Table 2.
Predicting season length (SEASL)
| Model | Variable | Weight | Coeff. | Std. error | z-value | Pr(>|z|) |
|---|---|---|---|---|---|---|
|
Early |
Intercept |
|
59.57 |
15.42 |
3.86 |
< 0.001 |
| NDWI10.21 |
1 |
85.23 |
31.52 |
2.7 |
0.007 |
|
| DAY_PREC2–13 |
0.99 |
-0.5 |
0.14 |
3.65 |
< 0.001 |
|
|
LST
8–19
|
0.78 |
1.5 |
0.85 |
1.76 |
0.079 |
|
|
Early + Late |
Intercept |
|
-19.11 |
28.45 |
0.67 |
0.501 |
| NDWI10–21 |
1 |
104.26 |
31.36 |
3.32 |
0.001 |
|
| LST16–27 |
0.98 |
3.78 |
1.26 |
2.98 |
0.003 |
|
| DAY_PREC20–31 |
0.79 |
0.29 |
0.1 |
2.88 |
0.004 |
|
|
LST
8–19
|
0.53 |
0.1 |
1.11 |
0.09 |
0.926 |
|
| DAY_PREC 2–13 | 0.4 | -0.28 | 0.16 | 1.73 | 0.083 |
The average weight and significance of variables remaining in the two best 'Early predictors only' and six best 'Early + Late predictors' models. Note that terms in italics are significant in some of the selected best models but not in others, and that overall, weighted model averaging procedures suggest that they are not significant.