Page 41 - Contributed Paper Session (CPS) - Volume 5
P. 41

CPS795 Nestor B.



                             Stochastic modeling of the impact of climate
                           change on the variability of daily rainfall in Benin
                                            Nestor BOKOSSA
                                   Univesity of ABomey-Calavi Cotonou, Benin

               Abstract
               In spite of uncertainties about their width and their variability, the impacts of
               the climatic  modifications observed as  well at the global level as  local are
               already pointed out in the countries of the sahelo-Saharan zone of Africa like
               Benin. The dryness which results from a deficit of the level of precipitations as
               well  as  the  distribution  of  precipitations  in  time  and  space  constitute  the
               climatic  attributes  which  make  this  phenomenon’s  degree  significant.  Our
               objective is to model and analyze the level of daily precipitations on the basis
               of  stochastic  model.  The  proposed  model  is  non  homogeneous  hidden
               Markov  model  with  a  special  structure  that  captures  the  semi-property  of
               hidden  semi-markov  model,  thus  the  model  allows  arbitrary  dwell-time
               distributions in the states of the Markov chain. Despite the Markov chain is
               non  homogeneous,  we  showed  that  it  is  semi-regenerative  process  and
               ergodic and identifiable in mild conditions. This model can reflect a stochastic
               seasonality of dailys precipitations, and then is able to modelize additional
               seasonal variability due to climate evolution. This model delimits both the end
               and  beginning  of  rainy  seasons,  but  also  through  these  parameters,  the
               quantity of precipitation and the regularity of the rainy season from one region
               to  another.    Apply  the model  to  data  of Kandi,  Parakou,  Natitingou,  Savé,
               Cotonou and Bohicon (six towns of Benin which have synoptic stations) we
               prouved  the  impact  of climate  change  in  daily  precipitation.  And  with  the
               likelihood ratio test, we dected the rupture on the daily precipitation data of
               each town.

               Keywords
               Markov  chain;  Hidden  Markov  model;  EM-Algorith;  Identifiability;  Mixture;
               breaking; likelihood ratio.

               1.  Introduction
                   About its latitude position (between 6 30 and 12 30 of North latitude),
               Benin  is  among  countries  in  which  the  climate  is  hot  and  wet  in  the
               intertropicale  zone.  The  drought  that  has  affected  many  African  countries
               south of the Sahara has not spared Benin, especially since the 1970s.   The
               consequences  of  this  phenomenon  are  dramatic  for  the  populations:
               reduction of water ressources and forest ressources, reduction of livestock,
               dam filling up, decrease of agricultural productivity, and brutal reduction of


                                                                   30 | I S I   W S C   2 0 1 9
   36   37   38   39   40   41   42   43   44   45   46