Published works

Accuracy is an important aspect of fire hotspot detection.  Errors in the H8-AHI WFABBA fire hotspot detection can lead to a loss of trust in a fire hotspot detection product.  Compared to MODIS and VIIRS polar-orbiting satellites hotspot detections, the WF-ABBA hotspot detection product over Australia 0400 UTC had minimum commission error rates of 31% (Winter), 35% (Summer), 48% (Spring) and 54% Autumn over 1 Dec 2015 – 30 Nov 2016.  The WF-ABBA algorithm was originally developed for America and was not tuned specifically for the Australian continent.  Here we create a new Himawari fire-hotspot algorithm tuned dynamically for 419 Australian bioregions.  The new algorithm 0400 UTC had minimum commission error rates, in comparison to MODIS/VIIR hotspot detections, of 6% (Summer), 8% (Spring), 20% (Winter) and 41% (Autumn) over 1 Dec 2015 – 30 Nov 2016.  These commission rates are a considerable improvement over the currently available (WF-ABBA) fire hotspot product.