We aimed to assess the climate and human factors that explain burnt area in Mexico. The largest burned area occurred in temperate forests, grasslands and hydrophilic vegetation, followed by fires in croplands, tropical forests and shrublands. Burnt area was related to temperature, precipitation, poverty, altitude and distance to water bodies.

We propose a new method to estimate burned area of wildfires. Using fire detections from multiple types of satellites, burned area can be estimated reasonably well when compared with burned area measurements from aircraft. This method worked well for large and small wildfires when tested on a variety of wildfires.

We propose an integrated framework that combines temporal classification and spatial clustering to extract fire footprints. This framework will provide more complete descriptive information than ground-based survey fire records from publicly available remote-sensing-based burn pixel products, which will assist in formulating a targeted and informed wildfire management plan and accurately account for fire-induced carbon emissions.

We developed a fire mapping method that uses active fire data from the MODIS and VIIRS instruments on board satellites to develop a high-resolution wildfire growth database. The satellite-derived fire events show excellent agreement with agency records (R² ≥ 0.95). This method can enable many fire-weather modelling and real-time fire applications.

We have proposed a deep learning model that makes advance predictions of fuel (vegetation) moisture content, a major driver of wildfire ignition and propagation. This proof-of-concept model produces accurate predictions with a 3-month lead time, demonstrating a potentially useful addition to wildfire management tools, facilitating early warning of high wildfire risk.

Stand conditions within a fire-excluded Oregon white oak ecosystem affected foliar moisture content. Encroached stands consistently had higher foliar moisture content, and were associated with stand density and leaf water potential. Better understanding of the role of stand conditions may contribute to improved modelling of potential crown fire behaviour.

Historically, fires were frequent across the upper Great Lakes Region, with regionally widespread, primarily late-season fires occurring in 1689, 1752, 1754, 1791, and 1891. We analysed a fire-scar network, because site-specific analyses can limit interpretations, and found regional drought synchronised fire years, with climate-driven fires likely burning millions of hectares.