[ncl-talk] Fwd: Sequences of values
Rashed Mahmood
rashidcomsis at gmail.com
Mon Mar 26 16:09:02 MDT 2018
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
> <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
> <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
> <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
> <http://mailman.ucar.edu/mailman/listinfo/ncl-talk>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20180326/4351a216/attachment.html>
More information about the ncl-talk
mailing list