Re: [ferret_users] FERRET v5.81: return=xaxis needs huge memory

[Ansley Manke <ansley.b.manke@xxxxxxxx>]

Hi Martin,
Thank you for reporting this, and for taking the time to make an 
example. This is a bug. What is happening is that the variable pvapour 
has a slightly different grid from the variable tatm. 
yes? show grid tatm
    GRID GIF1
name       axis              # pts   start                end
ETOPO60X  LONGITUDE          360mr   20.5E                19.5E(379.5)
ETOPO60Y  LATITUDE           180 r   89.5S                89.5N
normal    Z
TAX       TIME              3000 r   01-JAN-2000 00:00    18-MAR-2008 00:00


yes? show grid pvapour
   GRID (G002)
name       axis              # pts   start                end
ETOPO60X  LONGITUDE          360mr   20.5E                19.5E(379.5)
ETOPO60Y  LATITUDE           180 r   89.5S                89.5N
ABSTRACT  Z              9999999 r   1                     9999999
TAX       TIME              3000 r   01-JAN-2000 00:00    18-MAR-2008 00:00

In the code which decodes the string `pvapour,return=xaxis`, Ferret 
looks to see if the context is known for the variable pvapour, and if 
not, it loads the variable to get its full context.  The abstract Z axis 
of pvapour is triggering this in your example.  Of course, the context 
isn't needed just to get the name of the x axis of the variable, so we 
should be able to fix it without too much trouble.

Ansley

Martin Schmidt wrote:

> Hi all,
> I am confronted with the following:
>
> I am computing specific humidity from dew point temperature.
> Trying to cancel the modulo flag of the output data, ferret
> complains about missing memory.
>
> To demonstrate the problem, I have written some test
> procedure from public files:
>
> use etopo60
>
> define axis/t=0:3000:1/t0=1-jan-2000/unit=days tax
> define grid/x=rose[d=1]/y=rose[d=1]/t=tax tg
> let tatm = rose + 0*t[gt=tg]
>
> sp rm tatm*
> repeat/l=1:3000 save/file=tatm.nc/append tatm
> q
>
> Now restart ferret
>
> use tatm.nc
> let ahumi = 1
> let a1 = 6.1078
> let a2 = 17.2693882
> let a3 = 0.01
> let a4 = 237.29
> let pvapour = ahumi*a1*exp(a2*(tatm-a3)/(tatm+a4))
> can axis/modulo `pvapour,return=xaxis`
> **ERROR: insufficient memory: 194400000 words were requested.
>
> I only wanted to change an axis attribute .... 
> Surprisingly the following helps:
> set region/l=1
> can axis/modulo `pvapour,return=xaxis`
> can region 
> Also the following works
> let pvapour = 1*exp(2*(tatm-3)/(tatm+4))
> yes? can axis/modulo `pvapour,return=xaxis`
> !-> can axis/modulo ETOPO60X
>
> Something is wrong here, which I consider worth to be reported.
>
> Best,
> Martin Schmidt
>