Table 7.
Yield models’ output (quintals per hectare) related to real plant number without missing plants (Yield RVN) using NDVI filtered (NDVI_f), canopy thickness (Thick), NDVI filtered and canopy thickness (NDVI_f Thick) compared to yield ground estimation and yield harvested. The best resulting models are bold typed.
| Year | Vineyard | NDVI_f | Thick | NDVI_f Thick | Yield ground estimation | Yield harvested | ||||
|---|---|---|---|---|---|---|---|---|---|---|
| Linear | Trees | Linear | Exp GPR | Linear | SVM | |||||
|
Yield RVN (best models without cross validation) |
2017 | Caggio | 37.3 | 35.8 | 46.5 | 42.9 | 35.7 | 48.5 | 49.1 | 44.3 |
| 2018 | Caggio | 136.8 | 132.1 | 115.5 | 116.8 | 129.1 | 114.3 | 73.6 | 71.1 | |
| 2019 | Caggio | 126.2 | 116.4 | 118.0 | 114.0 | 124.5 | 106.6 | 78.5 | 110.8 | |
| 2019 | Solatio | 100.0 | 112.7 | 65.4 | 64.7 | 85.5 | 70.8 | 70.0 | 53.6 | |
| 2019 | Belvedere | 145.2 | 161.8 | 83.3 | 82.2 | 117.6 | 97.2 | 69.3 | 73.7 | |
| RF boosted | RF boosted | RF boosted | ||||||||
|
Yield RVN (best model with fivefold cross calibration) |
2017 | Caggio | 34.1 | 40.1 | 36 | 49.1 | 44.3 | |||
| 2018 | Caggio | 134 | 116 | 128.2 | 73.6 | 71.1 | ||||
| 2019 | Caggio | 118.6 | 115 | 118.2 | 78.5 | 110.8 | ||||
| 2019 | Solatio | 109.3 | 64.8 | 90.5 | 70.0 | 53.6 | ||||
| 2019 | Belvedere | 157.5 | 81.3 | 126.1 | 69.3 | 73.7 | ||||