Enhancing maize grain dry-down predictive models

• Published in: Agricultural and forest meteorology 2023-05, Vol.334, p.109427, Article 109427
• Main Authors: Chazarreta, Yésica D., Carcedo, Ana J.P., Alvarez Prado, Santiago, Massigoge, Ignacio, Amas, Juan I., Fernandez, Javier A., Ciampitti, Ignacio A., Otegui, Maria E.
• Format: Article
• Language: English
• Subjects: Argentina; corn; data collection; forests; grain yield; harvest date; Kansas; Kernel moisture; meteorology; Post-maturity drying; seeds; solar radiation; Sowing date; Zea mays; Zea mays L
• ISSN: 0168-1923
• DOI: 10.1016/j.agrformet.2023.109427

•Previous models considered post-maturity grain dry-down coefficient (k) as constant.•Solar radiation and vapor pressure deficit are important to model k.•New models to estimate kernel moisture loss and harvest maturity were developed.•New dry-down models predicted time to harvest maturity with an error of 2 to 3 days.•Models were trained with Argentina data and tested with Argentina and USA data. Predicting the optimal harvest date after crop physiological maturity is highly relevant for maize (Zea mays L.). While harvesting before achieving the commercial kernel moisture implies additional costs of grain drying, a delayed harvest of maize crops is linked to grain yield and quality losses. The main objective of this work was to identify weather variables affecting the post-maturity grain dry-down coefficient (k) in order to develop models to predict kernel moisture loss and time to harvest (harvest readiness) under a wide range of sowing date environments. Kernel moisture datasets from field experiments in Pergamino (Argentina) and Kansas (US) were used for training and testing post-maturity grain dry-down models. Two k coefficients were defined based on the solar radiation and the VPD explored during the pre- and post-maturity period (kpre and kpost). Models including kpre and kpost were tested under a wide range of sowing date environments, presenting high accuracy in predicting kernel moisture (R2 ∼ 0.80; RRMSE ∼ 0.15) and harvest readiness (R2 = 0.99; RRMSE ∼ 0.05). This study provides the foundation for developing an interactive digital platform to estimate harvest time to assist farmers and agronomists with this critical decision.