Table 1. Root mean square errors (RMSE) obtained with three different datasets for several statistical models: linear regression (L), quadratic regression (Q), cubic regression (C), dynamic linear models with and without trend (DLMs, DLM0), and linear-plus-plateau (LP).
| Models | ||||||||
| Dataset | Quantity | Units | L | Q | C | DLMs | DLM0 | LP |
| France | RMSE | t ha−1 | 0.60 | 0.54 | 0.50 | 0.47 | 0.38 | NA |
| Differencea | % | 55.38 | 40.85 | 28.95 | 22.82 | 0.00 | NA | |
| France (restricted) | RMSE | t ha−1 | 0.59 | 0.53 | 0.48 | 0.46 | 0.38 | 0.50 |
| Differencea | % | 57.62 | 42.32 | 29.13 | 23.03 | 0.00 | 32.30 | |
| Global | RMSE | t ha−1 | 0.40 | 0.35 | 0.32 | 0.23 | 0.20 | NA |
| Differencea | % | 96.98 | 74.03 | 59.76 | 15.11 | 0.00 | NA | |
a
The differences with respect to the lowest RMSE values are expressed as a percentage of the lowest RMSE values (RMSEmin); Difference = 100*(RMSE – RMSEmin)/RMSEmin.