Modelling average maximum daily temperature using r largest order statistics: An application to South African data

Area of Science:

• Environmental Science
• Climate Science
• Statistics

Background:

• Natural hazards like heat waves cause global destruction, impacting health and economies.
• Extreme heat events pose significant risks, necessitating effective disaster risk reduction strategies.

Purpose of the Study:

• To model average maximum daily temperature (AMDT) using extreme value theory for disaster risk reduction.
• To assess the frequency of extreme heat days and inform preparedness strategies for heat waves.

Main Methods:

• Applied the 'r' largest order statistics approach of extreme value theory to model AMDT from 2000-2010.
• Fitted a generalized extreme value distribution to annual maxima, using maximum likelihood estimation.
• Utilized the Entropy Difference Test (EDT) and graphical diagnostics for model selection.

Main Results:

• The Weibull distribution class was identified as appropriate for modeling AMDT in South Africa.
• Estimated extreme quantiles for specified return periods.
• Assessed the frequency of extreme heat days using South African meteorological data.

Conclusions:

• The 'r' largest order statistics approach effectively models AMDT for disaster risk reduction.
• Findings support enhanced preparedness for extreme heat events in South Africa.
• Statistical modeling of temperature extremes is crucial for climate change adaptation.