Re: Regridding and Plotting Satellite Data

In general, you always want to unpack a short value.
Note the scale_factor and add_offset. These are netCDF
convention attributes, specifically COARDS and CF.

See: http://www.ncl.ucar.edu/Document/Functions/Contributed/short2flt.shtml

The _FillValue should pertain to the short variable. It will
be carried over to type float in this case.

       aor = short2flt( a->Optical_Depth_Land_And_Ocean )
       printVarSummary( aor )
       printMinMax( aor, True)

Walter R Sessions wrote:
> Thanks for the quick response,
>
> Plotting the raw data values hasn't been a problem; the lat2d/lon2d
> worked perfectly for that.
>
> As for the gridding, I've attached a copy of the printVarSummary for
> aodr. It does contain the _FillValue attribute straight from the file,
> but it is -9999. Should that be -99 since the variable is of type
> 'short'? Does cssgrid have problems with short?
>
> Also pertaining to the variable type, the poisson function wanted a
> float or double variable.
>
> Walter
>
> printVarSummary(aodr) [raw data]
> **************************************************************************
> Variable: aodr
> Type: short
> Total Size: 54810 bytes
> 27405 values
> Number of Dimensions: 2
> Dimensions and sizes: [Cell_Along_Swath_mod04 | 203] x
> [Cell_Across_Swath_mod04 | 135]
> Coordinates:
> Number Of Attributes: 11
> _FillValue : -9999
> valid_range : ( -100, 5000 )
> long_name : AOT at 0.55 micron for both ocean (best) and land
> (corrected) with best quality data(Quality flag=3)
> units : None
> scale_factor : 0.001000000047497451
> add_offset : 0
> Parameter_Type : Output
> Cell_Along_Swath_Sampling : ( 5, 2025, 10 )
> Cell_Across_Swath_Sampling : ( 5, 1345, 10 )
> Geolocation_Pointer : Internal geolocation arrays
> hdf_name : Optical_Depth_Land_And_Ocean
> ************************************************************************
> On Sat, 2008-04-19 at 08:24 -0600, Dennis Shea wrote:
>
>> Hello
>>
>> [1]
>> The following is one example of directly plotting 'swath' data
>> http://www.ncl.ucar.edu/Applications/eosdis.shtml
>> See Examples 4 and 5
>>
>> [2]
>> NCL has two reserved attribute names "lat2d" and "lon2d".
>> These are used when the data are 'curvilinear'. Basically,
>> whenever the data have 2D lat/lon arrays, the variable being
>> plotted should have the lat2d/lon2d arrays associated with it
>>
>> lat2 = f->Latitude
>> lon2 = f->Longitude
>> aodr_at_lat2d = lat2 ; associate [assign] the 2d lat/lon
>> aodr_at_lon2d = lon2 ; to the variable to be plotted
>>
>> [3]
>> The cssgrid is supposed to ignore missing values [_FillValue].
>> Since this is a NASA generated hdf file, I'd be surprised
>> if they automatically created the _FillValue attribute. Typically,
>> NASA/HDF are blissfully ignorant of *netCDF* conventions.
>> WHat does the 'printVarSummary' say. Is there an _FillValue
>> attribute? If not, you can manually assign
>> aodr@_FillValue = -9999
>>
>> [4]
>> Just for 'fun' try
>> poisson_grid_fill( aodr, False, 1, 1500, 1e-2, 0.6, 0)
>>
>> plot the aodr in the swath region
>>
>> Good luck
>>
>> Walter R Sessions wrote:
>>
>>> Hello,
>>>
>>> I'm attempting to regrid MODIS satellite data for a correlation study
>>> and have been having difficulty getting it worked out.
>>>
>>> Since it is satellite data and there will be missing values, I followed
>>> the regridding flow chart and have been trying to use the cssgrid
>>> function. However, I've been running into the following problems.
>>>
>>> First, unless I drop the _FillValues, it appears to interpolate to them
>>> (in this case, -9999). The values of the contoured scalar are wholly
>>> positive, however the contour of the cssgrid derived plot is goes down
>>> to the -7e+06 and up into the positive millions.
>>>
>>> Second, I've been attempting to just plot a section of a swath in the
>>> North Pacific, which is all this file contains. The points where it
>>> does put data appear to be stretching from the appropriate area, south
>>> of Alaska, down past the equator.
>>>
>>> I've attached the script for reference.
>>>
>>> It occurs to me that it might be better to attempt to just use an
>>> average for the value within a defined grid spacing, but I've been
>>> having difficulty in finding how to do that within NCL.
>>>
>>> Advice on either how to average into a grid or to work with cssgrid is
>>> appreciated.
>>>
>>> Walter
>>>
>>>
>>> I've attempted several variations of resources on and off and still
>>> can't get anything that looks accurate.
>>>
>>> ;*************************************************
>>> ; plot_AOD.ncl
>>> ;************************************************
>>> load "$NCARG_ROOT/lib/ncarg/nclscripts/csm/gsn_code.ncl"
>>> load "$NCARG_ROOT/lib/ncarg/nclscripts/csm/gsn_csm.ncl"
>>> load "$NCARG_ROOT/lib/ncarg/nclscripts/csm/contributed.ncl"
>>>
>>>
>>> begin
>>> ;-open file and create pointer
>>> a =
>>> addfile("./MYD04_L2.A2008031.2300.005.2008033054757.pscs_000500293068.hdf","r")
>>>
>>> ;
>>> ;-input array od data (203x135ish)
>>> aodr = a->Optical_Depth_Land_And_Ocean
>>> printVarSummary(aodr)
>>>
>>> ;
>>> ;-get array of lat/lon coordinates
>>> latr = a->Latitude
>>> lonr = a->Longitude
>>>
>>> ;
>>> ;-put them both in 1D arrays
>>> aod1= ndtooned(aodr)
>>> lat1= ndtooned(latr)
>>> lon1= ndtooned(lonr)
>>> print("LAT_MAX: "+dim_max(lat1))
>>> print("LAT_MIN: "+dim_min(lat1))
>>> print("LON_MAX: "+dim_max(lon1))
>>> print("LON_MIN: "+dim_min(lon1))
>>> print("AOD_MAX: "+dim_max(aod1))
>>> print("AOD_MIN: "+dim_min(aod1))
>>>
>>> ;
>>> ;-just testing set - should always have missing data
>>> if(any(ismissing(aod1))) then
>>> print("MISSING AOD DATA")
>>> else
>>> print("No missing data")
>>> end if
>>>
>>> ;
>>> ;-generate output grid (half degree global)
>>> ; plon = fspan(-180,180,361)
>>> ; plat = fspan(-90,90,181)
>>> plon = fspan(-180,180,181)
>>> plat = fspan(-90,90,91)
>>> plon_at_units = "degrees east"
>>> plat_at_units = "degrees north"
>>> ;
>>> ;-future filtering placeholder
>>> aod_filter = aod1
>>>
>>> ;
>>> ;-grid data onto platxplon
>>> print("AOD_ONE_DIM : "+dim_max(aod1))
>>> aod_grid = cssgrid(lat1,lon1,aod_filter,plat,plon)
>>>
>>> ; aod_grid!0 = "lat"
>>> ; aod_grid!1 = "lon"
>>> ; aod_grid&lat = plat
>>> ; aod_grid&lon = plon
>>>
>>> ;-keep for reference
>>> ; aod_filter!0 = "lat"
>>> ; aod_filter!1 = "lon"
>>> ; aod_filter&lat= plat
>>> ; aod_filter&lon= plon
>>> ; aod_filter&lon_at_units = "degrees east"
>>> ; aod_filter&lat_at_units = "degrees north"
>>>
>>> aod_grid!0 = "lat"
>>> aod_grid!1 = "lon"
>>> aod_grid&lat = plat
>>> aod_grid&lon = plon
>>> aod_grid&lon_at_units = "degrees east"
>>> aod_grid&lat_at_units = "degrees north"
>>>
>>> ;
>>> ;-create workstation and define colormap
>>> wks = gsn_open_wks("x11" ,"ce")
>>> ;-gsn_define_colormap(wks,"WhBlGrYeRe")
>>> gsn_define_colormap(wks,"testcmap")
>>>
>>> ;
>>> ;-Turn on plotting options then set them
>>> res = True ; plot mods desired
>>>
>>> res_at_cnFillOn = True ; turn on color fill
>>> ;-res_at_cnFillMode = "RasterFill" ; rasterize plot
>>> res_at_cnLinesOn = False ; turn of contour lines
>>> ;-res_at_cnLevelSpacingF = 0.5 ; contour spacing
>>> res_at_gsnSpreadColors = True ; use full range of color map
>>> res_at_lbLabelStride = 4
>>> ;-res_at_pmTickMarkDisplayMode = "Always"; use NCL default lat/lon labels
>>> ;
>>> res_at_gsnAddCyclic = False
>>> res_at_mpMinLatF = -90 ; range to zoom in on
>>> res_at_mpMaxLatF = 90.
>>> res_at_mpMinLonF = -180.
>>> res_at_mpMaxLonF = 180.
>>>
>>> ;
>>> ;-1d monotonic arrays that represent the x/y axes
>>> ; res_at_sfXArray = plon
>>> ; res_at_sfYArray = plat
>>>
>>> ;-plot = gsn_csm_contour_map(wks,aod_grid, res)
>>> plot = gsn_csm_contour_map_ce(wks,aod_grid, res)
>>> end
>>>
>>> 113,1 Bot
>>>
>>>
>>>
>>> _______________________________________________
>>> ncl-talk mailing list
>>> ncl-talk_at_ucar.edu
>>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>>>
>>>
>>

_______________________________________________
ncl-talk mailing list
ncl-talk_at_ucar.edu
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Sat Apr 19 2008 - 11:57:36 MDT