Table 3.

Average Brier Score, Average Reliability, Average Resolution, and Average Ignorance for Below-Normal and Above-Normal Categorical Forecasts Obtained From the Leave-One out Cross-Validated Forecasts^a

	Below Normal				Above Normal
Model/Multimodel	$\stackrel{‒}{BS}$	$\stackrel{‒}{REL}$	$\stackrel{‒}{RES}$	$\stackrel{‒}{IGN}$	$\stackrel{‒}{BS}$	$\stackrel{‒}{REL}$	$\stackrel{‒}{RES}$	$\stackrel{‒}{IGN}$
Resampling (Res)	0.217	0.048	0.052	0.886	0.212	0.019	0.029	0.933
Regression (Reg)	0.199	0.015	0.037	0.883	0.210	0.018	0.029	0.847
MM1 (Res + Clim)	0.199	0.019	0.041	0.857	0.204	0.005	0.023	0.847
MM2 (Reg + Clim)	0.196	0.013	0.037	0.866	0.207	0.005	0.019	0.838
MM3 (MM1 + MM2)	0.197	0.011	0.035	0.861	0.205	0.005	0.021	0.840
MM4 (Res + Reg + Clim)	0.197	0.009	0.033	0.853	0.203	0.006	0.024	0.841
MM5 (Res + Reg)	0.203	0.032	0.051	0.864	0.204	0.016	0.033	0.868
MM6 (equal weights)	0.204	0.020	0.036	0.870	0.207	0.018	0.033	0.874
MM7 (long-term skill)	0.204	0.013	0.029	0.875	0.208	0.020	0.033	0.873

^aThe average ignorance score of climatological forecasts is log₂ n_c, where n_c is the number of categories of forecasts. For tercile forecasts, ${\stackrel{‒}{IGN}}_{\text{climatology}}=1.585$.