ISHS Acta Horticulturae 754: International Workshop on Advances in Grapevine and Wine Research
THE COHERENCE OF A NOVEL FORECAST MODULE IN R LANGUAGE FOR 'CHARDONNAY' HARVEST |
| Authors: | G. Bongi, E. Eccel, G. Toller |
| Keywords: | Vitis vinifera L., matrix languages, harvest prognosis, parameter economy |
| Abstract:
We assessed the feasibility of an R-language harvest forecast module for ‘Chardonnay’ in the grape cultivation areas of Trento province (Northern Italy) with care to minimize the parametric dimension. The model was structured as an interface between a distributed MySQL archive of local meteorology and a matrix of calibrated grape compositions. Trento summer variability was related to fall North Atlantic Oscillation (NAO DJFM) but, with low coherence near the mode. Years with moderate temperature variation resulted less predictable than cool or hot ones. In this first essay a model of ‘Chardonnay’ ripening drove harvesting date using as climatic input variable the 1st April to October-end heat sum above 10°C scaled for humidity sensitivity, and as output the ratio between brix degree and the product of pH by malic acid concentration of sampled grapes juice. Surprisingly it was possible to predict in advance of 4-6 weeks the harvest date and the maturation index at harvest with a bootstrap of 95% confidence interval of 1.2 to 2.4 days that amply satisfies our aim. The inter-annual variability dimensions in case is a rather new finding, that although being occurring for presence in the sample of one probable outlier, the 2003, poses basic research questions about the validity of zonations made without meteorological prognoses at seasonal level: these currently exceed the prognostic time of physical atmospheric circulation models and justify empirical models. Every year may produce different performances. This requires procedures with high temporal coherence and reduced dimensions: this module works on 6 parameters and produces reliable estimates of Winkler trend and maturation level, provided a discontinuous series of 10 daily temperatures and a previous calibration of grape composition tables relative to climate. | |
Download Adobe Acrobat Reader (free software to read PDF files) | |
