Burned data time series for Tropical Africa (1981–2000) with the Daily Tile NOAA/NASA Pathfinder AVHRR 8‐km Land dataset

Abstract

The NOAA/NASA Pathfinder AVHRR 8‐km Land (PAL) dataset provides an invaluable source of data for researching global changes on the Earth's surface due to its long time and spatial scales. However, because of errors, the original Global Daily PAL data (1981–1994) have been reprocessed into a tiled time series file configuration. Reprocessing involved calculating the correct solar zenith angles and relative azimuth angles, and re‐extracting radiances from the original Channel 1 and 2 reflectance values. The high dependence of the burned area algorithm (BAA) with respect to reflectance channel uncertainty led us, as a first goal, to compare the two dataset releases to determine the magnitude of their differences and how burned area estimations could be affected by them: the new burned data estimations from the Tile PAL release are around 30% higher than those from the original one. The second goal was to present and briefly analyse the monthly burned time series (BS) for Tropical Africa for the period 1981–2000: stational indexes show that the burned pixel peaks in the time series for northern Tropical Africa appear in winter (from December to February), while in southern Tropical Africa they are in summer (from July to September). Chronograms show a significant interannual seasonal variation in both (being more apparent in the southern region) possibly as a consequence of the influence of the El Niño Southern Oscillation (ENSO) phenomenon‐induced Sea Surface Temperature (SST) anomalies in the Pacific Ocean (annual average SST‐3.4 and delayed one year and a half BS anomalies were found in the period considered).