NCL Home > Documentation > Functions > Regridding, Interpolation


Performs areal conservative remapping from one rectilinear grid to another.


	function area_conserve_remap (
		loni [*] : numeric,  
		lati [*] : numeric,  
		fi       : numeric,  
		lono [*] : numeric,  
		lato [*] : numeric,  
		opt  [1] : logical   

	return_val  :  float or double



A one-dimensional array that specifies the longitudinal coordinates in degrees of the fi array. It must be monotonically increasing (W->E) with constant spacing and must be global.


A one dimensional array that specifies the latitudinal coordinates in degrees of the fi array. This must be a monotonically increasing array (S->N). It may be unequally spaced [eg, a gaussian grid]. It may be global or (say) 45S to 45N. Regardless of latitudinal span, it must be longitudinally cyclic (periodic).


An array of two or more dimensions. The two rightmost dimensions must of size nlati x nloni, and are the dimensions used in the interpolation. Missing values, designated by the _FillValue attribute, are not allowed. If any missing values are encountered, the remap will not be performed for that grid.


A one-dimensional array that specifies the longitude coordinates (degrees) of the return array. It must be monotonically increasing and have constant spacing.


A one-dimensional array that specifies the latitudinal coordinates in degrees of the return array. It must be monotonically increasing (S->N). It may be unequally spaced (eg, gaussian grid).


If the variable is True then optional arguments may be present.

If the input and/or output grids are Gaussian and of less than global extent, an optional parameter is required (opt@NLATi for input, opt@NLATo for output) giving the dimension of the *global* latitude set of which the grid is a subset. This is required in order to internally compute the global Gaussian weights from which local box areas are computed. These optional parameters are ignored for equally-spaced latitudinal grids.

There is an internal parameter, bin_factor, that specifies a grid contraction factor. The default bin_factor=1.0. In the uncommon case of binning from low-to-high resolution, an optional smoothing operator, "opt@bin_factor", should be employed since the default behavior will give discontinuous looking results. This parameter expands the area-of-influence in the binning process. It is defined relative to area of an output gridbox. So, for binning from 2x2deg to 1x1deg, a recommended "bin_factor" would be about 4.0 to smooth the data. *NOTE*: the remapping will not be conservative. Remapping is conservative only for "bin_factor=1." which is the default.

Return value

The returned value will have the same dimensions as fi, except for the rightmost two dimensions which will have the same dimension sizes as the lengths of lato and lono. The return type will be double if fi is double, and float otherwise.


area_conserve_remap uses local area-conservative binning to interpolate from high-to-low or low-to-high resolution rectilinear grids. The method accounts for fractional contributions of the input grid points to the scope of each output grid point. There is no provision for missing data. If any grid has a missing value the corresponding return grid will be all missing values.

Use the area_conserve_remap_Wrap function if metadata retention is desired. The interface is identical.

For more robust regridding, see the ESMF regridding examples, which show how to regrid data from and to rectilinear, curvilinear, or unstructured grids. Available in version 6.1.0 and later.

See Also

area_conserve_remap_Wrap, area_hi2lores_Wrap, ESMF_regrid, linint2, linint2_Wrap, latGau, latGauWgt, latRegWgt, NormCosWgtGlobe


Detailed applications of area_conserve_remap and area_conserve_remap_Wrap are available: specifically, examples 6-11.

Example 1

Consider a gaussian grid fi(ntim,klev,nlati,mloni) => fi(120,30,256,512).
Remap to a 2.5x2.5 grid => fo(120,30,72,144)

  fo = area_conserve_remap (loni,lati,fi, lono,lato, False)

  ; Use area_conserve_remap_Wrap if metadata retention is desired.
  ; fo = area_conserve_remap_Wrap (loni,lati,fi, lono,lato, False)

Example 2

Consider high resolution grid, fi(ntim,nlati,mloni), where nlati=400 and mloni=1440. The input latitudes span -49.875 to +49.875 (degrees_north). The input longitudes span -179.875 to +179.875 (degrees_east). The grid is cyclic and no missing values are present. Create a T85 gaussian grid that spans these latitudes [output size: 72x256]. Here, the opt=True and the attribute NLATo is set to the appropriate global value.

  f = addfile ("", "r")
  p = f->PRC
  printVarSummary( p )

  LON  = fspan(-178.5,178.5, 256)
  NLATG= 128           ; appropriate global value
  LATG = latGau (NLATG, "LATG", "latitude", "degrees_north")      ; nominally 90S -> 90N
  LAT  = LATG({-48:48})          ; extract gaussian latitudes for region

  opt = True
  opt@NLATo = NLATG              
  po  = area_conserve_remap (p&lon,p&lat, p , LON, LAT, opt)  ; (ntim,72,256)

  ; Use area_conserve_remap_Wrap if metadata retention is desired.
  ; po  = area_conserve_remap_Wrap (p&lon,p&lat, p , LON, LAT, opt)  ; (ntim,72,256)