Investigating the Correlation between Background Radiation and Weather Conditions: A Predictive Model Employing Machine Learning Techniques.

Abstract

Ensuring public safety and protecting environmental health rely heavily on dependable radiation monitoring systems. Accurately predicting variations in background radiation levels is paramount, particularly in environments where natural factors such as weather significantly influence these changes. Background radiation is known to be influenced by weather conditions such as temperature, humidity, atmospheric pressure, wind speed, and rainfall. These factors interact in intricate ways, making it challenging to discern natural fluctuations from potential radiation hazards. This study investigates the connection between background radiation and weather by analyzing data collected over three months at Oregon State University. Utilizing advanced machine learning techniques, particularly Gradient Boosted Regression Trees (GBRT), a predictive model was developed to estimate background radiation levels. Its performance was compared to a simpler linear regression model to illustrate the advantages of sophisticated techniques in comprehending intricate relationships. The results demonstrated that the GBRT model exhibited superior performance compared with the linear regression model in predicting background radiation levels. The GBRT model effectively captured the nonlinear relationships and interactions among the variables, resulting in enhanced accuracy and improved generalization. Key findings indicated that humidity and temperature exhibited consistent and significant correlations with radiation count rates, while factors such as rainfall and wind speed exhibited limited effects. Despite some challenges in predicting extreme values, the GBRT model demonstrated a robust ability to minimize prediction errors and effectively manage noisy data. This research underscores the significance of integrating environmental data into predictive frameworks and employing sophisticated machine learning algorithms to enhance the accuracy of radiation monitoring systems.