Re: points to grid

Dennis,

Actually the data is an index of plant diversity over land.
I tried the following and got decent results:

combmod=triple2grid(lon1,lat1,combmod1,lon,lat,False)
combmod@_FillValue=-999.
do i=2,357,5
   do j=2,717,5
     val=avg(combmod(i-2:i+2,j-2:j+2))
     combmod(i-2:i+2,j-2:j+2)=val
   end do
end do
b=addfile("/Volumes/biggrape/bigmike/cru_ts_2_10.1901-2002.tmp.nc","r")
tmp=b->tmp(0,:,:)
combmod=mask(combmod,ismissing(tmp),False)

So, I used triple2grid to produce a 0.5 degree dataset but it has
regularly spaced missing values. Then I spatially smoothed
at each grid point to fill in the gaps. Then I used the CRU data to
mask
out ocean values. Not a bad result but obviously not perfect.

Mike

On Dec 3, 2008, at 3:29 PM, Dennis Shea wrote:

> Mike
>
> I am not sure what is meant be "vegetation".
> My limited experience is that this type of data
> is more categorical. Something like:
>
>
> " 1 Evergreen Needleleaf Forest\n",
> " 2 Evergreen Broadleaf Forest \n",
> " 3 Deciduous Needleleaf Forest\n",
> " 4 Deciduous Broadleaf Forest \n",
> " 5 Mixed Forest \n",
> " 6 Closed Shrublands \n",
> " 7 Open Shrublands \n",
> " 8 Woody Savannas \n",
> " 9 Savannas \n",
> "10 Grasslands \n",
> "11 Permanent Wetlands \n",
> "12 Croplands \n",
> "13 Urban and Built-up \n",
> "14 Cropland Mosaics \n",
> "15 Snow and Ice (permanent) \n",
> "16 Bare Soil and Rocks \n",
> "17 Water Bodies \n",
> "18 Tundra " ;
>
> These data are nor well suited to any general interpolation method.
> These generally assume some level of smoothness/continuous data.
> If you were to average
> [ 10 Grasslands and 12 Croplands] =????=> 11 Permanent Wetlands
>
> ====
> You could use (1) triple2grid; (2) use poisson_grid_fill, (3) mask out
> ocean areas
>
> poissong grid_fill will give you numbers NUT are the
> the correct numbers?
>
> D
>
>
> Michael Notaro wrote:
>> Thanks for the suggestion.
>> triple2grid does no interpolation, so it produces data on a regular
>> 0.5x0.5 grid with periodic missing values.
>> Kind of like (D=data value, M=missing):
>>
>> DMMDMMDMM
>> MMMMMMMMM
>> DMMDMMDMM
>>
>> Mike
>>
>> On Dec 3, 2008, at 1:28 PM, Correia, James wrote:
>>
>>> You can try triple2grid I think.
>>> jimmyc
>>>
>>>
>>> On 12/3/08 11:18 AM, "Michael Notaro" <mnotaro_at_wisc.edu> wrote:
>>>
>>>> I have 11,000 random data points of vegetation with a latitude and
>>>> longitude for each point.
>>>> I want to create a global 2D lat/lon grid (0.5x0.5 degree) with
>>>> this
>>>> data, using interpolation. Data over the ocean
>>>> should either be all 0's or missing, either way. Which approach /
>>>> function is best for this?
>>>>
>>>> veg(11000)
>>>> lat(11000)
>>>> lon(11000)
>>>>
>>>> TO
>>>>
>>>> data(360,720)
>>>> _______________________________________________
>>>> 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
>
>
> --
> ======================================================
> Dennis J. Shea tel: 303-497-1361 |
> P.O. Box 3000 fax: 303-497-1333 |
> Climate Analysis Section |
> Climate & Global Dynamics Div. |
> National Center for Atmospheric Research |
> Boulder, CO 80307 |
> USA email: shea 'at' ucar.edu |
> ======================================================
>

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Wed Dec 03 2008 - 14:58:24 MST