[M-GI26] Data-driven approaches for weather and hydrological predictions

[MGI26-08] Parameter estimation of an atmospheric model using geostationary satellite observation to improve prediction of tropical cyclones: an idealized experiment

*Yuki Hirose¹, Futo Tomizawa¹, Le Duc^1,2, Yohei Sawada^1,2 (1.The University of Tokyo, 2.Meteorological Research Institute, Japan Meteorological Agency)

Keywords:tropical cyclones, parameter estimation, satellite observation

One of the major uncertainties in atmospheric models is the parametric uncertainty. It is important to infer appropriate parameters in various parameterizations from observation. Despite previous efforts on the calibration of parameters in atmospheric models, there is no existing work that calibrates parameters based on satellite data, which is the most important source of observation for tropical cyclones. In this study, we estimate the posterior distribution of parameters using brightness temperature observation from a geostationary satellite. By adopting the Structural Similarity Index (SSIM), an image similarity metric, as the evaluation metric, two parameters from cloud microphysics scheme and one from the boundary layer scheme demonstrated high sensitivity and were successfully estimated. The estimated posterior distribution of parameters not only improves the accuracy of the prediction of satellite images but also partly reduces errors in the prediction of tropical cyclone intensity. We demonstrate the potential of adjusting multiple parameters based on satellite data and the implications of model development to improve the accuracy of tropical cyclone simulations.