Predicting individual tree growth using stand-level simulation, diameter distribution, and Bayesian calibration

Table 3 Posterior probability distribution of the parameters in the individual tree diameter increment model. The posterior distribution is not analytical and it is characterized here by the mean values of the parameters, the standard deviation (SD) and the maximum a posteriori estimates (MAP). The prior was set as independent uniform distributions, of which the maxima and minima are listed

Parameters	Predictor	Prior	VB model			CB model			RB model			Recalibrated model
Parameters	Predictor	Prior	Mean	SD	MAP	Mean	SD	MAP	Mean	SD	MAP	Mean	SD	MAP
c₁	Constant	(− 100,100)	− 0.3014	0.0712	− 0.3129	− 1.5807	0.2388	− 1.5154	− 0.2250	0.8122	− 0.3144	− 1.3257	0.3396	− 1.1309
c₂	G_>d	(− 100,10)	− 0.1364	0.0011	− 0.1368	− 0.0579	0.0040	− 0.0578	− 0.0355	0.0096	− 0.0335	− 0.0905	0.0034	− 0.0901
c₃	lnG	(− 100,10)	− 0.2878	0.0213	− 0.2887	− 0.0480	0.0405	− 0.0027	− 0.5429	0.1739	− 0.5211	− 0.2751	0.0372	− 0.2796
c₄	\( \sqrt{d} \)	(− 10,100)	0.8385	0.0162	0.8366	0.6758	0.0623	0.6576	0.48914	0.1089	0.4764	0.7822	0.0858	0.7406
c₅	d²	(− 10,1)	− 0.0039	0.0001	− 0.0039	− 0.0024	0.0001	− 0.0023	− 0.0015	0.0003	− 0.0014	− 0.0030	0.0002	− 0.0029
c₆	\( {\overline{h}}_{30} \)	(− 10,100)	0.0636	0.0040	0.0662	0.0915	0.0166	0.0792	0.1408	0.0681	0.1420	0.1360	0.0136	0.1323

ISSN: 1297-966X