Re: Fwd: interpolation gridded data to rcm

From: Debasish <debasish_at_nyahnyahspammersnyahnyah>
Date: Tue Feb 08 2011 - 09:46:19 MST

Dear Dennis

Thanks for your response.

> I am not clear what you mean by
> "interpolate 2d randomly-spaced data to
> another 2d randomly-spaced data or 2d randomly
> spaced data to directly RCMs grid"
>
My primary aim is to interpolate 2d random spaced data to RCM grid

For example, I have 2d random spaced data x(npts) with lat(npts) and lon(npts). I would like to interpolate the data to RCM grid i.e. lat2d(m,n) and lon2d(m,n) and interpolated output should be xint(m,n)

I believe there is two ways to solve this

[1]
direct interpolation of 2d random spaced data on to RCM grid

[2]
First interpolate 2d random spaced data to another 2d random spaced data and then convert new random space data to RCM grid

Assume, function points2points(X[*],Y[*],z[*].X1[*],Y1[*],opts:logical)

step1: lat1d = ndtooned(lat2d)
            lon1d = ndtooned(lon2d)

step2: xint1d = points2points(lon(npts),lat(npts),x(npts),lon1d(m x n), lat1d(m x n), opts)

step3: xint = onedtond(xint1d, (/yc,x/))

> ---
>
> There is a function in NCL' contributed.ncl that may work.
> Like "natgrid" this can be slow!!!!
>
> function obj_anal_ic_deprecated(X[*],Y[*],Z[*],lon:numeric,lat:numeric \
> ,dcrit[*]:numeric, opt:logical)
>
> ===
> opt = 0
> zrcm = obj_anal_ic_deprecated(loni,lati,zi, dcrit, lon2d,lat2d, opt)
>
> input: loni[*], lati[*], zi[*]
>
> output: lon2d[*][*], lat2d[*][*]
>
> ; dcrit - 1D array containing successive radii of influence.
> ; Must be expressed in degrees latitude and should be
> ; monotonically de. eg: dcrit = (/5, 3/)
>
I believe that function "obj_anal_ic_deprecated" will works for me however I have few questions regarding the function.

[1]
Do you know which version the function is created? because when I try to use it according to the webpage example (http://www.ncl.ucar.edu/Document/Functions/Contributed/obj_anal_ic_deprecated.shtml) it does not recognized the function:
fatal:Undefined identifier: (obj_anal_ic_deprecated) is undefined, can't continue
I already load the "$NCARG_ROOT/lib/ncarg/nclscripts/csm/contributed.ncl" and I am using NCL version 5.1.1

[2]
I would like to clarify that in webpage of the function output lat, lon is 1dimensional so I need to follow the method [2] (mentioned above) or can I use 2dimension lat (lat2d) and lon(lon2d) too (method[1]). If like to use lat2d(m,n) and lon2d(m,n) and corresponding output(m,n), what are the modifications?

[3]
rscan (in webpage) or dcrit (in your script) can be one dimensional array of any length or is there any particular reason to choose length 2
 or length 3 with respect to the accuracy

Thank again for your help

Debasish

On 2/7/11 9:59 AM, Debasish wrote:
>
>> Hello,
>>
>> Thanks Dennis for your suggestion. However I believe the function "triple2grid2d" places randomly-spaced data onto the nearest locations of a grid instead of interpolation while I need to do a interpolation. I also checked http://www.ncl.ucar.edu/Document/Functions/regrid.shtml but did not find any function that can interpolate randomly-spaced data to another randomly-spaced data or RCMs grid.
>>
>> Therefore, I would like to know is there any function or any way in NCL to interpolate 2d randomly-spaced data to another 2d randomly-spaced data or 2d randomly spaced data to directly RCMs grid.
>>
>> Any help in this regard will be appreciated and sorry for any inconvenience
>>
>> Thanks
>>
>> Debasish
>>
>>
>>
>>
>> On 2011-02-03, at 14:36, Dennis Shea wrote:
>>
>>> NCL does not read Excel spradsheeys.
>>> You would have to convert to ascii, the use asciiread.
>>>
>>> "natvar" is high quality but is known to be slow.
>>>
>>> Really, if you have a large number of high res\
>>> grids you need interpolated to an RCM grid,
>>> there is not magic bullet. it will take time.
>>>
>>> The following *may* be useful.
>>>
>>>
>>> http://www.ncl.ucar.edu/Document/Functions/Built-in/triple2grid2d.shtml
>>>
>>> Good luck
>>>
>>> On 2/3/11 12:02 PM, Debasish wrote:
>>>> Dear NCL Users
>>>>
>>>> I guess I am getting missing outputs because of the assumption that xin
>>>> is gridded RCM data because neither lon proceeds left to right nor lat
>>>> proceeds from south-to-north.
>>>>
>>>> I have tried to use "natgrid" assuming xi (mxn) with lat(mxn) and
>>>> lon(mxn) but the program is hang. I guess it does not work too.
>>>>
>>>> Therefore, I would like to know how do I interpolate this data (high
>>>> resolution data) to RCM grid with lat2d(172,180), lon2d(172,180).
>>>> Meanwhile I counted that there are nearly 450 RCM grid over the input
>>>> domain; so I would like to interpolate the input data to this 450 CRCM
>>>> grid.
>>>> I am sending the part (total file has 17536 lat, lon and xin) of lat and
>>>> lon of the input data for your convenience .
>>>>
>>>> Any help in this regard will be appreciated
>>>>
>>>> Thanks
>>>>
>>>> Debasish
>>>>
>>>> =
>>>>
>>>>
>>>>
>>>> Begin forwarded message:
>>>>
>>>>> *From: *Debasish<debasish@sca.uqam.ca<mailto:debasish@sca.uqam.ca>>
>>>>> *Date: *1 février 2011 16:42:48 HNE
>>>>> *To: *Ncl Talk<ncl-talk@ucar.edu<mailto:ncl-talk@ucar.edu>>
>>>>> *Subject: **[ncl-talk] interpolation gridded data to rcm*
>>>>>
>>>>> Dear NCL users,
>>>>>
>>>>> I have gridded data xin (m,n) with lat(m,n) and lon(m,n) for smaller
>>>>> domain. I would like to interpolate the data to RCM grid lat2d(a,b)
>>>>> and lon2d(a,b) for bigger domian.
>>>>> I follow the followings steps:
>>>>>
>>>>> 1. I assume "xin" as RCM gridded data
>>>>> 2. convert lat2d and lon2d to 1d array lat1d and lon1d
>>>>> 3. Use "rcm2points" function
>>>>> xout1d = rcm2points(lat,lon,xin,lat1d,lon1d,0)
>>>>> 4. then convert xout1d to 2d
>>>>> but all the values of xout1d are missing
>>>>>
>>>>> Any suggestion to fix these problem will be appreciated
>>>>>
>>>>> Thanks
>>>>>
>>>>> Debasish
>>>>> _______________________________________________
>>>>> 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

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Tue Feb 8 10:31:43 2011

This archive was generated by hypermail 2.1.8 : Fri Feb 11 2011 - 16:11:42 MST