Page 45 - Contributed Paper Session (CPS) - Volume 5

CPS795 Nestor B.
From the model the length of state occurency are not significantly different, but
are slightly different inside the rainy saison. We have also detected the cities where the
number of rainy days and daily precipitation were high.
The fitting of begin, end, of rainy seasons, we noted three periods: The dry season,
the rainy season, and the intermediate period. The following figure shows the
evolution of the estimate rainy season length of tow cities

Evoulution ds durées des saisons pluvieuses de Savè Evolution des durées des saisons pluvieuses de Kandi

Années Années
Figure 3: duration of the rainy season of Savé and Kandi

We note also that, in period of good rainy season, the trend of daily priciptation
in rainy season increase, and the trend of daily precipitation in intermediate period
decrease, but we have the opposite in period of bad rainy season.

4. Conclusions
 The model allows to estimate the impact of climate change on rainfall variability,
and can be used for daily precipitation data from any country.
 From the model, we have shown that the break in the precipitation in Benin is
around the 70’s (precisely between 1964 and 1974). It should be noted that the
cities where it rains the least as kandi and Parakou were more quickly affected.
 the model through these parameters, is able to determine the cities where the
amount of water by rain and the number of rainy days are better or worse, this
can help water managers in decision-making.

Forthcoming Research
Applied the Model to data with = 3 to take account for the intermediate
period.
Set up the same model with the period (phase) variable from one rainy season to
another
Asymptotic normality and consistency of model parameter estimators.
Determination of Boostrap Confidence Interval for Non Homogeneous Hidden
Markov Models.

34 | I S I W S C 2 0 1 9