[Met_help] time staggering of traditional and object verification

[John Halley Gotway]

Jonathan,

Let me mention two things...

(1) We do not currently have plans for enhancing the MET tools to look for time offsets in the traditional metrics in a slick way.  However, that is certainly something you could already investigate
using clever scripting.  When you run Grid-Stat or Point-Stat, you're welcome to compare forecasts and observations that are offset in time.  For example, suppose you'd like to verify a forecast valid
at time T.  You'd typically verify it using observations valid at that same time T.  But you could also verify it with observations valid at times T-3, T-2, T-1, T+1, T+2, T+3.  Then see how the
performance differs for each time offset used.

When doing this, you'll want to be careful how you name the output files so that they don't overwrite eachother.  I'd suggest using the "output_prefix" option in the config files to differentiate the
output file names.  And you may want to modify the "model" name in the config file to distinguish between the verification output when using different time offsets.

So you'll probably want to define separate configuration files for each time offset used.  Or if you're scripting it, you could use environment variables in the configuration file, and be sure to set
them how you want them before each run.

(2) We are working on enhancing the MODE tool to extend it into the time dimension.  Currently, you could take the same approach in MODE as what I've suggested above.  But we're working on a version
of MODE that would actually do a 3D convolution - 2 dimensions in space and 1 dimension in time.  This create "potato" shaped objects - and the longer they are, the more the storm objects are moving
in time.  I'm not sure when this tool will be made available, but I'd guess in the next 6-12 months.

Hope that helps.

John

Case, Jonathan (MSFC-VP61)[Other] wrote:
> Hello John,
> 
> I was wondering whether the MET package may in the near future include a capability of performing time staggering of traditional verification and object feature matching?   
> 
> For example, there might be instances in which the model is slow to initiate convection by an hour or two, but the positioning might be accurate if verified against a time-staggered observed grid.  
> 
> Many thanks!
> Jonathan
> 
> --------------------------------------------------------------------------------------------------
> Jonathan Case, ENSCO Inc.
> NASA Short-term Prediction Research and Transition Center (aka SPoRT Center)
> 320 Sparkman Drive, Room 3062
> Huntsville, AL 35805
> Voice: 256.961.7504
> Fax: 256.961.7788
> Emails: Jonathan.Case-1 at nasa.gov / case.jonathan at ensco.com
> --------------------------------------------------------------------------------------------------
> 
> "Whether the weather is cold, or whether the weather is hot, we'll weather 
>   the weather whether we like it or not!"
> 
> 
>