Grid-Radius is an NGDC term for a method for qualifying the data in grids. The method involves omitting grid cell data where the distance to actual data values is more than a specific distance, know as the Grid-Radius value. Simply put, if a grid-radius of n cells is applied to a grid, this means that any cells for which the nearest original data value is more than n cells distant will be filled with the Empty Grid Cell value. If grid-radius is not applied to a grid, all the cells in the grid will contain values, no matter how far away the nearest original data was.
When creating the Coastal Relief Model grids NGDC did not apply grid-radius to the grids (except on the far periphery of the NOS data extent, see note below). But there are some areas where data coverage is sparse or even missing. For these areas the grid cell values are interpolated based on data which is sparse and/or far from the cell. Areas without data coverage may show up as artificially flat areas of bathymetry. You may want to omit these areas, especially if you are building high resolution grids.
If you wish to omit grid values where the actual data was not in close proximity, you may apply the Grid-Radius algorithm. (It is an option in the Grid Parameters Dialog.) when creating your grids.
NOTE: Grids on the far periphery of the NOS hydrographic data coverage have already had a Grid-Radius value of 110 applied to them, since grid cells any further from the data coverage would be totally speculative.
Note: Since there is not density data for land grids or for some lakes (such as Great Lakes Bathymetry) Grid-Radius cannot be applied to these areas. So Grid-Radius values which you set for your grid will be ignored in these areas.