THE OPERATIONAL APPLICATION OF THE LIGHTNING JUMP ALGORITHM FOR HAIL NOWCASTING IN CATALONIA.

Transcription

1 THE OPERATIONAL APPLICATION OF THE LIGHTNING JUMP ALGORITHM FOR HAIL NOWCASTING IN CATALONIA. Farnell C., Rigo T., Puig O. Servei Meteorològic de Catalunya (SMC). Universitat de Barcelona (UB). Grup de Climatologia.

2 Where is Catalonia? 2

3 Introduction Definition The LJ is a sudden increase of the total number of lightning rate (Williams et al. 1999). It is a consequence of strong updraft which favours the crash between ice particles and a higher separation of the charge (Williams, 2001). This tool is used for forecasting SEVERE WEATHER Hail > 2 cm Tornados / Waterspout Downburst Strong wind gust 3

4 Introduction Examples of Severe Weather recorded in Catalonia Font: Jesús Coines Font: F,F.Teixidó Font: Font: Farnell C Font: O. Rodríguez Font: F.F: Teixidó 4

5 Before the algorithm Lightning Location System (LLS) The LLS of the SMC operates in both frequencies, LF and VHF, allowing the detection of IC and CG flashes. Few networks in Europe have this capability (full IC detection) The LLS is composed of 4 detectors, with an efficiency of > 90% over the territory 5

6 Before the algorithm Detecting two types of flashes: intracloud and cloudtocloud VHF LF 6

7 Before the algorithm Multiplicity CG IC 7

8 Conceptual Outline Convection starts Separation of electrical charge Increase updraft. Increase electrical charge. The phenomena Warning is related triggered to ordinary thunderstorms are strong rain and wind and small hail. Severe phenomena

9 Lightning Jump Algorithm Adjustment The tool is an adaptation of research algorithms developed in the USA (Schultz, Gatlin ) with some different requirements. We have not considered the threshold to distinguish between severe or ordinary thunderstorms. The original algorithm takes into account Radar Lightning Data. We only consider Lightning Data. Why? The Lightning Data allow us to execute the algorithm every minute. In contrast, we obtain Radar Data every 6 minutes. 9

10 Lightning Jump Algorithm How to detect LJ warning? For each minute, we analyze each thunderstorms We transform the flashes to points on the map 24

11 Algorithm How to detect LJ warning? The points are joined between them Its center point characterizes the thunderstorm through : Coordinates Hour Flashes number Size 25

12 El Lightning Jump: una eina de predicció de temps sever How to detect a severe storm? REQUIRIMENTS Once we have the point which represents the thunderstorms We see the thunderstorm's track during the last 12 minutes And its evolution of electrical activity To make the algorithm work, it is necessary to accomplish: Continuity in space Continuity in time during the 12 minutes t 12

13 Algorithm How to detect LJ warning? This period of 12 minutes is divided into periods of 2 minutes. Then, we calculate the average of Total Lightning (TL) of each period. 13

14 Algorithm How to detect LJ warning? The averages are subtracted sequentially between them The standard deviation of all DFRDT is calculated 14

15 Characteristics Total Lightning (TL) More information in Rigo et al presentation This tool needs the TL (intracloud (IC) cloudtoground (CG)) to work well. Lightning Nº alerts ( ) Only CG CGIC (TL) 69 5,262 CLOUD TO GROUND TOTAL LIGHTNING 15

16 Characteristics Multiplicity More information in Rigo et al presentation The number of LJ taking into account multiplicity is higher than not considering the multiplicity (only the first node of IC). Without Multiplicity 2016 With Multiplicity Type Of LJ Nº alerts (2016 ) Nº alerts (2017) Without Multiplicity With Multiplicity LJ

17 How it works operationally Application to operative work 17

18 Results Experimental study PC: Percentage of correct predictions FAR: False Alarms SKILLS SCORES PC 73% FAR 10% LEAD TIME Moment of the record in surface Average of Lead Time 30 min 1 hour of anticipation 18

19 Results Experimental study /07/05 Lead Time 90 min 19

20 Results Operative 2016 SKILLS SCORES PC: Percentage of correct predictions FAR: False Alarms Average of Lead Time 56 min PC 93% FAR 7% 20

21 Results Operative /08/16 Anticipació Lead Time 40 min 40 min 21

22 Results Operative 2017 SKILLS SCORES PC: Percentage of correct predictions FAR: False Alarms Average of Lead Time 49 min PC 81% FAR 22% More than 2 hours of anticipation The MULTIPLICITY allows to increase the Lead Time 22

23 Results Operative 2017 Anticipació Lead Time 5 90 min 5 90 min

24 Campaing Picking Up The Stone 2017 & 2018 Preparation to external level Agreement with County Councils of Lleida and Girona Agreement with l ADV Terres de Ponent Agreement with Leon Universtity (ULE). Exemple de granímetre Part de l anàlisi d un granímetre. 24

25 Campaing Picking Up The Stone 2017 & 2018 Preparation to internal level Preparation of the web meteo.cat Radio and social media advertising. Preparation on: whatsapp, Manager of photos, Social media: twitter #meteocatpedra Instagram 25

26 Campaing Picking Up The Stone 2017 The hail observations collected during 2017 have been dobled thanks to #meteocatpedra 179 observacions 26

27 Campaing Picking Up The Stone 2017 Spatial Distribution of the Observations Observations LJ Warning LJ wm LJ wom 27

28 Campaing Picking Up The Stone 2017 Monthly Distribution Observations LJ Warning LJ wm LJ wom 28

29 Campaing Picking Up The Stone 2017 Hourly Distribution Observations LJ Warning LJ wm LJ wom 29

30 HAIL TEAM SMC Thank you for your attention see you in 2020 ;)