Predicting soil cone index and assessing suitability for wind and solar farm development in using machine learning...

Area of Science:

• Agricultural Science
• Soil Science
• Data Science

Background:

• Soil compaction is a critical factor affecting agricultural productivity and land suitability.
• Accurate prediction of soil compaction is essential for informed land management decisions.
• Existing methods may not fully capture the complex relationships between soil parameters and compaction.

Purpose of the Study:

• To develop and evaluate a novel machine learning approach for predicting soil compaction.
• To identify the most effective artificial intelligence (AI) technique for soil compaction prediction using soil cone index data.
• To demonstrate the practical implications of accurate soil compaction prediction for agriculture and renewable energy projects.

Main Methods:

• Utilized Support Vector Regression (SVR) to process input soil parameters.
• Applied Gradient Boosting (XGBoost), Decision Tree, Artificial Neural Networks, and Adaptive Neuro-Fuzzy Inference System.
• Evaluated model performance using metrics such as mean square error and correlation coefficient.

Main Results:

• The XGBoost model demonstrated superior accuracy and reliability in predicting soil compaction.
• XGBoost achieved a low mean square error and a high correlation coefficient compared to other AI techniques.
• The study successfully integrated SVR with other machine learning models for enhanced prediction.

Conclusions:

• The XGBoost model is highly effective for predicting soil compaction, offering a reliable tool for soil science.
• Accurate soil compaction prediction supports better soil management, enhances agricultural productivity, and informs land suitability evaluations.
• This AI-driven approach provides valuable insights for land use planning, sustainable farming, and renewable energy project assessments.