I have attached a zip file with a graph bmp. that I have fitted a nonlinear model (of the form b1*sin(b2*x+b6)+b3*sin(b4*x)+b5) to. It is not picking out the very low values for feb of each year however and since the nonlinear solver in my program is confined to only 6 parameters, what can I do to the function to make it account for these dates?!

The problem is I need to be able to integrate the function afterwards so I can't just use a conditional function. It has to be continuous. I was thinking of adding a sin function which has a dip around february to it but how would one create a function like that which is effectively zero for almost every other month?!