Satellite Remote Sensing of Potassium Emission Lines in Wildfire Spectra

Fire phase strongly impacts the species emitted by a fire but is difficult to estimate using remote sensing in the thermal infrared. Including visible and near-infrared (NIR) channels to measure the Visible Emission Fraction (VEF) of a fire allows accurate estimation of fire phase and emissions but, due to the relative weakness of visible and NIR emissions from fires compared to the sun, is only possible for nighttime measurements. However, the spectrum of burning biomass also contains prominent potassium emission lines at 767 and 770 nm which are only emitted during flaming combustion. These emission lines also have a much higher spectral radiance than the surrounding NIR blackbody emissions, increasing the sensitivity of instruments with sufficiently high spectral resolution to fire measurements in the NIR. This allows potassium emission lines from flaming fires to be detected from satellite instruments during the day, which is not possible for the surrounding visible and NIR signal. Using this property of potassium emission lines, a VEF-like product could potentially be developed that allows daytime estimation of fire phase and emissions. We show that potassium emission lines can be reliably detected under low-light conditions using TROPOMI and using daytime measurements with OCO-2, that the potassium spectral signal can be effectively isolated from solar radiation, atmospheric absorption, and other background signals using Principal Component Analysis (PCA) and that the isolated signal correlates well with Fire Radiative Power (FRP).