Compilation and accuracy assessment of long-term fire products retrieved from satellite data
Emissions of trace gases and aerosol particles from fires have a major share in the global carbon cycle. But large uncertainties exist due to the amount, frequency, and intensity of burning biomass emissions under changing climate. Over the last 150 years, fire records showed a decoupling from fire activities of the main drivers temperature and population density. The most important drivers of this phenomena remain debated. Gaps in our understanding of the relations between fire and climate over the mentioned period in comparison to the previous 2000 years shall be assessed for different regions of the world. Within this setting, the complex systemic linkages between climate, land use, fire, and vegetation will be addressed by the interdisciplinary Sinergia project Paleo fires from high-alpine ice cores. This research framework brings together expertise of ice core analysis and interpretation, fire reconstruction from natural archives, atmospheric modeling of fire tracers, and remote sensing of fire activities. The last mentioned area of research is the topic of this PhD thesis.
Hereby, the aim is to identify single fire events with different size and for different vegetation cover and corresponding observations of burned areas, which are based on available satellite active fire and burned area products as well as on own retrievals from NOAA AVHRR imagery of the last 30 years. This enables the calibration of identified highly-time resolved and well-dated fire reconstructions from the ice core data (since the 1980s) and supports modeling of backward trajectories. For the compilation of the data set of fire events the accuracy in terms of spatial and radiometric resolution of satellite based fire retrievals need to be assessed for different regions of the world with relevance to the ice core records. Therefore, an accuracy assessment will be performed using existent satellite products as well as the application and modification of retrieval methods (e.g. MODIS, AVHRR GAC and LAC data, ASTER, (A)ATSR). As the most suitable satellite fire products are only available since 2000 (MODIS and ASTER), the required time series relies heavily on NOAA AVHRR data. Only the later one provides the needed spectral and spatial resolution back to the 1980s. But reduced capability have to be dealt with (e.g. orbital drift, no on-board calibration of the solar channels, imprecise geocoding). The accuracy assessment of multi-spectral / multi-spatial retrieval of fire history will be one of the main tasks because up to now, no consistent long time series with rigorous validation to assess the aforementioned uncertainties exist. Even considering the processing issues of NOAA AVHRR, it is of high interest to retrieve a homogenous long term data set from one sensor, which provides the possibility to compare the annual fire frequency of different regions of the world for the last 30 years. This data set will be used to investigate the influence of climate change on fire frequency, intensity, and size in temperate, boreal, and tropical regions.