Heat Wave Characterization Using Extreme Value Theory and Parametric Bootstrap

Описание: José del Carmen Jiménez Hernández
Universidad Tecnológica de la Mixteca

Heat waves that occur in a year are considered extreme events that can affect public health and economic
activities such as animal husbandry and agriculture. Because of this, it is necessary to know certain
characteristics of heat waves since, based on these, appropriate measures and decisions can be taken. In
this paper, we propose to estimate three characteristics of heat waves: the distribution of the total number
of heat waves during a year; the distribution of the intensity or duration of heat waves during a year; the
distribution of the time corresponding to the beginning of heat waves in the year. The aforementioned will
be done using the extreme value theory and parametric Bootstrap. First, using real data of daily maximum
temperature recorded during 1970-2009 at three meteorological stations in the State of Sonora, Mexico, the
distribution of daily maximum temperature is estimated using the generalized extreme value distribution.
Then, using these distributions and specifying an absolute threshold, Bootstrap is carried out to calculate
heat waves, thus estimating the characteristics of interest. It is concluded that the generalized extreme value distribution fits the actual daily maximum temperature data well and that the estimated parameters contain
information about two temporal characteristics of the daily maximum temperature. Furthermore, through the
heat wave simulation process, the characteristics of the random phenomenon of daily maximum
temperatures seem to be well represented by the estimated models.