The northern high latitude peatlands represent a crucial store of carbon that are being increasingly threatened by climate change and fires. Peat fires represent a major fire phenomenon, which burn prominently by smouldering combustion and emit vast amounts of long-term stored carbon to the atmosphere. Peat fires also emit a suite of aerosols, which degrade air quality and impact upon human health. However, despite the importance of peat fires to atmospheric composition, the carbon cycle, climate and air quality, they are not currently represented in most Earth system models. Additionally, Earth observation products tend to omit peat fires, and consequently little is known about the large-scale impacts of peat fires on atmospheric composition and air quality. Consequently, within this thesis I developed INFERNO-peat, a novel parameterisation of northern high latitude peat fires in the JULES-INFERNO fire model based on the latest laboratory and field studies on peat fires and their dynamics. Here I show that by explicitly incorporating peat fires in a fire model, the spatial representation of burnt area and interannual variability in burnt area and carbon emissions are greatly improved. Additionally, I show that peat fires have major impacts on atmospheric composition and air quality in the northern high latitudes, especially in terms of NO2, CO, and PM2.5 emissions. Finally, I show how under future climate change scenarios, peat fire burnt areas and carbon emissions are expected to increase, particularly under high emissions scenarios, thereby exacerbating the impacts peat fires have on atmospheric composition, climate, the carbon cycle and air quality. Overall, in this thesis I demonstrate the critical need for representing peat fires in our fire models, and the necessity for obtaining accurate peat fire observations in order to better quantify their impacts on the climate system.