Nov 13, 2015

OSNW3|WxClimate Knowledge Base: 2015-16 Current Performance and Members Used in the Model Ensemble

The OSNW3|WxClimate daily forecast Trends began forecasting and accumulating verification data on October 1, 2015. I often look at them to analyze how the forecasts are performing. On November 13, 2015 I selected the MaxT Trend for Green Bay, WI.

October 2015 MaxT Trend for Green Bay, WI


November 2015 MaxT Trend for Green Bay, WI


A quick glance shows that some individual Members out performed the Ensemble, the Mean of all Members. I expect this to happen. Also, the early November miss could be blamed on the removal of the "Full Blend" analog for the 2015-16 testing. I would have to perform a deeper analysis to prove this however. It could just be a miss, and that is OK.

Daily Frequency Used To Project Weather Conditions


The Members are listed below.

M1 = 'Regional+' 15 Day Correlation Mode
M2 = Climate 'Region 6' Top Correlation
M3 = 'CONUS' Top Correlation

Consider this an attempt at Group Velocity. The 'Regional+' bucket is top R of individual stations. My reasoning to use this parameter is to capture the high frequency carrier wave based on standing wave notation, as I ignore all frequencies lower than 30. The 'Region 6' bucket is top R of all stations in the particular climate region. The 'CONUS' bucket is top R of all contiguous US stations. My reasoning to use these parameters is to capture the low frequency amplitude modulation. Molded into an envelope I attempt to project long-term weather conditions based on the amplitude of pattern correlation.

As always this is a work in progress. If there are any questions, comments, or suggestions on the material presented please let me know. Click on the images for a larger view. Thanks for reading!