STRAWBERRY FRUIT YIELD FORECAST BASED ON MONTECARLO METHODOLOGY AND ARTIFICIAL VISION

Abstract

The region of Huelva produces most of the strawberry export to Europe during winter season. Most farmers use the same planting date and strawberry cultivar, which leads to identical fruit yield curves or trends (upon weather changes). Producers negotiate a price with clients depending on fruit volume availability. Having a 3- to 4-days forecast on fruit harvest yield is thus a key negotiation tool to set fruit selling price. During the 2010-2011 campaign the Adesva R&D Center has carried out experiments to develop and validate a forecast equation on fruit harvest yield. The equation and forecast is based on Montecarlo (stochastic simulation) methodology. In order to build a forecast many potential parameters were taken into account such as: weather (temperature, humidity, radiation) forecast, previous weather, plant accumulated yield, and most important: fruits hanging on the plant with given maturity level (detected via simple artificial vision device). Successful correlation and forecast were achieved with this methodology. The accuracy of forecast achieved with this method was good enough to ensure price negotiation with success. This work leads to a wide world of applications in the field of precise agriculture. Similar correlations can be made now with fruit size or yield to irrigation or nutrition parameters.