Sources and implications of bias and uncertainty in a century of US wildfire activity data

Analyses to identify and relate trends in wildfire activity to factors such as climate, population, land use or land cover and wildland fire policy are increasingly popular in the United States. There is a wealth of US wildfire activity data available for such analyses, but users must be aware of inherent reporting biases, inconsistencies and uncertainty in the data in order to maximise the integrity and utility of their work. Data for analysis are generally acquired from archival summary reports of the federal or interagency fire organisations; incident-level wildfire reporting systems of the federal, state and local fire services; and, increasingly, remote-sensing programs. This paper provides an overview of each of these sources and the major reporting biases, inconsistencies and uncertainty within them. Use of national fire reporting systems by state and local fire organisations has been rising in recent decades, providing an improved set of incident-level (documentary) data for all-lands analyses of wildfire activity. A recent effort to compile geospatial documentary fire records for the USA for 1992–2013 has been completed. The resulting dataset has been evaluated for completeness using archival summary reports and includes a linkage to a widely used, remotely sensed wildfire perimeter dataset.