interpolating    mapping    searching

Interpolating Data

In many ways, the proliferation of image-processing software has put interpolation techniques at everyone's finger-tips. However, casual use of such techniques can result in misleading and alarming misrepresentations of data. Simply "warping" an image from one projection to another without regard for how the "warping" is defined may lead to erroneous conclusions simply because the interpolation was inappropriate.

In this section, we begin with an internal paper describing a case study for interpolation. The considerations of this case study can and should be more widely applied by any scientist considering interpolation of other kinds of source data. We also include downloadable software toolkits that are examples of flexible systems for interpolating passive microwave and MODIS data. The toolkits can be used to evaluate interpolation candidates for a particular application, and then to produce customized scripts for accomplishing it.

Interpolation : A SMMR case study of different interpolation techniques. Several methods for resampling SMMR Pathfinder data to the EASE-Grid format are compared. The following methods are described and their properties are discussed: drop in the bucket, nearest neighbor, bilinear, cubic convolution, sin(x)/x, inverse distance squared, Backus-Gilbert. Where appropriate each method is simulated by a 2-d transfer function and the frequency response is derived and is used to describe the expected properties of the method. Three methods, drop in the bucket, nearest neighbor, and inverse distance squared are tested on actual data.

Mapx : A mapping and gridding transformations library in C and Java. Mapx includes several command-line applications that implement various interpolation methods. See, for example, regrid.c.

AS2GT : AMSR Swath-to-Grid Toolkit. This suite of software tools subsets and grids AMSR-E swath data. Take a look at the file gsgrid.c to see examples of using forward navigation to implement interpolation methods that usually require reverse navigation.

MS2GT: MODIS Swath-to-Grid Toolbox. Developed at NSIDC, this set of software tools grids MODIS swath data. The file fornav.c is a good example of elliptical weighted averaging. Newer tools for gridding MODIS swath data have been developed elsewhere. Most users will find the MODIS Swath Reprojection Tool or the HDF-EOS To GeoTIFF Conversion Tool easier to use than MS2GT.