After the image displays in the frame, your selections are saved as default. To generate another plot, just select different value(s) (change to MRF, F024, etc), and click the "BIAS" buttons.
 
 

If you click the BIAS button:
The displayed image represents the AVERAGE BIAS  over the past "N" days for a given selection of  Area, Model, Forecast Hour, and Field.  The BIAS that is used here is  FORECAST minus OBSERVED based from  the initialization.  Therefore the AVERAGE BIAS is just the average of this BIAS calculation over the past 5 or 10 cycles.  The AVERAGE BIAS is plotted in color on the image.  Red indicates a high bias (model forecasting the respective field too high), and Blue is just the opposite (model forecasting the field too low). The yellow line is a BIAS of zero, and the green underlay is the mean pattern of the field over the past 5 or 10 cycles. These images help provide an objective method of discerning model bias over the given time periods.

IMPORTANT !!! This is a "grid-to-grid" comparison, i.e. the "observed" data used to verify the forecast is the initial analysis of the model.  Therefore in data void areas, the "observed" data is heavily weighted toward  the corresponding forecast from the previous model run.  The ECMWF data retrieved at WPC is on a 2.5 x 2.5 degree grid, therefore the ECMWF bias plots will appear more subdued compared to high resolution plots such as the AVN/MRF.   Further, the UKMET data retrieved at WPC is on a 1 x 1 degree grid through 72 hours, but on a 2.5 x 2.5 degree grid 96 to 120 hours.  The lower resolution 96 to 120 hour forecast grids are mapped onto the higher resolution 1 x 1 degree grid (of the 00 hour grid) in order to be processed in this grid to grid fashion.  This means that bias signatures will be potentially more subdued at these lower resolutions compared to other model runs verified on a higher resolution grid.

At times, non-NCEP model data does not arrive cleanly to WPC. This results in bias plots having an over abundance of  mostly blue or  red contours.  (These are also identified by looking at the MEAN contours in GREEN to see if those values look correct).  If this is the case, it is usually because of the 5 or 10 cycles that went into the statistical calculation, at least one of those cycles had a corrupt or incomplete grid.  The manner in which WPC retrieves this data is the culprit and at present is unavoidable.  This situation occurs most often with NOGAPS data.

ENSEMBLE PLOTS are labeled as 12 or 23.  What does that mean ???  These are the number of members in the ensemble and DIRECTLY RELATE TO THE CYCLE IN WHICH THEY WERE RUN.  Plots obtained from the GLOBAL ENSEMBLE (12) imply that 12 members from the 12Z contributed to the verification, whereas plots obtained from the GLOBAL ENSEMBLE (23) imply that 23 members (12 from the 12Z cyc1e and 11 from the previous 00Z cycle ) contributed to the verification. You  can only access these plots by selecting a cycle of 12Z.
 

If you click the BIAS  REMOVE button:
This button allow you to see what the latest model forecast looks like with the bias REMOVED from the run.
 

If you click the HELP AND INFO button:
Well..you got here ok, so that takes care of that.