Earth System Science Data (May 2023)

An adapted hourly Himawari-8 fire product for China: principle, methodology and verification

  • J. Chen,
  • Q. Lv,
  • S. Wu,
  • Y. Zeng,
  • M. Li,
  • Z. Chen,
  • E. Zhou,
  • W. Zheng,
  • C. Liu,
  • X. Chen,
  • J. Yang,
  • B. Gao

DOI
https://doi.org/10.5194/essd-15-1911-2023
Journal volume & issue
Vol. 15
pp. 1911 – 1931

Abstract

Read online

Wildfires exert strong influences on the environment, ecology, economy and public security. However, the existing hourly Himawari-8 fire product produced by the Japan Meteorological Agency (JMA) presents large uncertainties and is not suitable for reliable real-time fire monitoring in China. To fill this gap, the National Satellite Meteorological Center (NSMC) proposed an adaptive hourly Himawari-8 fire product for China based on the original Himawari-8 source by employing a dynamical threshold for fire extraction and a database of ground thermal sources. According to the visually extracted reference and consistency check, we found that the NSMC-Himawari-8 fire product effectively removed a majority of false fire alarms included in the original Himawari-8 fire product. Based on a rare field-collected ground reference dataset, we evaluated the reliability of original Himawari-8 and NSMC-Himawari-8 fire products across China. The overall accuracy of the raw Himawari-8 fire product was 54 % and 59 % (not considering the omission errors), respectively. As a comparison, by identifying more real fire pixels and avoiding a majority of false fire alarms, the overall accuracy of the NSMC-Himawari-8 fire product was 80 % and 84 % (not considering the omission errors), respectively, making it an ideal source for improved real-time fire monitoring across China. This research also provides a useful reference for employing a local dataset of underlying surfaces and thermal sources to enhance the accuracy of global fire products in specific regions. The NSMC-Himawari-8 fire product can be downloaded at http://figshare.com (last access: 17 April 2023) with the following DOI: https://doi.org/10.6084/m9.figshare.21550248 (Chen et al., 2022a).