aceCoupling machine learning with a biophysical model for maturity date prediction of apple fruit across China's apple planting regions
Background Accurate prediction of apple fruit maturity date is essential for optimizing harvest timing, fruit quality and market value under climate change. However, process-based crop models often show limited performance when extrapolated across large spatial scales, whereas machine learning models lack physiological interpretability. To address these limitations, this study has developed a hybrid framework integrating the process-based STICS model with machine learning approaches across China's apple planting…
"Wild Apples" by morebyless is licensed under CC BY 2.0. To view a copy of this license, visit https://creativecommons.org/licenses/by/2.0/.
Researchers developed a hybrid framework that couples the STICS biophysical crop model with machine learning to predict apple fruit maturity dates across China. Using phenology records from 24 sites and weather data from 250 stations for 1991-2020, they found a random forest integration improved prediction accuracy and interpretability at regional scales.
Accurate maturity forecasts matter for harvest scheduling, fruit quality and market value as growing conditions shift. The study suggests the hybrid approach can help growers adapt management to climate change, though performance varied by region and extrapolation of process-based models remains challenging.
- Study calibrated and evaluated six machine learning models using phenological observations from 24 sites and meteorological data from 250 stations during 1991-2020.
- Random forest achieved best performance with R2 > 0.65 and SHAP analysis identified chilling requirement, elevation and STICS-simulated maturity date as dominant drivers.
- Authors frame the tool as supporting optimization of harvest management and adaptation of apple production systems to climate change.
Hybrid STICS plus random forest integration improved accuracy and interpretability of apple fruit maturity date prediction to support harvest timing and climate adaptation across China's apple regions.
The rundown
Researchers combined the process-based STICS crop model with machine learning, testing six algorithms on 30 years of data from 24 phenology sites and 250 meteorological stations across China's apple planting regions.
The random forest variant performed best and SHAP analysis highlighted chilling requirement, elevation and the STICS-simulated date as key predictors, with authors reporting improved accuracy but noting regional variability in results.
Sources
- Peer-reviewedJournal of the Science of Food and Agriculture2026-08-01
How should this claim be treated?
The debate