Giesco
RESEARCH ON SAMPLING METHODOLOGIES FOR VINEYARD YIELD FORECASTING IN DIFFERENT CULTIVARS AND TRELLIS SYSTEMS INVESTIGACIÓN DE METODOLOGÍAS DE MUESTREO PARA EL PRONOSTICO DEL RENDIMIENTO DEL VIÑEDO EN DIFERENTES CULTIVARES Y SISTEMAS DE CONDUCCIÓN
OYARZUN KUSCHEL, Carlos1; ARMINGOL, Erick2; CUEVAS VALENZUELA, José2; GONZÁLEZ ROJAS, Álvaro2* ¹Ecole Supérieure d’Agriculture, Angers, France Center for Research and Innovation, Viña Concha y Toro, Chile 2 * Corresponding author: alvaro.gonzalez@conchaytoro.cl
Abstract: Vineyard yield forecasting is a key issue for vintage scheduling and for optimizing vineyard and winery resources. Errors in yield forecasting are due to high spatial variability present in the vineyard and in some cases to the absence of an adequate sampling method. This project aims to identify the main sources of variability and error in yield prediction, in order to increase the accuracy of the forecast and to define a sampling methodology. In Maule Valley, 7 blocks (30 ha in total) of different cultivars, trellis systems and grape-quality classification were selected. The number of experimental units (126 in total) was based on the total plants per block. Bunch counts were performed at flowering, fruit set, veraison and harvest, and evolution curves of bunch and berry weight were built. Bunch number was the main source of variability and berry weight the least variable parameter. No significant difference in bunch weight at veraison and harvest was found in any of the studied blocks, but a significant difference in berry weight at veraison and harvest was found in block 3 (Syrah clone 470), with a variation of -9%. Models built using field bunch and berry weight measurements at veraison (models V-2 and V-3) presented better performance in almost all blocks compared to models using historical data (FS-1, FS-2 and V-1).
Keywords: yield prediction; spatial variability; yield components; correction factor; model performance
Abstract: Vineyard yield forecasting is a key issue for vintage scheduling and for optimizing vineyard and winery resources. Errors in yield forecasting are due to high spatial variability present in the vineyard and in some cases to the absence of an adequate sampling method. This project aims to identify the main sources of variability and error in yield prediction, in order to increase the accuracy of the forecast and to define a sampling methodology. In Maule Valley, 7 blocks (30 ha in total) of different cultivars, trellis systems and grape-quality classification were selected. The number of experimental units (126 in total) was based on the total plants per block. Bunch counts were performed at flowering, fruit set, veraison and harvest, and evolution curves of bunch and berry weight were built. Bunch number was the main source of variability and berry weight the least variable parameter. No significant difference in bunch weight at veraison and harvest was found in any of the studied blocks, but a significant difference in berry weight at veraison and harvest was found in block 3 (Syrah clone 470), with a variation of -9%. Models built using field bunch and berry weight measurements at veraison (models V-2 and V-3) presented better performance in almost all blocks compared to models using historical data (FS-1, FS-2 and V-1).
Keywords: yield prediction; spatial variability; yield components; correction factor; model performance