[ncl-talk] stat_dispersion

[jbuzan]

That’s what is confusing, because moment is defined in contributed.ncl.  
It is possible that there is actually nothing wrong.  The cluster was having issues.  I’ll try to run the script again.

   statx    = new ( nStat, typeof(sFillV), sFillV)
   if (.not.allFill) then
       nwork    = 0l                    ; force nwork as 'long' int
       nwork    = dimsizes(work)        ; so (say) 3*nwork will work
      ;statx(0) = avg(x)                ; mean
      ;statx(1) = stddev(x)             ; std. deviation
       moment   = dim_stat4( work )
       statx(0) = moment(0)             ; mean
       statx(1) = moment(1)             ; std. deviation

       if (nwork.ge.10) then
           qsort( work )                    ; sort into ascending order
           statx(2) = min ( work )

-Jonathan




On Dec 3, 2014, at 11:24 AM, Mary Haley <haley at ucar.edu> wrote:

> If you are getting an error "moment not found", then this sounds like you might be referencing "moment" before you calculate it.  
> 
> For example, if I wrote a script that had just one line in it:
> 
>   print(moment)
> 
> I will get this error, because "moment" hasn't been set yet:
> 
>   fatal:Variable (moment) is undefined
> 
> Is this what you're getting?  It would help if you could include the exact error message, and the code that produces this error.
> 
> --Mary
> 
> On Wed, Dec 3, 2014 at 8:57 AM, jbuzan <jbuzan at purdue.edu> wrote:
> That’s what was confusing me.  It was saying something on the order of ‘moment not found’, and when I looked at the code:
>  moment   = dim_stat4( work )
> 
> So I figure something is wrong with moment.  I commented out all instances related to moment.
> 
> -Jonathan
> 
> 
> 
> On Dec 3, 2014, at 10:49 AM, Mary Haley <haley at ucar.edu> wrote:
> 
>> Hi Jonathan,
>> 
>> When you say dim_stat4 bombed, what exactly was the error message?  This is a built-in function, so you don't need to load any NCL scripts to use it.
>> 
>> --Mary
>> 
>> 
>> On Sat, Nov 29, 2014 at 3:47 PM, jbuzan <jbuzan at purdue.edu> wrote:
>> Hi Mary,
>> 
>> Sorry for the delayed response!
>> 
>> I realized there was a two problems:
>> 1) I had an error in the original dataset.
>> 2) I figured out how to use IND to feed only non-missing values to the exceedance (the original issue).  
>> 
>> I did have a problem with dim_stat4.   When it was trying to calculate kurtosis and skewness for stat_dispersion (dim_stat4 2,3), the contributed file would blow up saying it cannot find dim_stat4.  I turned it off, but I can’t seem to figure out why dim_stat4 bombed.
>> 
>> -Jonathan
>> 
>> 
>> 
>> 
>> On Oct 9, 2014, at 6:38 PM, Mary Haley <haley at ucar.edu> wrote:
>> 
>>> Jonathan,
>>> 
>>> Did anybody ever respond to this one?  
>>> 
>>> If not, the code is quite complex, and it would be useful if you could provide a full script and data.
>>> 
>>> --Mary
>>> 
>>> 
>>> On Fri, Sep 26, 2014 at 3:30 PM, jbuzan <jbuzan at purdue.edu> wrote:
>>> I am having some issues with stat_dispersion.
>>> Perhaps my code ‘fixes’ cause the issue.  I am not sure.
>>> 
>>> I have 2 datasets:
>>> time: 1460 lat: 192, lon 288
>>> time: 2920 lat: 192, lon 288
>>> I have added a second year to the original year (both 1460 time steps).  The stat_dispersion, in theory, should have the same minimum and maximum values.
>>> 
>>> The data is partitioned by Temperature and by a difference field.
>>> Here is a snippet of the code that does the partitioning:
>>> do k = 0, dimsizes(numvar)-1
>>> ;do k = 16, 16
>>>    thresholdstart = k*5.d + 200.d
>>>    print(thresholdstart)
>>>    thresholdend = k*5.d + 205.d
>>>    print(thresholdend)
>>> ;   location = ind(temperature1d.lt.thresholdend)
>>>    location = ind(temperature1d.lt.thresholdend.and.temperature1d.ge.thresholdstart)
>>> ;   printVarSummary(location)
>>> ;   print(location)
>>>    if(.not.all(ismissing(location)))
>>>         tbin = wetbulb1d(location)
>>>         else
>>>         tbin = value
>>>         tbin at _FillValue = value
>>>    end if
>>>  ;  print(tbin)
>>> 
>>>    if(.not.all(ismissing(tbin)))
>>>         tbin at _FillValue = -999.d
>>>         delete(tbin at _FillValue)
>>>         numarray(:,k) = stat_dispersion(tbin,opt)
>>>         else
>>>         numarray(:,k) = value
>>>    end if
>>>    delete(tbin)
>>>    delete(location)
>>>   ; print(numarray)
>>> end do
>>> numarray = where(numarray.eq.-999.d,value,numarray)
>>> 
>>> replace_ieeenan (numarray, value, 0)
>>> numarray at _FillValue = value
>>> numarray at missing_value = numarray at _FillValue
>>> numarray!0 = "stat"
>>> numarray!1 = "bin"
>>> numarray&bin = numvar
>>> printVarSummary(numarray)
>>> 
>>> Here is a snippet of the code that grabs the values:
>>> box_var = numvar
>>> box_var at long_name = longnamea
>>> box_var at units = unitsa
>>> ;box_var(:,0) = field1c(22,:)
>>> box_var(:,0) = field1c(2,:)
>>> box_var(:,1) = field1c(24,:)
>>> box_var(:,2) = field1c(8,:)
>>> box_var(:,3) = field1c(25,:)
>>> box_var(:,4) = field1c(14,:)
>>> ;box_var(:,4) = field1c(26,:)
>>> 
>>> field1c of 14 and 2 are the maximum and minimums from Stat_Dispersion.
>>> I think the issue is that stat_dispersion field becomes too large, and drops something.  It might be how I characterize the missing values.
>>> 
>>> Any help would be appreciated.
>>> 
>>> -Jonathan
>>> 
>>> 
>>> _______________________________________________
>>> ncl-talk mailing list
>>> List instructions, subscriber options, unsubscribe:
>>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>> 
>> 
> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20141203/b9fb8ae0/attachment.html