[Met_help] Multiplicative Bias used in PointStat (UNCLASSIFIED)

[John Halley Gotway]

Barbara,

Multipicative Bias is simply the ratio of the mean forecast value to the mean observation value.  It's listed in Appendix C of the MET User's Guide (METv1.1/doc/MET_Users_Guide.pdf) on page C-8.

The term "bias" is used for both continuous data (i.e. raw rainfall amounts) and thresholded categorical data (i.e. rainfall amounts > 0.1 inches).  When applied to continuous data, it's often called
"multiplicative" bias and defined as fbar/obar.  When applied to categorical data, it's often called "frequency" bias and defined as (hits + misses)/(hits + false alarms).

For example, when multiplicative bias = 1.1, your forecast is 10% too high on average.  And when frequency bias = 1.1, you're forecasting the event 10% too often.

Here's a good link for more information about both types of biases:
http://www.bom.gov.au/bmrc/wefor/staff/eee/verif/verif_web_page.html

Hope this helps.

John


Sauter, Barbara (Civ, ARL/CISD) wrote:
> Classification:  UNCLASSIFIED 
> Caveats: NONE
> 
> Hello and happy 2009,
>      Can someone explain to me (hopefully with more words than
> equations) what multiplicative bias is?  It is one of the statistics
> generated by the MET PointStat tool in the cnt (continuous statistics)
> output file.  I tried looking it up with Google and Wikipedia, but did
> not find an explanation.
> 
> Thank you,
> Barb Sauter
> U.S. Army Research Laboratory
> White Sands Missile Range, NM
> (575) 678-2840
> bsauter at arl.army.mil
> 
> Classification:  UNCLASSIFIED 
> Caveats: NONE
> 
> _______________________________________________
> Met_help mailing list
> Met_help at mailman.ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/met_help