Wildland fires are a type of vegetation fire that burn in a rural or wild landscape and affect many countries worldwide. They are an important mechanism in ecosystem maintenance, although in certain cases wildland fires can adversely affect both people and the environment. A wildland fire can interact with the surrounding topography, vegetation and weather in a complex manner, which makes microscale prediction of wildland fire behaviour difficult in many situations. This thesis focused on the application of the Weather Research and Forecast (WRF) numerical weather prediction (NWP) and WRF-Fire coupled atmosphere-fire models to investigating aspects of atmospheric interactions with wildland fire. The research covered a wide range of atmospheric scales, from a seasonal mesoscale analysis of fire weather conditions across New Zealand to a microscale analysis of complex atmosphere-fire interactions over idealised terrain. The first study investigated the suitability of WRF modelling of fire weather conditions for the 2009/10 wildland fire season in New Zealand. The WRF model horizontal grid spacing was 8 km and the model output was directly compared with near-surface fire weather conditions meaured and derived at 23 weather stations located throughout New Zealand. The analysis considered the air temperature, relative humidity, wind conditions, rainfall and the New Zealand Fire Weather Index (FWI) and Continuous Haines Index (CHI) on observed high-end fire weather days. WRF typically underpredicted the air temperatures and relative humidities, whereas it typically overpredicted the wind speeds, CHI and the number of high-end fire weather days. WRF was assessed to be unsuitable for accurately modelling particular aspects of fire weather, such as the wind speed and direction, in mountainous terrain and near complex coastlines. Further research is needed to investigate how varying the horizontal resolution in WRF affects the assessed accuracy of modelled fire weather conditions. The second study investigated the behaviour of the Haines Index (HI), CHI and FWI, and their associated atmospheric properties for the 2009/10 wildland fire season in New Zealand. The analysis demonstrated that there was a large degree of spatial variability in fire weather conditions throughout New Zealand, particularly in or near mountainous terrain. The fire weather severity was highest in the eastern South Island and appeared to be closely associated with mesoscale atmospheric processes over mountainous terrain, although the relationship between these atmospheric processes and fire weather condi- tions requires further investigation. The HI and CHI were both limited in their utility at measuring aloft fire weather conditions in high altitude regions. Finally, the fire weather conditions associated with the 36 largest wildland fires of the fire season were evaluated, although no statistical relationships were found between the wildland fire size and either the CHI or FWI. The third study investigated the fire weather conditions across the South Island associated with an extreme foehn event on 6 February 2011. Mountain waves developed in the northwesterly synoptic flow over the Southern Alps and were found to directly influence the fire weather conditions near the surface and aloft in the lee of the mountains. A hydraulic jump along the foothills of the Canterbury Plains resulted in a downslope windstorm with wind speeds exceeding 80 km/h. Further south, large amplitude mountain lee waves directly influenced the near-surface wind speeds and atmospheric stability aloft. The foehn winds were associated with peak air temperatures over 35˚C in the eastern South Island, which are significantly higher than the climatological average. The FWI indicated widespread extreme near-surface fire weather conditions in the lee of the mountains. The subsequent passge of a cold front on 7 February brought a marked reduction in fire weather severity across the South Island. The fourth study investigated atypical wildland fire behaviour on steep leeward slopes through a series of idealised WRF-Fire simulations. The analysis considered both the leeward flow characteristics over a triangular ridge line and the fire spread from an ignition point at the base of the leeward slope. The fire spread was modelled for two different fuel types and with two-way atmosphere-fire coupling both enabled and disabled. The modelled fire spread in the heavy fuel type with coupling enabled closely resembled the fire channelling wildland fire behaviour phenomenon. The initial fire spread was initially dominated by upslope fire spread to the mountain ridge line at an average rate of around 2.0 km/h. This was followed by a phase of intermittent rapid lateral fire spread close to the ridge line at a maximum rate of around 3.6 km/h. The intermittent rapid lateral fire spread was driven by strongly circulating horizontal near-surface winds that were associated with updraft-downdraft interfaces. These updraft-downdraft interfaces formed due to an interaction between the strong pyro-convection and terrain-modified winds. The presented research collectively demonstrated the versatility and effectiveness of NWP and coupled atmosphere-fire modelling for studying various aspects of atmospheric interactions with wildland fire. The research further highlighted the effects of atmospheric processes over complex terrain on fire weather conditions and wildland fire behaviour. Although three of the studies in the thesis had a regional focus on New Zealand, the research outcomes should benefit end users in fire management worldwide.