning location for 21 April 2005, 2134UTC.

Surface observations from meteorological stations network, indicated wind gust in excess of 70km/h, and temperature drop around 7C in one hour. Accumulated precipitation in those stations indicated around 30mm/h, comparable to radar observations. Similar to previous events, the predominant area of lightning occurrence was in the most intense gradient of surface equivalent potential temperature, as presented in Figure 17. This feature should be looked more carefully in the future, specially compared to cloud base height, instability and lightning activity, as well.

Figure 15 – Lightning information from 1800 to 2200UTC, for 21 April2005. This supercell storm passed near Curitiba around 2100UTC.

Figure 16 – Temporal evolution of VILMAX and lightning stroke rate for 21 April 2005 event.

Figure 17 – Storm path (lightning strokes) and surface equivalent potential temperature for 21 April 2005.

3.4 Supercell Event on 14 June 2005 This event occurred during the night, affecting towns mostly along an important power transmission

line which connects one of the most important hydro-electrical power plants in Brazil, Itaipu plant, and the metropolitan areas of Sao Paulo and Rio de Janeiro. When this storm reached the transmission line, nine towers fell down, usually built to support gust winds in excess of 200km/h. In this case, as the storm did not move near meteorological surface stations, wind gust was not registered. Synoptic conditions indicated an environment with instability, moist and warm for this time of the year, after the passage of a cold front through the area, as indicated in the infrared image and radiosonde for Iguassu Falls, Figures 18 and 19, respectively. Figure 20 presents radar and lightning information by the time the storm reaches the power transmission line. Although the radar is about 340km distant, with beam width of 0.95deg and beam elevation around 6km, reflectivities above 50dBZ were registered, indicating the severity of a supercell structure with a high reflectivity core, and consequently strong vertical circulation, possible hail and wind gusts, as later observed. A hook echo was also observed during some period of observation. During this day the meteorologists expected some storms to occur in the region due to the indication from numerical models and atmospheric conditions in the area, but all indications for such a severe event were only possible when signature from satellite IR image and lightning activity were integrated. When the radar image showed strong reflectivity in such a distance from the antenna, they were also confirming the warnings issued to Civil Defense in the area during that day.

Figure 18 – Infrared satellite and lightning information at 2045UTC, 14 June 2005 event.

Figure 19 – Iguassu Falls radiosonde profiles before (14June 1200UTC) and after (15June 0000UTC) the supercell event of 14 June 2005.

Figure 20 – PPI at 0.1deg elevation reflectivity data from the weather radar and lightning location for 14 June 2005, 2220UTC. Indication of power transmission lines in red.

Figure 21 – Storm path with cloud-to-ground lightning strokes for 14 June 2005, from 1900 to 0200UTC.

Figure 22 – Storm path with cloud-to-ground lightning strokes for 14 June 2005, from 1900 to 0200UTC.

Lightning activity was observed throughout more than 7 hours of this storm life cycle. Although other small convective systems were observed around this supercell storm, it was possible to isolate the lightning strokes related to this storm and analyze it in comparison to lightning stroke rate and polarity during its life cycle as presented in Figures 21 and 22, respectively, during all the time of observations.

Figure 22 also presents the temporal evolution of lightning stroke rate and polarity for this event, indicating some periods of predominantly positive polarity. Since the data utilized in this work is obtained simply from the real-time processing, some investigation will be done in the future to confirm this polarity distribution, as the storm occurred in the boundary of the lightning detection network. When comparing the surface distribution of equivalent potential temperature and lightning occurrence as in Figure 23, once again it is observed that the storm moved along an area of strong gradient, indicating this feature as a possible area of severe weather occurrence.

Figure 23 – Storm path with cloud-to-ground lightning strokes and surface equivalent potential temperature variation for 14 June 2005, from 1900 to 0200UTC.

4 CONCLUSIONS In the last few years, a collection of severe weather events such as tornadoes, supercells, squall lines and even a hurricane-like storm took place in the southern parts of Brazil which have caused several damage to the population and properties. Those severe storms affect constantly Parana State, in the south of Brazil, where the experience

with operational systems integration for analysis and forecasting in conjunction with professional knowledge and timely dissemination of warnings has helped the population, government, Civil Defense and industries to minimize the damage they generally provoke with terrible economic and social impacts. This paper describes the experience of lightning data analysis and severe weather monitoring and forecasting in a meteorological operational center located in a region particularly prone to hazardous weather. With a strong interest in research and development of applications for severe weather analysis and forecasting and for the operative use of lightning information, we have developed not only tools for the integration of lightning, radar, satellite, numerical models and automatic weather stations for nowcasting purposes, but also applications for management and distribution of lightning information for several operative end users, such as analysis and correlation of lightning strikes with power supply faults. The result of these developments are routinely used during nowcasting activities, specially related to preventing damages and injuries, and also disruptions in the power system of Parana companies. This paper also presents preliminary results of the analysis of electrical and hydrometeorological characteristics of severe storms, specially with hail and intense precipitation. The set of storms presented here have no indication of tornadoes. However, other cases will be investigated in the future to analyze more carefully these severe weather events. Although predominantly negative CG mesoscale convective systems are common, positive CG storms such as the leading line trailing stratiform squall line and a supercell storm presented here are related to strong severe weather events, and will be studied further. With this work, we expect to have a better understanding and improvement of our forecasting abilities related to those storms with strong precipitation and lighting. The integration of observation systems in an operational environment such as described in this paper are the initial steps to achieve these goals. 5 REFERENCES Beneti, C., L. Calvetti, M. Jusevicius, and R. B. Gin, 2004: Analysis of lightning and hydrometeorological conditions during severe thunderstorm events in the south of Brazil: preliminary results. Preprints, 1st Int. Conf. Lightning Physics and Effects. Belo Horizonte, MG, Brazil. 12-16. Beneti, C., and F. Sato, 2000: Integration of weather radar and lightning detection systems in Brazil. Preprints, 16th Intern. Lightning Detection Conf., Tucson, AZ. USA. Carey, L. D., and K. M. Buffalo, 2005: Environmental control of cloud-to-ground lightning polarity in severe storms during IHOP. Preprints, 2005 Conf. Meteor. App. Lightning Data. Amer. Meteor. Soc., USA. Cummins, K. L., Murphy M.J., Bardo E.A., Hiscox W., Pyle R.B., 1998: “A combined TOA/MDF technology upgrade of the US NLDN”, J. Geophys. Res., 103, 9035-9044. Gin, R. B., and C. Beneti, 2003: Cloud-to-ground lightning flash density of SE of Brazil in 2001. Preprints, VII Intern. Symp. Lightning Protection, Curitiba, PR, Brazil.

Gilmore, M. S., and L. J. Wicker, 2002: Influences of the local environment on supercell cloud-to-ground lightning, radar characteristics, and severe weather on 2 June 1995. Mon. Wea. Rev., 130, 2349-2372. Goodman, S. J., 2003: Atmospheric electrical activity and the prospects for improving short-term weather forecasting. Preprints, 12th Intern. Conf. Atmos. Electricity, Versailles, France. Pinto, I. R. C. A. et al, 1992: Coordinated study of a storm system over South American continent”, J. Geophys. Res., 97, 18205-18213. Smith, S. B., J. G. LaDue, and D. R. MacGorman, 2000: The relationship between cloud-to-ground lightning polarity and surface equivalent potential temperature during three tornadic outbreaks. Mon. Wea. Rev., 128, 3320-3328. Vasconcellos, C. A. M., C. Beneti, F. Sato, 2003: Fault analysis and lighting monitoring and analysis system, features and usage. Preprints, VII Int. Symp. Lightning Protection. Curitiba, PR, BR. 23-26. Zandona, C., C. Beneti, F. Sato and S. Scheer, 2005: An integrated tool for numerical weather models and environmental information visualization. Preprints, XVIII Braz. Symp. on Computer Graphics and Image Processing. Natal, RN, Brazil.