Spatio-temporal estimation of tree canopy geometry in three-dimensional space (3D) was carried out using a LiDAR scanner in commercial apple production (Malus × domestica Borkh. ‘Gala’/M9) at 55, 85, 115, 140 days after bud break (DABB) for two years. Leaf area (LA) was measured by defoliating trees to calibrate corresponding LiDAR-based points per tree (PPT). Estimation of LA was improved when points of woody parts were removed from PPT, resulting in leave-one-out cross validation adjacent coefficient of determination (R2adj,CV) of 0.92 and root mean squared error (RMSECV) of 4.52%. Spatio-temporal LA was obtained for each tree (n = 4506) showing mean values of 6.25 m2 and 7.15 m2 with high standard deviation of 3.64 m2 and 2.83 m2 in 2018 and 2019, respectively. The growth rate of foliage was calculated with sigmoid growth function quantifying the full development of canopies at DABB105 and DABB95 in 2018 and 2019, respectively. Apparent soil electrical conductivity (ECa) and LA were correlated with fruit size. Accordingly, k-nearest neighbour models were built to predict fruit quality at harvest from first year data, validated on second year data. Based on geoposition and ECa, classification accuracy for fruit size in the test set validation was 48.1%, whereas classification with geopositioning and LA resulted in 67.9% accurate classification. Results highlight the spatio-temporal variation of canopy growth considering a high sample number. Furthermore, results support the future use of LA data instead or in addition to soil data in decision support systems aimed at optimising orchard management practices and, particularly, quantifying the impact of orchard management on fruit size.