Ratio limits · January initialization · FMAM and its months

The operational correction limits the spread ratio r = σ_obs/σ_model to 0.5–2. This study runs the same screening and confirmation for that correction with wider limits (1/3–3, 1/4–4) and none, all with the same fallback (r = 1 where the model spread is below 1 mm), to see whether the limits are worth changing. ✓ = better than the operational limits with Holm-adjusted p < 0.05 across the 3 alternatives × 5 targets and not worse over the test years. Run at commit fb7425f.

Confirmed better than the operational limits — published (blended) probabilities: none; rainfall amounts: MVS, ratio limits 1/3-3 (Feb), MVS, ratio limits 1/4-4 (Feb, Mar) and Variance Scaling (unbounded) (Feb, Mar).

Probabilities

Blended tercile RPSS through the full pipeline, outer leave-one-year-out 1993–2016
Method	Feb	Mar	Apr	May	FMAM
Mean-Variance Scaling (operational, 0.5–2)	+0.027	+0.002	+0.001	+0.028	+0.032
MVS, ratio limits 1/3-3	+0.029	+0.002	+0.001	+0.028	+0.033
MVS, ratio limits 1/4-4	+0.029	+0.003	+0.002	+0.028	+0.033
Variance Scaling (unbounded)	+0.029	+0.003	+0.002	+0.028	+0.033

Amounts

Amount CRPSS, leave-one-year-out 1993–2016
Method	Feb	Mar	Apr	May	FMAM
Mean-Variance Scaling (operational, 0.5–2)	+0.070	-0.044	-0.016	+0.007	+0.025
MVS, ratio limits 1/3-3	+0.076 ✓	-0.041	-0.016	+0.007	+0.025
MVS, ratio limits 1/4-4	+0.076 ✓	-0.040 ✓	-0.016	+0.007	+0.025
Variance Scaling (unbounded)	+0.076 ✓	-0.040 ✓	-0.016	+0.007	+0.025

Where the operational limits act

Share of the corrected area, mean over the leave-one-year-out folds, 1993–2016
Target	Lower limit active (r would be < 0.5)	Upper limit active (r would be > 2)	Spread–error, operational	Spread–error, no limits
Feb	20%	9%	1.02	1.05
Mar	2%	9%	0.93	0.95
Apr	0%	6%	0.95	0.96
May	0%	6%	0.96	0.97
FMAM	0%	7%	0.98	0.99