Interpolating large datasets

Dear Matlab Central,

I want to interpolate a very large evenly gridded data set (in this case 14000x12000 points) onto a lower res dataset, like 1400x1200 or even lower. As you might expect Matlab can't do this easily. Even if I turn a script on and then go home and return the next day, it has not finished :P (I use griddata)

Does anybody know of a way to speed this up, or to split the dataset up into small parts that are than evaluated one by one? I tried this latter way (splitting it up into seperate grids of 2000x2000 and interpolating this on 200x200 grids) but feel that there will be problems at the boundaries of these interpolations.

It is quite a simple data set. It only consists of 0s and 1s (it's a mask) so I'd presume it should be possible to do this large yet rather simple interpolation :(

Can anybody help me? Thanks in advance.

As said I now use griddata together with meshgrid to do this.

[Jan]

Which kind of interpolation do you want? Is a blockwise mean sufficient?

Answer 1

[JMvanwessem]

Hah. Back then I didn't see your reactie Jan. But I found an alternative way of doing it. Now I am back to this and I really do want to solve this problem of interpolating large datasets. It really should be a quite general process, shouldn't it?

The matter has shifted a bit though. I want to interpolate two quite similar data-sets onto each other (it is on the same "region" but the latlon breakpoints are just slightly shifted). Like interpolating a 5000x5000 dataset onto a 4000x4000 dataset. It contains the same kind of data (height) data so I would just like to linearly interpolate it. But matlab cannot get this going. Can anybody just help me to know how to approach this problem?

Thanks.

P.S. Blockwise mean sounds good, but is that really what I want?