Artificial Intelligence Applied to Lightning Incidence Warning in the Eastern Amazon
The Amazon region presents high lightning incidence rates, observed by satellite and ground detection systems.Most of the time the lightning incidence is related to severe storms.This work presents a methodology to predict lightning in the eastern Amazon region using Sferics Timing and Ranging Network-STARNET data and atmospheric sounding data obtained by polar orbit satellites.The intelligent tool used to predict the occurrence or not of lightning was a feed-forward artificial neural network (ANN).The ANN was used to predict the occurrence (1) and non-occurrence (0) of lightning.We have tested the methodology to forecast lightning within 5 hours.The artificial neural network was trained with the Levenberg-Marquardt algorithm.In order to train and test of the ANN were used eight study regions, located in the northeastern of Pará state in Brazil.The results showed that the ANN was able to recognize patterns of severe thunderstorms and predict lightning in advance.
Artificial neural network atmospheric sounding polar orbit satellite lightning forecast
Elton Rafael Alves Ad(o)nis Ferreira Raiol Leal Márcio.N.G.Lopes José Pissolato Filho
UNIFESSPA Marabá, Brazil UFPA Belém, Brazil CENSIPAM Belém, Brazil UNICAMP Campinas, Brazil
国际会议
4th International Lightning Protection Symposium (ILPS2018第四届雷电防护国际研讨会)
深圳
英文
91-97
2018-10-24（万方平台首次上网日期，不代表论文的发表时间）