Skip to main content

Table 2 Model comparison and selection for fir, oak, and birch. AIC Akaike information criterion, AUC area under receiver operating characteristic curve

From: Climatic information improves statistical individual-tree mortality models for three key species of Sichuan Province, China

Modela

Equationb

Fit

Test (cross validate)

AIC

AUC

Chi square

P

AUC

Chi square

P

FirI1

4.94 + 0.345*relBAI’

69.34

0.617

3.98

0.86

0.582

11.83

0.16

FirS1

2.74-0.244*DCI + 3.130*CC

62.87

0.645

3.74

0.88

0.628

7.34

0.50

FirC1

0.59-0.316*Tavg-0.296*Pre1 + 0.0100*Pre

28.28

0.733

14.47

0.07

0.715

14.35

0.07

FirIS1

3.65 + 0.480*relBAI’-0.283*DCI + 3.620*CC

53.84

0.682

3.94

0.86

0.662

11.71

0.16

FirIC1

0.41 + 0.730*relBAI’ + 0.0800*D-0.0008*D2-0.336*Tavg-0.358*Pre1 + 0.010*Pre

9.87

0.792

12.48

0.13

0.770

9.44

0.31

FirSC1

6.22-0.282*DCI + 2.770*CC-0.259*Tmax-0.260*Pre1 + 0.00380*Pre

19.10

0.746

14.43

0.07

0.727

6.35

0.61

FirISC1

8.82 + 0.700*relBAI’-0.361*DCI + 3.370*CC-0.333*Tmax-0.278*Pre1 + 0.00430*Pre

0.00

0.795

5.40

0.71

0.780

10.62

0.22

OakI1

4.51 + 0.334*relBAI’

79.05

0.585

8.43

0.39

0.560

4.17

0.84

OakS1

1.96-1.7379*BAL’ + 4.680*CC2

45.95

0.711

21.05

0.01

0.699

16.20

0.04

OakC1

3.24-0.0773*Pre1 + 0.0200*SAND

69.78

0.643

36.38

0.00

0.664

25.43

0.00

OakIS1

0.83 + 0.450*relBAI’-0.0007*D2-2.906*BAL’ + 0.100*SMD + 6.540*CC2

28.22

0.764

9.65

0.29

0.749

8.21

0.41

OakIC1

5.95 + 0.610*relBAI’ + 0.06*D-0.001*D2-0.1102*Tmax7 + 0.0200*SAND

60.47

0.705

10.58

0.23

0.726

14.84

0.06

OakSC1

6.99-2.972*BAL’ + 4.060*CC2-0.174*Tmax7 + 0.150*Tmin1-0.143*Pre1 + 0.040*SAND

20.76

0.803

10.02

0.26

0.773

9.35

0.31

OakISC1

1.77 + 0.640*relBAI’-2.077*BAL’-0.270*DCI +4.810*CC2-0.248*Pre1 + 0.013*Pre7 + 0.039*SAND

0.00

0.845

9.25

0.32

0.828

12.97

0.11

BirchI1

4.86 + 0.762*relBAI’

66.68

0.672

9.40

0.31

0.666

10.44

0.24

BirchS1

0.58-6.661*BAL’ + 0.0033*SMD2 + 4.54*CC2

41.25

0.783

25.01

0.00

0.760

13.08

0.11

BirchC1

4.59 + 0.403*Pre1-0.0310*Pre12-0.189*CN

104.44

0.665

33.94

0.00

0.641

37.69

0.00

BirchIS1

1.95 + 0.58*relBAI’-5.775*BAL’ + 0.003*SMD2 + 4.32*CC2

14.61

0.796

12.70

0.12

0.773

14.13

0.08

BirchIC1

5.61 + 0.80*relBAI’ + 0.08*D-0.115*Pre1-0.0908*CN

57.40

0.716

16.21

0.04

0.703

20.14

0.01

BirchSC1

-0.27-6.525*BAL’ + 0.15*SMD + 5.62*CC-0.118*CN-0.0438*CLAY

37.85

0.784

12.25

0.14

0.773

17.54

0.02

BirchISC1

4.53 + 0.620*relBAI’-6.501*BAL’ + 4.80*CC2 + 0.0031*SMD2-0.0024*Pre-0.0332*CLAY

0.00

0.821

11.41

0.18

0.810

12.09

0.15

  1. aAll the models were named with species plus one of the seven compositions of three scales plus the serial number of each model (1–n), such as FirI1–FirIn, FirIS1–FirISn, FirISC1– FirISCn, etc.
  2. b relBAI’ = log (relBAI + c); c was introduced to deal with 0 values. c = (25 % −quantile) 2/ 75 % − quantile. In addition, the quantiles were calculated for relBAI in all data set without 0 (Wunder et al. 2008). BAL’ was log-transformed as the same method of relBAI
Back to article page

Annals of Forest Science

ISSN: 1297-966X