dear Pankaj Dey

 You can refer few research articles I have added in Research Gate for public viewing. those will be useful to you please

assuming you have good data..

1 come up with the variates you want to analyze

2 fit the the distributions

- you might need to check for any ties in the data, and if yes, then address it.. e.g. through jittering for example - as in papers by Prof Carlo de Michele & Prof. G Salvadori

3 Analyse the dependance structure of the variates

4 Select the most appropriate copula from the set of Archimedian copulas or other copulae families

5  then use the selected copula to derive or determine what you want for FFA ---  conditional distributions, joint return periods, conditional return periods, trivariate return periods etc...

-- R loaded with copula packages, provides a good starting point!!!!