How to interplate observations to model grids

[xshao]

Hi,
I am going to do ocean data assimilation, and I have
ocean surface observation data with latitude and longitude
information. Now I want to interplate the observations to
model horizontal [lat lon] girds.
So my question is, which method should I use to interplate
the observations to the model horizontal [lat lon] grids?
Also, could you recommend some references associated with
the interplation method？
Anyone could help me to figure it out?
Thanks in advance.

[kate]

The code for converting from lat,lon to i,j is in ROMS/Utility/grid_coords.F for both FLOATS and STATIONS.

[guokunlv]

There are different assimilation methods.But as far as I know, many of the assimilation method don't need to interp scatter observations to the model grid. Just like the Optimal Interpolation:it caculate observation and model discrepency on observation point and then interpolate the incremental to model space based on background error Covariance and observartion error covariance. Also you can research on the 4DVAR in ROMS.

[wzhu]

I am also trying to do the job like this. I am trying to use Matlab to interpolate the 1 degree temperature data (WOA09) onto the open boundary line. It is obvious a kind of 2-D interpolation, but what we want are not the meshgrid data, we need the data on the model grid nodes. But it seems that it does not work, I am still trying.

Any suggestions?

[arango]

The 4D-Var observation NetCDF file for ROMS is very special since it needs several observations variables and a particular metadata design. Check tutorial 13 in WikiROMS. If you are doing 4D-Var data assimilation, you need to study and master all these tutorials. Otherwise, you will not be able to do data assimilation in your application. These algorithms are complex and you need to master them and read the appropriate literature.

All the Matlab scripts to process observations for data assimilation are in the matlab repository, directory matlab/4dvar.

[guokunlv]

your jobs are something like: climatoligy nuding, I think. Your just relax the boundary to WOA09 climatology. In ROMS probably there are some ana_funciton to do this work. Also , if you have to change the nudging coefficient to get an satisfactory result.