Assessing the Accuracy of a 2021 Kansas‐Oklahoma Tallgrass Ecosystem Burn Area Map

Systematic spring season burning of the tallgrass prairie in eastern Kansas and northeastern Oklahoma has been used by ranchers and ecologists for over 150 years to encourage forage production and reduce invasive species propagation and woody species encroachment in the Kansas Flint Hills and Oklahoma Osage Tallgrass ecoregion. On average, 8100 km 2 is burned yearly during the March to April burn window. Despite the beneficial effects resulting from spring burning, air resource managers must also account for the particulates emitted from the combustion of these fuels. The Kansas Department of Health and Environment (KDHE) has mapped spring burn area over the Flint Hills (Kansas) and Osage (Oklahoma) regions since 2011 by processing satellite data from the National Aeronautics and Space Administration (NASA) Moderate Resolution Imaging Spectroradiometer (MODIS) sensor onboard both the Aqua and Terra satellite platforms. Understanding the accuracy of these maps is useful for understanding PM2.5 impacts to the local communities and the surrounding states. We conducted both a sample unit–based and an area‐based assessment of the 4 April 2021 burn area map to determine map accuracy and identified potential sources of error, both “commission” (i.e., mapped a burn area where there was no burn) and “omission” (i.e., did not map a burn area where a burn actually occurred). The results indicate that the KDHE 2021 map achieved high overall accuracies (>94%), yet these accuracies were not consistent over the entire area. Omission and commission errors were three to four times larger in areas that had higher burn area percentages, partially due to increased burn edge, promoting larger numbers of mixed 250-m cells. Another source of error was from the prior year (2020) fall burn scars that were occasionally identified as spring burns over the 2021 burn season. The area-based assessment method provided additional information from the sample unit–based method and was useful in identifying anomalous assessment results dependent on the geographical location within the tallgrass system.