The most obvious suggestion would be a sum of normal distributions, one for each of the five peaks. But pinning down means and variances of five different Gaussians seems unpleasant without some kind of theoretical justification.
– SemiclassicalJan 19 '21 at 21:15