Predicting Corn Water Use with Advanced AI Using Weather and Crop Data

Jim Crocker
4th March, 2024

Predicting Corn Water Use with Advanced AI Using Weather and Crop Data

Image Source: Natural Science News, 2024

Key Findings

  • In Northern China, a new model predicts maize water loss to the air with improved accuracy
  • The model combines weather, soil, and crop data, optimized by AI algorithms
  • The best AI-enhanced model was 2.7-4.8% more accurate than previous methods
Understanding the amount of water that crops lose to the atmosphere, a process known as evapotranspiration (ET), is crucial for farmers to efficiently manage irrigation and conserve water resources. This is particularly important for crops like maize, a staple food for many regions around the world. Traditional methods for estimating maize ET rely heavily on weather data and certain fixed values, but these estimates can be uncertain and may not always reflect real-world conditions accurately. Recent research from Henan University of Science and Technology has made significant strides in improving the accuracy of these estimates[1]. By collecting detailed data from the Yucheng Station in Northern China, the researchers were able to develop a more precise model to predict ET for summer maize. This model doesn't just use standard meteorological data; it also incorporates information about soil moisture and the crop itself, which are critical factors influencing ET. The researchers utilized a back-propagation neural network (BP), a type of artificial intelligence that can learn from data and make predictions. To enhance the model's precision, they applied three nature-inspired optimization algorithms: sand cat swarm optimization (SCSO), hunter-prey optimizer (HPO), and golden jackal optimization (GJO). These algorithms mimic natural processes to find the most accurate model parameters. The results were clear. The hybrid model that combined the BP neural network with the SCSO algorithm (SCSO-BP) outperformed the standalone BP model significantly. The SCSO-BP model improved accuracy by 2.7-4.8% in terms of the coefficient of determination (R2), which measures how well the model explains the observed data. It also reduced the root mean square error (RMSE) by 17.2-25.5%, the mean absolute error (MAE) by 13.9-26.8%, and improved the Nash-Sutcliffe efficiency (NSE) by 3.3-5.6%. These statistics are ways of measuring the difference between the model's predictions and the actual observed values, with lower values indicating a more accurate model. When compared to existing models for maize ET, the SCSO-BP model ranked highest in accuracy, with an R2 of 0.842 and an RMSE of 0.433 mm/day, among other impressive metrics. This level of precision is a significant improvement and provides a valuable tool for calculating daily ET for maize in similar climates across northern China. The findings from this study build upon earlier research that also aimed to refine ET estimation using advanced computational techniques. For instance, a study in the Beas-Sutlej basin of Himachal Pradesh, India, developed multi-layer perceptron artificial neural network (MLP-ANN) models that showed promising results in predicting ET using various meteorological data[2]. Similarly, another study explored the use of support vector regression (SVR) combined with grey wolf optimizer (GWO) to estimate ET in Algeria, also yielding very positive outcomes[3]. These studies, like the one from Henan University of Science and Technology, demonstrate the growing potential of AI and optimization algorithms in enhancing agricultural water management practices. The research from Henan University of Science and Technology not only advances our understanding of ET in maize crops but also showcases the potential of combining different types of data and sophisticated algorithms to achieve more accurate predictions. This approach can help farmers make more informed decisions about irrigation, leading to better water conservation and potentially higher crop yields. The success of the SCSO-BP model also encourages further exploration of hybrid AI models in other agricultural contexts, potentially transforming how we manage and utilize water in farming globally.

AgricultureBiotechPlant Science

References

Main Study

1) Estimating maize evapotranspiration based on hybrid back-propagation neural network models and meteorological, soil, and crop data.

Published 10th January, 2024

https://doi.org/10.1007/s00484-023-02608-y

Related Studies

2) Modelling the reference crop evapotranspiration in the Beas-Sutlej basin (India): an artificial neural network approach based on different combinations of meteorological data.

https://doi.org/10.1007/s10661-022-09812-0

3) Artificial intelligence models versus empirical equations for modeling monthly reference evapotranspiration.

https://doi.org/10.1007/s11356-020-08792-3


Related Articles

An unhandled error has occurred. Reload 🗙