An Extreme Value Approach to Modeling Risk of Extreme Rainfall in Bangladesh

Detta är en Kandidat-uppsats från Lunds universitet/Matematisk statistik

Författare: Emmy Andersson; Evelina Nilsson; [2018]

Nyckelord: Mathematics and Statistics;

Sammanfattning: The extreme value theory has been applied on daily rainfall in the five most exposed areas of Bangladesh between the years 1980-2016 in order to esti- mate extreme rainfalls for the next 10, 50 and 100 years. These types of computations are necessary for optimising planning and preparations for ex- treme future rainfalls which can lead to minimising property damage and ultimately saving lives. Generalised extreme value distribution is fitted to annual maxima accord- ing to the block maxima method. In addition, the generalised Pareto distri- bution is fitted to the daily rainfall according to the Peaks-Over-Threshold method. The different parameters were estimated with a 95 % confidence interval both through the delta- and the profile likelihood methods. There- after, the return period is computed according to each model using the same estimated confidence intervals.

  HÄR KAN DU HÄMTA UPPSATSEN I FULLTEXT. (följ länken till nästa sida)