Carbon monoxide concentration forecasting in Santiago, Chile

Area of Science:

• Environmental Science
• Atmospheric Chemistry
• Data Science

Background:

• Santiago, Chile faces air quality challenges, primarily measured by particulate matter (PM10).
• Current air quality models struggle to accurately predict harmful pollution days.
• A strong correlation exists between particulate matter and carbon monoxide (CO) levels.

Purpose of the Study:

• To develop and evaluate a neural network model for forecasting next-day carbon monoxide (CO) concentrations.
• To assess the utility of CO forecasting as a complementary tool for air quality prediction in Santiago.
• To compare the performance of a neural network model against linear regression for CO forecasting.

Main Methods:

• Utilized 3 years of air quality data from a monitoring station in Santiago.
• Developed a neural network model to predict maximum 8-hr moving average CO concentrations.
• Compared neural network model predictions with those from linear regression models.

Main Results:

• The neural network model demonstrated potential for forecasting CO concentrations.
• The neural network approach offers flexibility in parameter adjustment for predictive accuracy.
• Data from the most polluted zone in Santiago was used for model development.

Conclusions:

• Forecasting carbon monoxide (CO) concentrations can enhance existing air quality prediction systems.
• Neural network models show promise for improving air quality forecasts in urban environments.
• Further development of CO forecasting models could aid in managing air pollution in Santiago.