Dear Experts;
I have point base data in text files (no of sample are 324) in such a way first column of Lat, Second of Lon and third column of temperature. I want to create interpolation surface and want to select which interpolation(IDW,nearest,bilinear,natural) is best for my data set.
In literature , i find for selection of best interpolation, i have to use cross validation and validation( will find out RMS error and correlation) . I search for these terms but not able to understand what these are? Suggest me and tell me how can i use cross validation and validation( for calculation of RMS error and correlation) in matlab?? How can i divide my data in train and testing(not sure what these terms are?) data sets?
I have upload one of text file with this question
The data are from a uniformly spaced grid, right? Do you want to be able to get a temperature estimate at additional locations (not just the grid centers?) One good thing about kriging (which you did not plan to use or mention) is that it provides a mean and variance. IDW will not give you the same kind of variance results. Kriging can also be forced to reproduce the data that you have -- so the estimate for say 21.5, 62.5 will match the data observed. Is there any measurement error in the data values? I would also check for some basic spatial dependence in the data. Kriging works by modeling that dependence. It might be easier to provide some direct help if we know what you are trying to achieve.
Thanks @Morton. Yes this is uniformly space grid data(for lat,lon point of view) . I plan to use kringing but matlab has not as such build in tool to perform it. I want to apply this interpolation also. I would like to perform cross validation and validation so that i able to get RMS error and correlation among different interpolation technique. After calculating RMS error and correlation, i offer best interpolation technique.( Please check it whether i am going right or not?)
Yes i have fill NaN values in my data set by linear interpolation. Hope i answer all requirements.
This book: MATLAB® Recipes for Earth Sciences by Martin H. Trauth should get you comfortable, firstly in the terminologies (cross-validation, etc) and examples (in Matlab), as well. If you can download the data and follow closely, you should be able to find your way out. It helped me a lot in interpolating wind velocity in my PhD work.
Dr. Bawe, I tried the method of cross-validation from your suggested book and I get the mean and std, but I want to make sure that the best method that has the lowest std?!
dear @Nour Khashan I hope this publication will fine u helpful
https://www.researchgate.net/publication/307900696_Missing_Data_Imputations_for_Upper_Air_Temperature_at_24_Standard_Pressure_Levels_over_Pakistan_Collected_from_Aqua_Satellite
Article Missing Data Imputations for Upper Air Temperature at 24 Sta...
- Badges
- Science topic
- Similar topics
- Organizational Psychology
- Training