To get your calculation to complete without error you need only regrid your 1 year (365.2425 day long) climatology back onto the Gregorian calendar. Simply replace the definition
Note that @LIN is the default ... shown here just to emphasize that linear interpolation is used.
Here's the background. A daily climatology is not a clean problem no matter how you slice it. I suppose your intent for a "daily climatology of 366 days" might be that the final result contains a Feb. 29 date, which is computed from 1/4 as many input points as all other days. Ferret will not readily calculate things this way (though I imagine it could be forced to do so with sufficient torture applied ;-) ). For better or worse, Ferret's regridding algorithm between calendar types doesn't regard the proclamations of Pope Gregory as scientifically based. (The logic being that in a leap year the earth does not actually take an additional day to revolve around the sun.) Instead they regard each individual year (1961, 1962, ...) as representing the same time interval, regardless of how the calendar may divide it into days. I offer this just as background. (It can play havoc with daily (24 hour) climatologies!)
Here is a script to illustrate. (Note: For this email list it is helpful to pose questions and solutions that do not depend on access to particular files.)
define axis/t=1-jan-1961:31-dec-1990:1/unit=days tlong
On 4/15/2011 3:07 AM, Peter Szabo wrote: