Hi Robbi,

Thanks for your answer. I am using Kriging for the interpolation. The semivariogram gives me the information about the spatial dependency of the dataset, and which points (in terms of their location and direction) should be used to estimate my variable at unsampled positions. However, the semiovariogram does not provide an insight on the optimised spatial resolution for my dataset; I mean the grid cell size for the surface to be created by interpolation.

I just found an interesting paper that may help me:

Finding the right pixel size, published in 2006 by Tomislav Hengl at Computers & Geosciences (Link below).

Regards,

Simone

http://www.sciencedirect.com/science/article/pii/S0098300405002657

Thanks Robbi, Michael and Haider for your contributions to my question. It was very helpful. I have analysed the different spatial patterns of my data and found some interesting results for the recommended spatial resolutions if an interpolation is used. They varied from 1x1km to 12x12 km. I am now further investigating the relationship between data quality (input) and scale (output) and discussing this in terms of the scale of different urban issues. I will post the article here in the near future.

Regards,

Simone

Dear Simone,

In the case of the inverse distance interpolation, the optimum power for each integration time should be used while for the method of kriging the semi-variograms should be used.

see the following link; this study was to assess the performance of several rainfall interpolation methods as evaluated by a conceptual hydrological model .