\noindent {\large \textbf{\sf Fit 3:}}
\begin{verbatim}
> (Fit3=fbvpot(cbind(cluster.peaks,cluster.peaks2),c(1.2,1.3),model="log"))

Estimates
  scale1    shape1    scale2    shape2       dep  
 0.11134  -0.08502   0.10141  -0.03622   0.22080  

Standard Errors
 scale1   shape1   scale2   shape2      dep  
0.22361  0.14318  0.26121  0.15698  0.04865  
\end{verbatim}

Here a bivariate extreme value distribution has been used to model the dependence between extreme sea surges at these two locations.  Spatial dependence is clearly an issue here, as the estimate of the logistic dependence parameter is 0.22; incorporating this dependence into our analysis could improve return level inference.\textrm{\textcolor{blue}{\bf [2 marks]}}

We are told in the question that there is a rainy season (November--May).  Thus, there could be issue of seasonal variability that we have not taken into account.  One solution might be to restrict attention to data in the rainy season; another might be to allow parameters in the model to vary by season, either smoothly or on a seasonal-piecewise fashion.  \textrm{\textcolor{blue}{\bf [2 marks]}}