Jump to Main Content
Description and testing of a weather-based model for predicting phenology, canopy development and source–sink balance in Vitis vinifera L. cv. Barbera
- Cola, Gabriele, Mariani, Luigi, Salinari, Francesca, Civardi, Silvia, Bernizzoni, Fabio, Gatti, Matteo, Poni, Stefano
- Agricultural and forest meteorology 2014 v.184 pp. 117-136
- Vitis vinifera, atmospheric precipitation, canopy, data collection, decision support systems, leaf area, meteorological data, model validation, models, phenology, prediction, shoots, temperature, time series analysis, vines, vineyards, weather stations
- A dynamic crop growth model based on availability of a minimal set of weather data and elemental vineyard and plant characteristics is described and its accuracy at predicting phenology, leaf area development, light interception and pending yield evaluated. Outputs calibration based on field measurements was carried out over 2011–2012 on hedgerow trained cv. Barbera (Vitis vinifera L.) vines having either low-density (LD, ≅10 shoots per m of row) or high density (HD, ≅ 60 shoots per mof row) canopies. Model validation was then carried out based on a five year (2003–2007) independent dataset characterized by a wide inter-annual variability of meteorological conditions. During calibration, MAE, CRM and R2 values evinced high-to-very high model accuracy at predicting phenology, leaf area development, total light interception and pending yield for any year×treatment combination. The validation procedure showed that, despite large inter-annual weather variability, final yield was very accurately predicted (R2=0.96 in LD and R2=0.94 in HD), whereas precision at simulating final total leaf area was higher in HD (R2=0.92) than in LD (R2=0.63). Due to its flexibility the model can also be easily applied to the analysis of long time series of meteorological data coming from weather stations where only daily maximum and minimum temperature and daily precipitation are available and it can also be run in a forecast mode. Moreover, the broad range of released outputs renders the model an ideal tool to be used in decision support system applications which typically rely on real time or forecast estimation of vine development parameters.