DATE2016-05-31 18:46:19
IDABSTRACT20160531184619-1072
CONTACTn.kamperakis@gmail.com
PRESENTATIONORAL-PARALLEL
INVITED0
IDSESSION1
TITLEA DATA MINING APPROACH TO PREDICT WATERSPOUTS OVER IONIAN SEA (WEST GREECE)
AUTHORSNikolaos Kamperakis (1,2), Ioannis T. Matsangouras (1,2), Panagiotis T. Nastos (1), Ioannis Pytharoulis (3)
AFFILIATIONS
  1. Laboratory Of Climatology And Atmospheric Environment, Faculty Of Geology And Geoenvironment, University Of Athens Athens (Greece)
  2. Hellenic National Meteorological Service Athens (Greece)
  3. Department Of Meteorology And Climatology, School Of Geology, Aristotle University Of Thessaloniki Thessaloniki (Greece)
ABSTRACTExtreme weather phenomena have been considered of high concern by the scientific community so that to mitigate the impacts and contribute to the adaptation and resilience of the society. Tornadoes and waterspouts have been characterized as the most violent of all small-scale natural phenomena as they are associated with extremely high winds, inside and around the funnel, causing extended damage and in many cases loss of life. Frequent areas of waterspout formation are identified over specific areas along west parts of Greece. The goal of this study is to examine the forecast ability of waterspout occurrence by implementing a data mining method on a database of tornadic activity. There is a need to initiate a forecast task regarding waterspout development over these frequent areas along western Greece. In order to support this task the Laboratory of Climatology and Atmospheric Environment has launched an online reporting system (tornado.geol.uoa.gr) to record any tornadic activity over Greece. The database concerns waterspout activity that occurred during the last 16 years (2000-2015). Concerning the data mining method the Waikato Environment for Knowledge Analysis (WEKA) was used, which is a collection of machine learning algorithms developed by the University of Waikato, New Zealand, under a GNU General Public License. Data mining, an interdisciplinary subfield of computer science, adopts algorithms and data analysis tools, involves methods at the intersection of artificial intelligence, machine learning, statistics, and database systems to discover patterns and relationships in data that may be used to make valid predictions. Significant research has been carried out investigating the use of diagnostic variable sets or instability indices, as forecasting tools or parameters to identify favorable atmospheric conditions of severe convective weather. Atmospheric variables as wind speed/direction and temperature at specific isobaric levels of middle/low troposphere were taken into consideration as inputs in data mining process. In addition, the sea surface temperature and several thermodynamic indices were also used as variable regarding air-sea interaction and unstable weather conditions during waterspout days, respectively. The ERA-Interim reanalysis dataset was used as source of the atmospheric variables to the data mining method.
PAGE6
STATE1