UNCERTAINTY QUANTIFICATION OF DUST SOURCE MAPPING

Numerous remote sensing studies have attempted to address the complex problem of dust source identification. This has proven to be a challenge, as dust storms on the order of 5 to 100 km rarely originate from a single source, but rather from the amalgamation of several point source dust plumes. These smaller plumes stem from multiple point sources (1 to 10 km across) that can evade coarser-resolution satellite detection. In this presentation, we demonstrate a well-established technique for mapping dust plume point source locations using ArcGIS and Moderate Resolution Imaging Spectroradiometer (MODIS) satellite imagery and quantify the role analyst subjectivity plays in the uncertainty associated with this methodology. Four analysts independently mapped dust plume point sources in Southwest Asia from MODIS imagery collected between 15 May and 31 August for the years 2012 to 2016. The overall mean maximum distance between point sources mapped for the same dust plume and estimates of the amount of error associated with this technique are provided. The influence of additional factors on reproducibility, such as the total number of analysts who subjectively identified individual point sources and analyst confidence levels, will also be explored.