Re: If Statement - conditional expression yields a missing value

Hi Mike,

What Daniel suggested should work.

The important thing about the "where" function is that the conditional argument (the first argument) must be operating on an array that is the same size as what you have on the left side of the equal sign. The second and third arguments must also either be the same size, or a scalar.

So, in Daniel's code below:

do hr=0,23
height_new(:,hr)= where(temp.lt.5,-999,height(:,hr))
end d

"temp" must be the same size as the leftmost dimension of "height_new", because that is what is being reference on the left side of the equal sign, and also in the third argument.

--Mary

On Sep 16, 2013, at 4:50 PM, Daniel Barandiaran <dbarandiaran@gmail.com> wrote:

> I have come across this problem before, it happens when the value being looked at in the boolean expression is a missing value, if I'm not mistaken. Also, I looked at the where statement that you commented out and doesn't work. I have also done something similar, but without the subscripting. I'm not an expert on using the where function, but perhaps you can try the following:
>
> do hr=0,23
> height_new(:,hr)= where(temp.lt.5,-999,height(:,hr))
> end do
>
> this would not be inside the i loop. You can do similar where statements without the subscripting for the 1-D variables including temp, outside the hr loop.
>
> doing the height_new(:,:) would make the entire array -999. which is probably not what you want. Also, the third argument in where says what gets assigned to height_new if the boolean is false. The way you have it written, temp would be assigned to height, which is also probably not what you want.
>
> Once again, I am not the expert here, but it is worth a shot at least, and of course if any of the other users out there see a mistake in what I wrote I am sure they will chime in.
>
> Good luck!
>
>
>
> On Mon, Sep 16, 2013 at 4:26 PM, Mike Pirhalla <glacierbayncl@gmail.com> wrote:
> I’m making a series of Hovmueller plots by plotting height levels in 24 hour time series by 124 days.
>
> The plot simply plots height levels, and works perfectly for what I need it to do. I also have temperature data for each corresponding height. I want filter what I am plotting to only plot heights where the temperature is greater than 5. When I attempt to set those temperature values and heights to -999 that are less than 5, I get this error: “fatal:The result of the conditional expression yields a missing value. NCL can not determine branch, see ismissing function.”
>
> Perhaps I’m going about this in an incorrect manner. The code for this portion is below, Thanks very much for any help.
>
>
>
> ; read in the data from the external file
>
> do i=0,(nrow-1)
>
> n=i
>
> date(i) = a(n,0)
>
> day(i) = a(n,1)
>
> hour(i) = a(n,2)
>
> lev(i) = a(n,3)
>
> height(i) = a(n,4)
>
> temp(i) = a(n,5)
>
> end do
>
>
>
> height_new=new((/124,24/),float)
>
>
>
> do i=0,123 ; loop over all of the days in season (124)
>
> a0=i*24 ; each day has 24 hours
>
> a23=a0+23 ; last hour of day
>
> height_new(i,0:23)=height(a0:a23) ;new variable for heights for each hour of the day
>
>
>
> ;height_new(:,:)= where(temp(i).lt.5,-999,temp(i)) ; this didn’t work
>
>
>
> if (temp(i) .lt. 5) then
>
> temp(i)=-999
>
> height_new(:,:)=-999
>
> day(i)=-999
>
> hour(i)=-999
>
> date(i)=-999
>
> lev(i)=-999
>
> end if
>
>
>
> …skipping to the end I plot:
>
> plot= gsn_csm_contour(wks,height_new(:,:),res)
>
>
>
> ---
>
> Michael Pirhalla
>
> Graduate Research Assistant
>
> University of Alaska Fairbanks
>
> Dept. of Atmospheric Sciences
>
>
> _______________________________________________
> ncl-talk mailing list
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>
>
>
>
> --
> Danny Barandiaran
> Department of Plants, Soils, and Climate
> Utah State University
>
>
> _______________________________________________
> ncl-talk mailing list
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Mon Sep 16 18:02:47 2013