# Re: [ferret_users] error converting monthly to yearly values

```Hi,

You want the upper bound to your time axis to be a year later.
i.e. for years 1979 to 2010 you need times encompassing 1-jan-1979 to
1-jan-2011
so

DEFINE AXIS/EDGES/T0=1-jan-1900/UNITS=days tann79_10 =
DAYS1900(X[i=1979:2011:1],1,1)

should work.

Russ

On Wed, 2012-06-13 at 17:22 +1000, samir pokhrel wrote:
> Thanks a lot Ansley,
>
> I have gone through the FAQ and tried to define axis based on days1900
>
> DEFINE AXIS/EDGES/T0=1-jan-1900/UNITS=days tann79_10 =
> DAYS1900(X[i=1979:2010:1],1,1)
>
> let yr=mn[gt=tann79_10@ave]
>
> but this is giving 31 records instead of 32 records.
>
> Also I feel if I use repeat command to calculate the average of every
> alternate 12 records like
>
> repeat /name=nm/range=1:384:12 (let evp_yr = evp[l=`nm`:`nm+11`@ave];
> save /file=yr_79_10.nc/append yr)
>
> which will give me perfect result but I am not able to append it in
> time, how to do it?
>
> Samir
>
>
> On Tue, Jun 12, 2012 at 11:27 PM, Ansley Manke
> <ansley.b.manke@xxxxxxxx> wrote:
>         Hi -
>         Do the edges of the grid cells on the input monthly time axis
>         line up with the edges of the yearly axis you have defined?
>          If not, you could be seeing a discrepancy in the average from
>         that, with contributions from parts of grid cells at the start
>         and end of each year.
>
>         Try this:
>
>          yes? show axis/t=1-jan-1979:1-jan-1981 t[gt=mn]  ! shows axis
>         details for the input monthly time axis
>
>          yes? show axis/t tax  ! shows the box edges and step size for
>         the axis you defined
>
>         Do the values of TBOXLO at the start of the years match?  This
>         http://ferret.pmel.noaa.gov/Ferret/faq/regridding-a-time-series-with-ave
>
>         Ansley
>
>
>         On 6/12/2012 6:04 AM, samir pokhrel wrote:
>                 Dear Ferret users,
>
>                 I am trying to convert monthly fields to yearly fields
>                 by defining axis of 360 calender days
>
>                 define
>                 axis/t=1-jan-1979:30-dec-2010:12/unit=month/cal=360_days/edges tax
>
>                 let yr = mn[gt=tax@ave]
>
>                 This seems to be much closer than other approach,
>                 still there is a difference if i check the values by
>                 averaging every alternate 12 records or by using CDO.
>
>                 Is there any other efficient way to make it more
>                 reliable
>
>
>                 Samir Pokhrel
>
>                 Climate and Global Modelling Division
>                 Indian Institute of Tropical Meteorology
>                 NCL Post, Pashan
>                 Pune-411008
>                 INDIA
>
>
>
> --
> Samir Pokhrel
> Climate and Global Modelling Division
> Indian Institute of Tropical Meteorology
> NCL Post, Pashan
> Pune-411008
> INDIA

```