Harvest date estimation of ‘Gala’ apples based on environment temperature using artificial intelligence

Abstract

Agroclimatic variables in different time windows were analyzed using Artificial Intelligence techniques to estimate the fruit growing season extension and harvest start date for ‘Gala’ apples (Malus domestica (Suckow) Borkh.) Meteorology and phenology data were collected from five orchards in Central Chile, between 2004 and 2019. The attributes derived from air temperature during the first days of fruit growing season showed the high relationship with harvest start date: The number of hours below 18 °C from full bloom to 35 d after (R = 0.9) and growing degree hours accumulated from full bloom to 45 d (R = -0.84). Different models were developed with these attributes. Simple and multiple linear regression models were the most accurate for explain the length of the total fruit growth period until harvest. The 35 d after full bloom time window was the most effective, with an R2 = 0.82, for estimating harvest start date of ‘Gala’ apples. These results contribute to the apple growers demand to schedule fruit harvest and processing, especially in a climate change scenario.