On the outstanding need for a long-term, multi-decadal, validated and quality assessed record of global burned area: Caution in the use of Advanced Very High Resolution Radiometer data

Abstract

A long-term, multi-decadal, record of global burned area is needed to understand the interplay between human activity and climate change on fire occurrence and behavior. At present, the longest consistent satellite-based record has been provided by the Moderate Resolution Imaging Spectroradiometer (MODIS) on-board NASA’s Terra and Aqua satellites. A long-standing objective within the remote sensing community has been to compile a long-term record of pre-MODIS global fire activity using data acquired with the older Advanced Very High Resolution Radiometer (AVHRR). The recent production and release of one such new data set – the AVHRR-based FireCCILT10 burned area product generated under the auspices of ESA’s Climate Change Initiative (CCI) – warrants a fresh consideration of the practical issues confounding the use of AVHRR observations. Here we compare the new 36-year FireCCILT10 product with the AVHRR Long Term Data Record (LTDR) input data used to generate the data set. Despite its recent provenance, we found significant temporal artifacts in the first half, and in the last year, of the FireCCILT10 burned area record, that we show are associated with the orbit drift of the NOAA polar orbiting satellites on which the AVHRR sensor suite resides. Based on these findings, and having noted similar but previously unacknowledged artifacts in two older long-term AVHRR burned area products, we formulate several practical recommendations with respect to use of the FireCCILT10 data set and the development of future long-term AVHRR-based fire data sets. The paper findings reinforce the ongoing need for a long-term, multi-decadal, validated and quality assessed, record of global burned area.