Gumbricht, T, McCarthy, TS, McCarthy, J, Roy, D, Frost, PE, Wessels, K (2002). "Remote sensing to detect sub-surface peat fires and peat fire scars in the Okavango Delta, Botswana". SOUTH AFRICAN JOURNAL OF SCIENCE, 98(8-Jul), 351-358.
As a consequence of the constantly shifting water distribution in the Okavango Delta, Botswana, peat accumulated in former wetlands adjacent to receding channels is consumed by fire in successive layers as the peat desiccates. Burning peat has been recorded around the dying Thaoge channel since the mid-19th century. In this study, five different remote sensors were used to detect peat fires in and around the Okavango Delta. ATSR data were used to study the annual cycle from 1999 to 2000, AVHRR data and MODIS data were used to monitor the 2000 fire season, and the MODIS Airborne Simulator and Landsat ETM data were employed for high-resolution studies of single fire events. Peat fires were detected using absolute thresholds in the mid-infrared spectrum, the difference between mid and thermal infrared brightness temperatures, and a threshold in the visible to near-infrared spectrum. Lower thresholds than normally used had to be adopted to detect the fires, which led to an overestimation of fires over other areas. The MODIS Airborne Simulator successfully identified the state of peat fires. Making use of a standard fixed threshold, active-fire detection algorithms, ATSR, AVHRR and MODIS data identified only intense surface fires. Peat fires were distinguished only by using a priori knowledge of either spatial distributions or temporal history. Landsat ETM data could not be used to detect these fires. The peat fires have an annual cycle, with a maximum during the dry winter, coinciding with the surface fire season, and a dormant state during the wet summer. Night-time imagery was necessary to detect sub-surface fires. The present area of peat fires in the Okavango is approximately 100 km(2), which is expected to make negligible contributions to regional trace gas and particulate emissions.