Prediction of harvest start date in highbush blueberry using time series regression models with correlated errors

► The blueberry harvest start date is predicted using heat unit requirements. ► Annual temperature trend models are used to predict heat unit accumulation. ► The prediction error achieved was less than 2 days from the actual start of harvesting. ► All the prediction errors are within the variability of heat unit models. ► These models can be used by blueberry farmers to predict harvest start dates.