Overcoming Obstacles to Analyzing Large-Scale Smoke Plumes with MOPITT

The MOPITT (Measurements of Pollution in the Troposphere) instrument on the NASA Terra satellite has produced a nearly twenty-year long global record of measurements of tropospheric carbon monoxide (CO) concentrations. MOPITT products have been thoroughly validated and are commonly applied, for example, in the estimation of CO emissions from both fossil fuels and biomass burning. The focus of this research is on issues inhibiting the use of MOPITT retrieval data for characterizing CO distributions within large-scale smoke plumes. In such scenes, MOPITT retrievals face several challenges which can result in observations either being discarded or biased. For example, we show that the operational MOPITT cloud detection algorithm discards potentially useful cloud-free MOPITT observations due to the treatment of strongly polluted scenes by the MODIS cloud mask algorithm. In a case study based on observations of the 2017 wildfires in Northern California, we then show that adapting the MOPITT cloud detection algorithm to be less reliant on the MODIS cloud mask exposes high CO concentrations both near the source regions and in associated smoke plumes extending hundreds of km. In addition, we investigate methods for filtering the MOPITT Level 2 data in order to effectively exclude CO retrievals strongly weighted by the retrieval a priori. Without filtering, such low-quality retrievals tend to mask the retrievals which most clearly delineate the smoke plume.