Re: confidence interval

Thank Mateus and Nicole!
Lei

On 7/2/09, Mateus Teixeira <mateus.teixeira_at_gmail.com> wrote:
> Hi Lei,
>
> Let's assume that you have two files (or one only) which contain your
> confident intervals boundaries and you read them (it) in NCL into two
> arrays.
> So, you can create two additional arrays, one for each boundary, like
>
> do i = 0, nlat-1 ; <- for latitude
> do j = 0, nlon-1 ; <- for longitude
> if( myField( i, j ) .ge. upConfInterval( i, j ) ) then
> ; <= for upper boundary of confidence interval
> myMask_upConfInterval( i, j ) = 1
> else
> myMask_upConfInterval( i, j ) = 0
> end if
> if( myField( i, j ) .le. lowConfInterval( i, j ) )
> then ; <= for lower boundary of confidence interval
> myMask_lowConfInterval( i, j ) = 1
> else
> myMask_lowConfInterval( i, j ) = 0
> end if
> end do
> end do
>
> Now, you can mask your field with this information, like
>
> maskUp = myField ; <= just to
> don't loose dims info
> maskUp = mask( myField, myMask_upConfInterval, 1 )
>
> maskLow = myField
> maskLow = mask( myField, myMask_lowConfInterval, 1 )
>
> Finally, you only have to plot using the scheme that I showed before, using
> this arrays. But, you'll have two plots, one for lower boundary and another
> for upper boundary.
>
>
> Best regards,
>
> Mateus
>
>
>
> 2009/7/2 Lei <dream916_at_gmail.com>
>
>> Hi Mateus,
>> Thank you for your suggestions. The figure provided is exactly what I
>> want to get. A few more questions about your scripts:
>> 1. are both the "mask" and "contourField" netCDF files created by
>> putting
>> 0 for below 90% confidence and 1 for above 90% confidence?
>> 2. Confidence intervals have two boundaries (up and bottom). Is it
>> possible to plot different colors for passing 90% confidence for up limit
>> and low limit?
>> Thanks,
>> Lei
>>
>>
>> On Wed, Jul 1, 2009 at 9:47 PM, Mateus Teixeira
>> <mateus.teixeira_at_gmail.com
>> > wrote:
>>
>>> Hi Lei,
>>>
>>> You can create another array in which you put 0 for below 90% confidence
>>> and 1 for above 90% confidence. After, you can plot this array as filled
>>> contours, without its contour, then plot above this field, the rainfall
>>> field.
>>>
>>> As a result, you'll have your rainfall field with shades where it is
>>> above
>>> 90% significant.
>>>
>>> See below
>>>
>>> mapa = True
>>> mapa_at_gsnFrame = False
>>> mapa_at_gsnDraw = False
>>> mapa_at_gsnAddCyclic = False
>>> mapa_at_cnFillOn = True ; c/
>>> preenchimento de cores
>>> mapa_at_cnLinesOn = False ; s/ linhas
>>> mapa_at_cnLineLabelsOn = False ; s/ rotulos nas
>>> linhas
>>> mapa_at_cnMonoFillColor = True ; uma unica cor
>>> =
>>> sombreado
>>> mapa_at_cnFillColor = "grey" ; cor do
>>> sombreado
>>> mapa_at_lbLabelBarOn = False ; s/ barra de
>>> cores
>>> plot0 = gsn_csm_contour_map_ce( wks, mask, mapa ) ; plota sombreado
>>>
>>> mapa2 = True
>>> mapa2_at_gsnFrame = False
>>> mapa2_at_gsnDraw = False
>>> mapa2_at_gsnAddCyclic = False
>>> mapa2_at_cnFillOn = False
>>> plot1 = gsn_csm_contour( wks, contourField, mapa2 )
>>>
>>> overlay( plot0, plot1 )
>>> draw( plot0 )
>>>
>>> As you can see, plot0 is the shaded field and plot1 is the contour plot.
>>> See, also, an attached figure showing the result of this plot.
>>>
>>> I hope I could help you
>>>
>>> Best regards,
>>>
>>> Mateus
>>>
>>>
>>>
>>>
>>>
>>> 2009/7/1 Lei <dream916_at_gmail.com>
>>>
>>>> Hi,
>>>> I would like to highlight (shade) the regions with precipitation
>>>> above or below 90% confidence limits. For each grid cell, the
>>>> confidence limits are different. Is there any efficient way to do
>>>> this? Thanks,
>>>> Lei
>>>> _______________________________________________
>>>> ncl-talk mailing list
>>>> List instructions, subscriber options, unsubscribe:
>>>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>>>>
>>>
>>>
>>>
>>> --
>>> Mateus da Silva Teixeira
>>>
>>> Registered Linux User #466740
>>>
>>
>>
>
>
> --
> Mateus da Silva Teixeira
>
> Registered Linux User #466740
>
_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Fri Jul 03 2009 - 07:04:57 MDT