Wildfires needs three components to occur: the right weather, abundance of inflammable objects, and a spark. As the planet continues to demonstrate even more clear signs from the global warming, this combination will happen more frequently. There are two opposite scenarios that demonstrate how and why this phenomena is occurring with more intensity:

• Wet and densely forested ecosystem has a large amount of fuel. A single wet season is irrelevant for the increase of the fuel and the soil moisture limits the risks for wildfires. However, as the dry seasons continue to get drier and hotter, soil moisture is set to diminish. Having this abundance of inflammable objects and the right climate conditions, all it takes to a fire to start is a spark. A recent study made by the Colorado University shows that 59% of the wildland fires were started by human activity and this number is even higher in areas with human population. Furthermore, with the advance of global warming, lightning strikes are becoming even more frequent, increasing the risk of natural sparks as well.
• Relatively dry ecosystems dominated by grass and low-density shrub offers a low amount of fuel and this limitates the spread of large fires. Should these ecosystems receive an above normal precipitation, the additional growth of the vegetation will increase the inflammable material and the excess moisture will be quickly dried out in the dry seasons. The spark risks are the same as mentioned before.

Wide-scale wildfires damage analysis should consider a wide array of variables, not only the burned area. Multi-dimensional metrics (such as carbon emissions, human lives lost, direct economic losses, potential hazards in air and water quality), that identifies the areas that have the most risk is essential for governments to promote landscapes that are more fire resilient and minimize risks for human populations and infrastructure.

As the fire seasons continues to get bigger and longer, firefighting needs to be more efficient. Fire suppression has already proven itself to have a poor efficiency. It is vital that the governments target their efforts in identifying the most susceptible areas to fire, taking steps to reduce the damage caused by the fire season.

The Capybara project aims to firstly identify which areas are in more danger based on historical satellite imagery and weather data. Then classify the level of danger these areas represent based not only on the area that was burned, but also on the climate effects for the region. Furthermore, in the future will be possible to classify the economical and humane damages. The project consists in creating an API to provide our analysis and a Power BI dashboard that consumes the analysis and enables researchers and policy makers to manipulate the data the way they desire.

This project has the goal to create a machine learning model that identifies which areas are more susceptible to wildfires, based on historical weather and fire data. Then classify the damage these fires caused to predict which area will eventually cause more damage in the future. As a result, policy makers would be able to focus on the mitigation, prevention and preparation for the fire season, by acting in the areas that are most critical for the firefighting. Finally, capybaras would be able to live worry free on the great brazilian wetlands.