As of met-6.1, the config_constants.h file contains 4 structs which include interpolation options: InterpInfo, RegridInfo, NbrhdInfo, and HiRAInfo.
First, create a new struct named InterpOpts which includes all the interpolation options, as show in the attached version of config_constants.h. Factor the interpolation options out of the 4 structs listed above and just include an instance of InterpOpts.
Add a parse_conf_interp_opts() utility function and update the parsing functions for the other structs to call it.
Next, add a new ShiftDataInfo struct which includes an instance of InterpOpts as well. In the MET config files, let the user define:
After reading gridded data, we'd apply a rigid shift of the data as defined by fr_lat_lon and to_lat_lon. We could also support this for point observations by updating the observation lat/lon locations.
This is all useful for doing feature-relative verification where we're tracking a feature, like a hurricane.
Need to do some more brainstorming here. Could also consider reading the TCST output from TC-Pairs to define what the lat/lon shift should be. Could also read the ascii output from MTD for the same reason.
Think about how this should work for grid_stat and point_stat, but also for series_analysis where the times differ.
[MET-914] created by johnhg
As of met-6.1, the config_constants.h file contains 4 structs which include interpolation options: InterpInfo, RegridInfo, NbrhdInfo, and HiRAInfo.
First, create a new struct named InterpOpts which includes all the interpolation options, as show in the attached version of config_constants.h. Factor the interpolation options out of the 4 structs listed above and just include an instance of InterpOpts.
Add a parse_conf_interp_opts() utility function and update the parsing functions for the other structs to call it.
Next, add a new ShiftDataInfo struct which includes an instance of InterpOpts as well. In the MET config files, let the user define:
shift_data = {
fr_lat_lon[] = { 30, -100 };
to_lat_lon[] = { 35, -110 };
method = NEAREST;
width = 1;
}
After reading gridded data, we'd apply a rigid shift of the data as defined by fr_lat_lon and to_lat_lon. We could also support this for point observations by updating the observation lat/lon locations.
This is all useful for doing feature-relative verification where we're tracking a feature, like a hurricane.
Need to do some more brainstorming here. Could also consider reading the TCST output from TC-Pairs to define what the lat/lon shift should be. Could also read the ascii output from MTD for the same reason.
Think about how this should work for grid_stat and point_stat, but also for series_analysis where the times differ. [MET-914] created by johnhg