A Hidden Markov Model Applied to the Daily Spring Precipitation over the Danube Basin

Table 3

Fitting by NHMM with seven states of precipitation at the 19 stations, taking into account as predictors (first 10 PCs) for EOF or MEOF decomposition; log𝐿 is log likelihood and BIC represents Bayesian Information Criterion.

Predictors	Variance (%)	log𝐿	BIC
SLP (AE)	83	−125405	256125
SLP (E)	92	−125790	250563
G-850 hPa (AE)	86	−125799	256913
G-850 hPa (E)	94	−122435	250185
SH-850 hPa (AE)	38	−122999	251314
SH-850 hPa (E)	55	−122873	251062
RH-850 hPa (E)	55	−123494	251855
TT-850 hPa (E)	85	−125337	255990
MEOF (slp-G-850-SH-850 hPa)-(AE)	61	−126035	257385
MEOF (slp-G-850-SH-850)-(E)	71	−122651	250619
MEOF (slp-G-850-TT-850 hPa)-(E)	82	−122605	250527
SLP-pp (E)	87	−121812	248939
G-850 hPa-pp (E)	89	−121494	248304
TT-850 hPa-pp (E)	81	−122111	249539
MEOF (slp-G 850 hPa-pp)-(E)	89	−122208	249731