8 compared the within-season yield predictive performance of two simulation crop models, one model based on stochastically generated climatic data, and the other on mean climate data. Numerous studies have used simulation crop models for forecasting applications. Simulation crop models make agricultural predictions such as yield, flowering time, and water stress using management, crop cultivar and environmental inputs as well as science-based equations of crop physiology, hydrology and soil C and N cycling 5, 6, 7. Given the strengths of each of these technologies, and that each may favor additional information for making more informed predictions, there may be benefits integrating them to further increase prediction accuracy. These technologies have each provided unique capabilities and significant advancements in the prediction performance, however, they have been mainly studied separately. Introduction Advances in machine learning and simulation crop modeling have created new opportunities to improve prediction in agriculture 1, 2, 3, 4. This result indicates that weather information alone is not sufficient and ML models need more hydrological inputs to make improved yield predictions. Finally, based on feature importance measure, it has been observed that simulated APSIM average drought stress and average water table depth during the growing season are the most important APSIM inputs to ML. Furthermore, we investigated partial inclusion of APSIM features in the ML prediction models and we found that soil and weather-related APSIM variables are most influential on the ML predictions followed by crop-related and phenology-related variables.

The results suggest that adding simulation crop model variables (APSIM) as input features to ML models can make a significant difference in the performance of ML models, and it can boost ML performance by up to 29%. Five ML models and six ensemble models have been designed to address the research question. The main objectives are to explore whether a hybrid approach (crop modeling + ML) would result in better predictions, investigate which combinations of hybrid models provide the most accurate predictions, and determine the features from the crop modeling that are most effective to be integrated with ML for corn yield prediction.

Archontoulis 2, Isaiah Huber 2 1 Department of Industrial and Manufacturing Systems Engineering, Iowa State University, Ames, Iowa, USA 2 Department of Agronomy, Iowa State University, Ames, Iowa, USA * Corresponding author Abstract This study investigates whether coupling crop modeling and machine learning (ML) improves corn yield predictions in the US Corn Belt. Other areas of model improvements are also discussed.1 Coupling Machine Learning and Crop Modeling Improves Crop Yield Prediction in the US Corn Belt Mohsen Shahhosseini 1, Guiping Hu 1*, Sotirios V. Improved modeling of the responses of wheat growth to extreme temperatures during winter and spring periods, and to varying planting densities is needed for better future prediction. These alerted us that the model should be used cautiously to simulate growth and yield over a wide range of sowing dates and planting densities. Additionally, the simulation behaved in a complex manner, and the errors varied greatly with different combinations of sowing dates and planting densities. Simulation of yield achieved poor results with decreased planting density, with average RMSEn of 9-12%, 11-12%, and 16-19% at high, medium, and low density, respectively. However, errors of simulation in phenology and yield were increased with delay in sowing date, with the average absolute root mean square errors of 2 d, 3 d, and 3-4 d in phenology, and the normalized root mean square error (RMSEn) of 7-12%, 11-16%, 16-22% in yield at early, medium, and late sowing dates, respectively. The results show that the APSIM-Wheat model could capture a large part of the variation in phenology, biomass and yield for the same variety across sites. This study presents an evaluation of the APSIM-Wheat model using data from field experiments consisting of three sowing dates, two and three crop varieties and three planting densities in a split-split plot design at three ecological sites from 2008 to 2010 in the North China Plain. However, less information is available on how a wheat model performs in simulation of wheat response to changes in varieties, sowing dates and planting densities across space. Wheat is one of the most important crops in the world, and wheat models have been widely used to study yield responses to changes in management and climate. Evaluation of the APSIM-Wheat model in terms of different cultivars, management regimes and environmental conditions.

Apsim model