Fighting Climate Change with Machine Learning

Queretaro, Mexico

The concern of rising pollution levels is one that’s shared globally, and one city in Mexico is leveraging AI to find a solution. The Office of Sustainable Development in Querétaro, Mexico has partnered with Valkyrie Intelligence to develop a predictive model powered by Machine Learning for forecasting air pollution levels in its capital city.  

Santiago de Querétaro, known for its rapid urbanization and industrial growth, is familiar with the challenge of air pollution. Recognizing the urgency to mitigate these issues, the Office of Sustainable Development has taken a proactive approach by turning to data science in search of a solution to regulate pollution and help keep citizens safe from its harmful effects. The city of Querétaro has been measuring pollutant contamination levels for years using their Calidad del Aire Querétaro webpage (CeMCAQ), but until now, they did not have the capability to predict future pollution levels. By using a complex machine learning model that takes into account environmental, weather, and seasonal factors combined with historical data, government officials are able to forecast pollution levels days in advance with a high degree of accuracy. This is valuable information for the Office of Sustainable Development, as this reduces unnecessary restrictions imposed on businesses and citizens when air pollution is above certain safety thresholds.  

The Office of Sustainable Development currently monitors real-time air quality from six measuring stations across the city of Santiago de Querétaro. Based on the real-time contamination levels, officials determine if certain protocols should be enforced to mitigate pollution and protect citizens. These protocols can include school closures or delays, factory/industry closures, outdoor activity restrictions, and road closures. Basing these decisions on a science-backed predictive model provides more accuracy and assurance for government officials in enacting safety measures, reducing risks associated with not taking action as well as those associated with enforcing protocols unnecessarily.   

Leveraging AI and ML Predictive Power 

Valkyrie’s model, nicknamed Cirrus, is an air pollution forecasting tool that uses machine learning to predict contamination levels of harmful pollutants based on historical & real-time pollution levels, as well as weather, calendar, and seasonal data. 

The predictive model, developed by Valkyrie Intelligence, leverages historical and real-time data from various sources such as weather conditions, industrial activity, and seasonality. Machine learning algorithms analyze these complex datasets, identifying correlations and patterns that humans may miss. Powered by Machine Learning, Cirrus uses stacked ensemble modeling with a combination of machine learning algorithms, including time series cross-validation and cyclic transformations of date features. By considering multiple variables simultaneously, the model generates forecasts of pollution levels with a higher degree of accuracy than human instinct alone. All of these data insights, as well as comparisons of historical predicted vs actual values are displayed in an interactive and user-friendly dashboard tool for use by SEDESU officials.

An image of the AI-powered front end dashboard Valkyrie created to help fight climate change in Mexico with Machine Learning

Screenshot of Cirrus with city measuring stations and pollution level predictions across Santiago de Querétaro

There are several benefits this model can offer state officials in Querétaro. With higher accuracy of forecasted air pollution, government officials can reduce unnecessary school and industry closures, which are some of the common contingencies enacted. This also adds tangible value for private businesses and factories that experience revenue loss due to closures enacted to keep pollution levels at bay. On the flip side, knowing highly-accurate forecasted pollution levels in advance assists officials in ensuring contingencies are enacted when they are necessary, and allows for timely decision-making and communication.  

Another major benefit the forecasting model provides is explainability of the factors that most affect air pollution in Santiago de Querétaro. By analyzing historical data, the model has learned how different factors such as seasonality, day of the week, holidays, and various weather conditions affect air quality. Not only that, but the model can provide insight on each of those factors individually as well as combined, providing more depth of understanding into the role each factor plays in air quality. For example, trends showed air quality spiked during winter months and tapered off in the summertime. This knowledge of seasonality, combined with a given day’s weather conditions and the presence of a holiday, all give more detail and higher accuracy in predicting that day’s air quality.    

Queretaro, Mexico

What This Means for The Future of AI in Latin America 

This initiative sets a precedent for the adoption of AI technology in Latin America, and states across Mexico are coming together to explore novel technologies for research and decision-making that impact the population’s welfare. This collective effort reflects a shared commitment to developing cutting-edge solutions for sustainable development that support citizen wellbeing. 

The collaboration between SEDESU and Valkyrie marks a significant milestone in harnessing the potential of AI to address environmental issues. By leveraging advanced AI models, we can surpass the limitations of human intuition and achieve a higher level of precision in predicting and strategizing against pollution. Beyond just climate change, machine learning’s predictive capabilities can be applied to numerous other practical issues, business problems and social/political topics. With its superior monitoring, predicting, and analyzing abilities, AI holds tremendous promise for creating an entirely new landscape for future generations with almost infinite opportunities for application. 


515 Congress Ave. Austin, TX 78701 Suite #1425



(512) 947 – 6472

Stay in touch

Stay Updated
This field is for validation purposes and should be left unchanged.