This is a request from Partha Bhattacharjee at NOAA/EMC. He's already running the Series-Analysis tool and would like an easy way of dumping a time series of the data at each grid point. He was running Grid-Stat for each time in his series (length 50) but then has 50 NetCDF output files to slice and dice to extract a time series. Enhancing Series-Analysis to dump the paired fcst/obs values would be very convenient for him and much easier to extract a time series of paired data.
I thought about adding support for CLIMO, but that's a big(ger) can of worms.
The big wrinkle here is that we'd need to add a new dimension to the output NetCDF file. There are many ways the series can be defined:
same init time, multiple lead time
multiple valid times
same valid time, multiple lead times (revision series)
same timing info, multiple levels
The question is how refined should the output series info be?
Suggest that the add a new "series" dimension, and also add variables for:
fcst_init(series), fcst_valid(series), fcst_lead(series)
obs_init(series), obs_valid(series), obs_lead(series)
Should this always be added or only when MPR output is requested?
TaraJensen
commented
5 years ago
This is a request from Partha Bhattacharjee at NOAA/EMC. He's already running the Series-Analysis tool and would like an easy way of dumping a time series of the data at each grid point. He was running Grid-Stat for each time in his series (length 50) but then has 50 NetCDF output files to slice and dice to extract a time series. Enhancing Series-Analysis to dump the paired fcst/obs values would be very convenient for him and much easier to extract a time series of paired data.
Suggest adding support for: output_stats = { ... mpr = [ "FCST", "OBS" ]; }
I thought about adding support for CLIMO, but that's a big(ger) can of worms.
The big wrinkle here is that we'd need to add a new dimension to the output NetCDF file. There are many ways the series can be defined:
The question is how refined should the output series info be?
Suggest that the add a new "series" dimension, and also add variables for: fcst_init(series), fcst_valid(series), fcst_lead(series) obs_init(series), obs_valid(series), obs_lead(series)
Should this always be added or only when MPR output is requested?