[ncl-talk] Fwd: Sequences of values

Tue Mar 27 01:51:23 MDT 2018

please find data in this link : ERA-I_1_1_1979-2017_Tmax_Tmin_NTA.nc  

| 
| 
|  | 
ERA-I_1_1_1979-2017_Tmax_Tmin_NTA.nc

 |

 |

 |

Actually there isn't any error message. It just kills the script and shows up this message:"killed".

-----------------------------
GUIGMA 

    On Monday, 26 March 2018, 23:09:03 GMT+1, Rashed Mahmood <rashidcomsis at gmail.com> wrote:  

 Sorry I meant to send it to ncl-talk. here it is:
---------- Forwarded message ----------
From: Rashed Mahmood <rashidcomsis at gmail.com>
Date: Mon, Mar 26, 2018 at 3:02 PM
Subject: Re: [ncl-talk] Sequences of values
To: Kiswendsida Hyacinthe GUIGMA <karongseba at gmail.com>

That's too many loops! ...I can not test without data set, what error messages you get when you run this?

On Mon, Mar 26, 2018 at 1:20 PM, Kiswendsida Hyacinthe GUIGMA <karongseba at gmail.com> wrote:

Please find attached an example of a code. My data spans 0 to 35N and 20W to 50E with resolution in both directions of 1*1. It is too big to be uploaded in NCL_talk

Kindest regards

-----------------------------
GUIGMA 

    On Monday, 26 March 2018, 20:40:32 GMT+1, Rashed Mahmood <rashidcomsis at gmail.com> wrote:  

 How does your script look like? and what is the size of input data?

On Mon, Mar 26, 2018 at 8:45 AM, Kiswendsida Hyacinthe GUIGMA <karongseba at gmail.com> wrote:

Hi Rashed and everyone,Thank you for your suggestion.I also had the idea of using many loops. It works perfectly for no large data. But when it comes to apply it to a 3D data covering half of the Africa domain, it slows down the execution and ends up killing the process. That's why I was thinking that there exist an other way to do that using not as much iterations.
Best

-----------------------------
GUIGMA 

    On Saturday, 24 March 2018, 18:30:38 GMT, Rashed Mahmood <rashidcomsis at gmail.com> wrote:  

 Hi Guigma,
For 1 dimensional series you can use the following. however, you would need to adapt this to your 3D data sets through loops/ind function and need to do TESTING, this is a very time consuming process.
begin
;  X1 = (/0,1,0,0,1,1,1,1,1,0,0,1,0,1, 1,1,1,0,0,1,0/) ; orignal 
   X1 = (/0,1,0,0,1,1,1,1,1,0,0,1,0,1, 1,1,1,0,0,1,0,1,1,0,0,1,0,1,0, 1,1/) ; for testing

   X2 = new(dimsizes(X1),integer)

 do n=0,dimsizes(X1)-1
     if(n.eq.0)
       if(X1(n).eq.1.and.X1(n+1).eq. 0)
         X2(n) = 0
        else
         X2(n) = X1(n)
       end if
     end if

    if(n.gt.0.and.n.lt.dimsizes( X1)-1)
      if(X1(n).eq.1)
         if(X1(n-1).eq.0.and.X1(n+1). eq.0)
            X2(n) = 0
          else
            X2(n) = X1(n)
         end if
       else
          X2(n) = X1(n)
      end if

    end if

    if(n.eq.dimsizes(X1)-1)
      if(X1(n).eq.1.and.X1(n-1).eq. 1)
        X2(n) = X1(n)
       else
        X2(n) = 0
      end if
    end if
 end do
    print(X1+" ... "+X2)

end

Note I have NOT tested it as much one should so I would let you to test and find out if that is what you are looking for.

Cheers,
Rashed

On Sat, Mar 24, 2018 at 6:44 AM, Kiswendsida Hyacinthe GUIGMA <karongseba at gmail.com> wrote:

Thank you for your reply and for your help. Actually, before to post my question, I tried to make a use of dim_numrun_n but it didn't solve my problem.
You misunderstood a bit my [1]. I said that I successfully did that... So it is OK for that. 
For [2] , what the dim_numrun_n  function allows is just to do a count. What me I'd rather like is to set to _FillValue (or to 0 if use of where function) all values that are not in a sequence of at least 04 days. As an example I would like to have something as follows:X1=(/0,1,0,0,1,1,1,1,1,0,0,1, 0,1,1,1,1,0,0,1,0,/) ==========> X2= (/0,0,0,0,1,1,1,1,1,0,0,0,0,1, 1,1,1,0,0,0,0,/).In other words it is to replace all "isolated" 1s by 0.
Kindest regards

-----------------------------
GUIGMA 

    On Friday, 23 March 2018, 22:17:11 GMT, Dennis Shea <shea at ucar.edu> wrote:  

 Sorry, I'm I am not looking at any code.

   tcrit = 30.

   f  = addfile(...)
  T = f->T                  ; (time,lat,lon)

[1] "setting values below the threshold to _FillValue" is simple.

https://www.ncl.ucar.edu/ Document/Functions/Built-in/ where.shtml

   Tcrit = where(T.ge.tcrit, T, T at _FillValue)
   copy_VarMeta(T, Tcrit)
===========
[2] You do not need to do [1]

Maybe you could use the following function:

=========
http://www.ncl.ucar.edu/ Document/Functions/Built-in/ dim_numrun_n.shtml
***Look at the examples***
=========

  T10= where(T.ge.tcrit,1, 0)  
  T10 at long_name = "Exceed=1; No=0"
  copy_VarCoords(T, T10)
  printVarSummary(T10)
 r0 = dim_numrun_n (T10, 0, 0)  ; (ntim,nlat,mlon)
 r1 = dim_numrun_n (T10, 1, 0)  

 copy_VarCoords(T, r0)
 copy_VarCoords(T, r1)

;---Pick one grid point to see what is happening

  LAT = ..
  LON = ..
  print(T(:,{LAT},{LON})+"  "+Tcrit(:,{LAT},{LON})+"  "+T10(:,{LAT},{LON}) \
                        +"  "r0(:,{LAT},{LON})+"   "+r1(:,{LAT},{LON})    

On Fri, Mar 23, 2018 at 3:29 PM, Kiswendsida Hyacinthe GUIGMA <karongseba at gmail.com> wrote:

Dear NCL users,I have 3D (time,lat,lon) array of temperature from which I would like to keep only sequences (consecutive days) exceeding a certain threshold.  This consist concretely in setting values below the threshold to _FillValue.  The length of sequence is set to 4. So far,  have been able  to set to _FillValu only individual days that exceed the threshold. Then I would be pleased if you could provide me with a method to also set to _FillValue values that are not into sequences of at least 04 days.
Please find attached my data and my code.

Regards

-----------------------------
GUIGMA
______________________________ _________________
ncl-talk mailing list
ncl-talk at ucar.edu
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/ mailman/listinfo/ncl-talk

______________________________ _________________
ncl-talk mailing list
ncl-talk at ucar.edu
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/20180327/cf34a90f/attachment.html>