Comment on amt-2022-166

Abstract

Uncertainty-bounded satellite retrievals of volcanic ash cloud properties such as ash cloud-top height, effective radius, optical depth and mass loading are needed for the robust quantitative assessment required to warn aviation of potential hazards. Moreover, there is an imperative to improve quantitative ash cloud estimation due to the planned move towards quantitative ash concentration forecasts by the Volcanic Ash Advisory Centers. Here we apply the Optimal Retrieval of Aerosol and Cloud (ORAC) algorithm to Advanced Himawari Imager (AHI) measurements of the ash clouds produced by the June 2019 Raikoke (Russia) eruption. The ORAC algorithm uses optimal estimation to consolidate a priori information, satellite measurements and associated uncertainties into uncertainty-bounded estimates of the desired state variables. Using ORAC, we demonstrate several improvements in thermal infrared volcanic ash retrievals applied to broadband imagers. These include: an improved treatment of measurement noise, accounting for multi-layer cloud scenarios, distinguishing between heights in the troposphere and stratosphere, and the retrieval of a wider range of effective radii sizes than existing techniques by exploiting information from the 10.4 μm channel. Our results indicate that 0.73 ± 0.40 Tg of very fine ash (≤ 15 μm, radius) was injected into the troposphere and stratosphere during the main eruptive period from 21 June 18:00 UTC to 22 June 10:00 UTC. The total mass of very fine ash decreased from 0.73 Tg to 0.10 Tg over ~48 h with an e-folding time of 20 h. We also estimate a distal fine ash mass fraction of 0.47 ± 0.3 % based on the total mass of very fine ash retrieved and the ORAC-derived height time-series. Several distinct ash layers were revealed by the ORAC height retrievals. Generally, ash in the troposphere was composed of larger particles than ash present in the stratosphere. We also find that our implementation of the ORAC algorithm was reliable out to four days and was able to track the median ash cloud at concentrations below peak ash concentration safety limits (< 4 mg m-3) if typical ash cloud geometric thicknesses were assumed. The ORAC height retrievals for the Raikoke case study have a bias and precision of -2.22 km and 2.85 km, respectively, based on comparisons with CALIOP and GOES-17 height validation data. The dataset generated here provides uncertainties at the pixel level for all retrieved variables and could potentially be used for dispersion model validation or implemented in data assimilation schemes. Future work should focus on improving ash detection, improving height estimation in the stratosphere and exploring the added benefit of visible channels for retrieving effective radius and optical depth in opaque regions of nascent ash plumes.