data interpolating or resampling and which method?

Hi everyone

I have a a large number of daily precipitation nc data files for an area of 1780 sqrkm that every single time has data of daily precipitation in form of 13*5 martix (13 longitude and 5 latitude). i have to transform this data to 1 km resolution and also change the temporal to another time intervals( for example 3 days). I know that downscaling of this data carries uncertainities but i have no other choice and want to select the best method to do that with least uncertainity. my region located in arid regions and included about 800 sqrkm plain part. Based on these information is there anyone have experienced about the best method and advise? If dont so please at least guide me about which one is better curve fit or surface fit? and which method most suited my regions (linear, nearest, pchip, cubic spline or v4, ...). I have recently started learning matlab and stucked a number of days on this problem. any advice or guide or help would be highly appreciated.

[Sadegh Dalfardi]

I used the fit function in matlab and it gave me some information but also i need any suggested advice if there is, and also apreciate any helps, advices and experiences.

FO=fit(X,Y,FT)