Table 5.

Candidate models with reach- and sub-catchment scale explanatory variables predicting the median radiant water temperature and the probability of occurrence of cool water in the Pend Oreille River. For each model, the number of parameters (K), −2 log-likelihood (−2logL), Akaike’s information criterion adjusted for small sample size (AIC_c), the difference in AIC_c between the given model and the best-performing model (ΔAIC_c), and Akaike weight (w_i) are shown.

Model	K	−2LogL	AICc	ΔAICc	wϊ
Reach scale
Median water temperature
Confinement, Sum all dense vegetation, z-Area, z-Maximum depth	14	−12.98	60.0	0.00	0.52
Confinement, z-Area, z- Maximum depth	11	−16.09	60.2	0.16	0.48	*
Probability of occurrence of cool-water areas
Confinement, z-Area, z-Maximum depth	6	−25.57	64.2	0.00	0.30
Confinement, Sum all dense vegetation, z-Area, z-Maximum depth	6	−25.57	64.2	0.00	0.30
Confinement, z-Area	4	−27.26	65.0	0.84	0.20	*
Confinement, Sum all dense vegetation, z-Area	4	−27.26	65.0	0.84	0.20
Sub-catchment scale
Median water temperature
Distance from dam, Percent developed land	10	−5.25	34.3	0.00	0.28
Distance from dam, Percent developed land, Tributary area	10	−5.21	34.3	0.01	0.27
Distance from dam	9	−6.22	34.7	0.38	0.23	*
Distance from dam, Tributary area	10	−5.95	34.7	0.40	0.23
Probability of occurrence of cool water
Distance from dam, Percent developed land	8	−19.87	59.6	0.00	0.50	*
Distance from dam, Percent developed land, Tributary area	8	−19.87	59.6	0.00	0.50

An asterisk (*) indicates a model selected based on parsimony in addition to AIC_c scores and weights.