A Dynamic Enhancement With Background Reduction Algorithm: Overview and Application to Satellite-Based Dust Storm Detection

Steven D. Miller, Richard L. Bankert, Jeremy E. Solbrig, John M. Forsythe, Yoo-Jeong Noh, Lewis D. Grasso
2017 Journal of Geophysical Research - Atmospheres  
This paper describes a Dynamic Enhancement Background Reduction Algorithm (DEBRA) applicable to multispectral satellite imaging radiometers. DEBRA uses ancillary information about the clear-sky background to reduce false detections of atmospheric parameters in complex scenes. Applied here to the detection of lofted dust, DEBRA enlists a surface emissivity database coupled with a climatological database of surface temperature to approximate the clear-sky equivalent signal for selected
more » ... sed multispectral dust detection tests. This background allows for suppression of false alarms caused by land surface features while retaining some ability to detect dust above those problematic surfaces. The algorithm is applicable to both day and nighttime observations and enables weighted combinations of dust detection tests. The results are provided quantitatively, as a detection confidence factor [0, 1], but are also readily visualized as enhanced imagery. Utilizing the DEBRA confidence factor as a scaling factor in false color red/green/blue imagery enables depiction of the targeted parameter in the context of the local meteorology and topography. In this way, the method holds utility to both automated clients and human analysts alike. Examples of DEBRA performance from notable dust storms and comparisons against other detection methods and independent observations are presented. Plain Language Summary A picture being worth a thousand words is not always a good thing. When a satellite picture of a complex environmental scene contains too much information, it can be hard to make sense of it all. Simple attempts to distill the information into colorful displays to "enhance" a certain feature of can be helpful, but sometimes they can do more harm than good. Problems arise when parts of ourthe image share properties with ourthe feature of interest, masquerading as false alarms and confusing our interpretation. This work attempts to reduce the chances of false alarms happening by accounting for them ahead of time. The Dynamic Enhancement Background Reduction Algorithm, applied in this paper to satellite-detected dust storms, accounts for land surfaces that "look like dust" under nondusty conditions and then adjusts the sensitivity of the detection tests accordingly. The result is a numerical gaugemeasure of our confidence in there being dust present. DEBRA can be communicated as simple, visually intuitive imagery where the only colors involved pertain to the feature of interest---the rest is portrayed as grayscale. The resulting picture may no longer be worth a thousand words, but its added utility to forecasters speaks volumes.
doi:10.1002/2017jd027365 fatcat:kikalthvkrdwfgs7hnucqph3py