process 'storms' for the multi-site temporal model

[JavierDiezSierra]

The option process 'storms', for the multi-site model, simulates series with lower values of cross correlation for some set of hiperparameters.

The main problem resides in the fact that the scale parameter (obtained from the function xi_months) is the same for both storms in the current implementation. The function to calculate the scale parameter use the mean rainfall per gauge, but when 2 independently storms are used it is not clear which amount of rainfall assign to each one of them.

The following options underestimate the variance and skewness:

def scale_funtion_normal(x, landa, ipsilon, eta, alpha):
    return (((landa*mt.gamma(2)*ipsilon))/(x*eta))

def scale_funtion_storms(x, landa1, landa2, ipsilon1, ipsilon2, eta1, eta2, alpha1, alpha2):
    return ((((landa1*mt.gamma(2)*ipsilon1))/(eta1)) + (((landa2*mt.gamma(2)*ipsilon2))/(eta2)))*(1/x)

Another option might be obtain two different scale parameters, one per storm, and calculate each one with Rainfall_mean/2