michelleharrold
commented
4 years ago This issue came up again for the AF project on 8/17/2020.
The group discussed using the grid-diag tool for grid-to-grid evaluations, but this is not sufficient for grid-to-point evaluations. Would still be a worthwhile endeavor, but would need to secure funding.
While Point-Stat writes the MPR output line type and Grid-Stat writes matched pairs to a NetCDF file, it would be convenient to enhance both of them to write a new line type to describe the distribution of the raw forecast, observation, and climatological mean values for each verification task.
Could choose multiple line types for each data type or a single one for call three: FHIST, OHIST, CHIST for forecast, observation, and climo mean histograms... analogous to RHIST. Or DHIST for Data Histogram or MPRHIST. For counts, suggest using HIST in the now. For rates (i.e. divide by the total), suggest using PDF in the line type name.
Would need to add a new config file option to define the bin size for each variable. Recommend that the default bin size be set to bad data... which would result in a runtime error. Therefore, users requesting this output line type would be required to define the bin size for each variable.
Once this output is written, METviewer would need to be enhanced to load it, aggregate across multiple cases (enforcing compatible bin sizes), and plot it.
This request was made by Jamie Wolff on 11/25/2019 for her cloud evaluation project with MMM.