Yield prediction with machine learning algorithms and satellite images

Area of Science:

• Agricultural Science
• Remote Sensing
• Machine Learning

Background:

• Barley yield prediction is vital for global food security.
• Integrating remote sensing, field, and meteorological data offers a promising approach.
• Machine learning models are key for multi-resource data integration in yield estimation.

Purpose of the Study:

• To investigate the impact of different time intervals on barley yield prediction accuracy.
• To develop and evaluate a multi-resource data-based estimation model for barley yield.
• To compare the performance of various machine learning techniques for yield prediction.

Main Methods:

• Dividing the barley growth period into three distinct phases for analysis.
• Developing a model that integrates field, remote sensing, and meteorological data.
• Evaluating four machine learning algorithms, including Gaussian Process Regression.

Main Results:

• Gaussian Process Regression demonstrated superior performance among the tested algorithms.
• The model achieved a coefficient of determination (r²) of 0.84.
• Accurate yield estimation was possible one month prior to harvest, with specific error metrics provided.

Conclusions:

• Yield estimation accuracy is influenced by agricultural zones and temporal training data.
• The developed model shows significant potential as a tool for barley yield prediction.
• Multi-source data combined with machine learning offers a robust approach to agricultural yield forecasting.