Viña Concha y Toro’s Center for Research and Innovation (CRI), since its foundation in 2014, has promoted the concept of industry 4,0 and digital transformation through the Strategic Programme of Intelligent Viticulture Industry.
In 2016, the Agricultural Management of Concha y Toro expressed the need for a more accurate harvest yield forecasting system, compared to those in the industry. As a solution, the CRI proposed to develop a new methodology based on systematic cluster sampling, cluster weight measurement, integration of satellite systems to monitor the variability of vineyard vigor and the application of predictive mathematical models based on machine learning algorithms.
‘The Harvest Volume Forecast project is of great importance to us, as it uses data science and especially techniques in predictive modelling, we are looking to achieve something unique in the wine world, obtaining rates that are much lower than the industry average of 20-30% and achieving the most homogeneous indicator among the company’s wide diversity of assets. The challenge has been a long one, but with promising results,’ explains Eduardo Herrera, Head of Management Control of the Agricultural Management of Viña Concha y Toro.
The yield forecast seeks to estimate, in due time, the amount of kilos of grapes that will be harvested at a specific branch, in the next season. ‘To carry out an early forecast with adequate accuracy (error less than 10% per quarter) and accuracy is critical for harvest operations and this is not trivial, as there is a high spatial and temporal variability in the vineyards of performance components,’ explains José Cuevas, Leader of R&D and Innovation in Engineering of the CRI.
The Center is working on the development and implementation of a digital platform that will allow monitoring key variables for crop yield, applying predictive models, generating reports and establishing recommendations that support the decision-making of the agricultural team and the generation of cost-effective actions. ‘This new platform has included a joint and multidisciplinary work of the CRI, Agricultural Management and IT, together with external companies that provide precision farming technologies.’, highlights José Cuevas.
For the 2019-2020 season, a pilot was carried out in Lourdes Vineyard, located in the Maule Valley, obtaining an average error in the forecast of yield in paints of close to 8% per branch. For the 2020-2021 season, the pilot was extended to eight estates and an area close to 200 hectares, from the Maipo Valley to the Maule Valley. ‘The pilots have the commitment and direct collaboration of the managers of each vineyard, the agricultural operations teams and the technical assistants of the Agricultural Management of Viña Concha y Toro,’ says José Cuevas.
‘Fortunately, the Idahue Vineyard was selected to work on this project, as the significance and scope of the project results make it very attractive to us. Knowing the expected harvest volumes more precisely is not just a number, but it has a tremendous scope in organising and planning any harvest, in all its processes, from the Agricultural Management to Winemaking, and also through a user friendly method,’ points out Ítalo Barrientos, Manager of Idahue Vineyard, located in the Cachapoal Valley.
With the completion of the pilots and the definitive validation of the technologies developed, the CRI expects to have a new performance forecasting system that is cost-effective and scalable.