| Re: Interpolation from irregular to regular grid [message #49724 is a reply to message #49666] |
Thu, 10 August 2006 10:54   |
James Kuyper
Messages: 425
Registered: March 2000
|
Senior Member |
|
|
adisn123@yahoo.com wrote:
> Hi,
>
> WHile reading a paper relavant to my project, I found one thing really
> questioning.
>
> THe paper used an interpolation method to obtain a regular
> three-dimensional grid
> from the irregular three -dimensional data set.
>
> I wonder whether there is any advantages in reforming an irregular data
> grid to a regular grid.
> Whatever the irregular data points are, aren't there the real data? If
> you change to a regular grid, then don't the data values slightly
> change?
>
>
> That is,
> my question will be
> when, and why do we want to change from an irregular grid to a regular
> grid in 3D?
The main reason is convenience. Writing routines to work with
irregularly gridded data is very complicated and tricky. Routines for
handling regularly gridded data are much easier to write, far more
efficient, and correspondingly easier to find. The improved performance
is especially noticeable in interpreted languages like IDL, where the
kinds of loops you have to write to handle an irregular grid are very
inefficient, and cannot be easily replaced with the much faster matrix
operations.
Converting irregularly gridded data to a grid inherently involves loss
of information, and introduction of extra, spurious data. Making
reasonable error estimates on the results of such an analysis can be
very tricky. I'd avoid it, if you can, but in this case "if you can" is
a big "if".
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
