Re: Ratio between two fields with some null values

Dennis, Maria,

I usually do something like this...

ncl > a = (/10, 0, 6, 0, 0, 3/)*1.0
ncl > b = (/ 2, 3, 1, 0, 6, 0/)*1.0
ncl > b@_FillValue = 1e20
ncl > epsilon = 1.0e-4
ncl > b = where(b.lt.epsilon,b@_FillValue,b)
ncl > c = a/b

...since taking the ratio of values which are not quite zero but almost
is usually not very meaningful and gives nasty plot results. An example
is precipitation, where the reference (b) might give very little precip
(i.e. no precip) but the experiment (a) actually gives a considerable
amount. The ratio would simply be some huge (and meaningless?) number
which I don't want to plot.

Cheers,
Oli
 
________________________________________

Oliver Fuhrer
Numerical Models

Federal Departement of Home Affairs FDHA
Federal Office of Meteorology and Climatology MeteoSwiss

Kraehbuehlstrasse 58, P.O. Box 514, CH-8044 Zurich, Switzerland

Tel. +41 44 256 93 59
Fax +41 44 256 92 78
oliver.fuhrer@meteoswiss.ch
www.meteoswiss.ch - First-hand information
  
 

> -----Original Message-----
> From: Dennis Shea [mailto:shea@ucar.edu]
> Sent: Dienstag, 13. Juli 2010 17:35
> To: M.G.Frontoso@leeds.ac.uk
> Cc: Fuhrer Oliver; ncl-talk@ucar.edu
> Subject: Re: Ratio between two fields with some null values
>
> The "where" statemen is quite useful but not for the case.
>
> ncl > a = (/10, 0, 6, 0, 0, 3/)*1.0
> ncl > b = (/ 2, 3, 1, 0, 6, 0/)*1.0
> ncl > c = a/b
> fatal:divide: Division by 0, Can't continue
> fatal:Div: operator failed, can't continue
> fatal:Execute: Error occurred at or near line 3
>
> ncl > q = where(b.ne.0.0, a/b, 1e20)
> fatal:divide: Division by 0, Can't continue
> fatal:Div: operator failed, can't continue
> fatal:Execute: Error occurred at or near line 13
>
>
> ncl > b@_FillValue = 0.0
> ncl > c = a/b
>
> ncl 6> print(c)
>
> Variable: c
> Type: float
> Total Size: 24 bytes
> 6 values
> Number of Dimensions: 1
> Dimensions and sizes: [6]
> Coordinates:
> Number Of Attributes: 1
> _FillValue : 0
> (0) 5
> (1) 0
> (2) 6
> (3) 0
> (4) 0
> (5) 0
>
> ===
> I am not sure this is what you want.
>
> For example c(1) = a(1)/b(1)/ = 0.0/3.0 = 0.0
>
> or, do you want
>
> ncl > b@_FillValue = 1e20
> ncl > d = a/b
> ncl > print(d)
>
> Variable: d
> Type: float
> Total Size: 24 bytes
> 6 values
> Number of Dimensions: 1
> Dimensions and sizes: [6]
> Coordinates:
> Number Of Attributes: 1
> _FillValue : 1e+20
> (0) 5
> (1) 0
> (2) 6
> (3) 1e+20
> (4) 0
> (5) 1e+20
>
>
>
>
> On 7/13/10 6:29 AM, Oliver.Fuhrer@meteoswiss.ch wrote:
> > Maria,
> >
> > The NCL where statement (
> > http://www.ncl.ucar.edu/Document/Functions/Built-in/where.shtml )
> > usually does the trick!
> >
> > Cheers,
> > Oli
> >
> >
> > ________________________________________
> >
> > Oliver Fuhrer
> > Numerical Models
> >
> > Federal Departement of Home Affairs FDHA
> > Federal Office of Meteorology and Climatology MeteoSwiss
> >
> > Kraehbuehlstrasse 58, P.O. Box 514, CH-8044 Zurich, Switzerland
> >
> > Tel. +41 44 256 93 59
> > Fax +41 44 256 92 78
> > oliver.fuhrer@meteoswiss.ch
> > www.meteoswiss.ch - First-hand information
> >
> >
> >
> >> -----Original Message-----
> >> From: ncl-talk-bounces@ucar.edu
> >> [mailto:ncl-talk-bounces@ucar.edu] On Behalf Of Maria Frontoso
> >> Sent: Dienstag, 13. Juli 2010 14:04
> >> To: ncl-talk@ucar.edu
> >> Subject: Ratio between two fields with some null values
> >>
> >> Dear all,
> >> I would like to plot the variable ratio between two fields A
> >> and B, but some of these fields have sometimes a *null*
> >> value, so I get the following error message
> >>
> >> *********************************************
> >> fatal:divide: Division by 0, Can't continue
> >> fatal:Div: operator failed, can't continue
> >> *********************************************
> >>
> >> Since I don't mind to have an array with *missing* values
> >> (i.e. the values coming from the division by zero), I would
> >> like to ask you if there is a way to by-pass that and plot
> >> the ratio in any cases, regardless by the values of the two
> >> initial fields A and B.
> >>
> >>
> >> Thanks,
> >> Maria
> >> _______________________________________________
> >> 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
>
_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Tue Jul 13 09:47:12 2010