Improving weather forecasting with a new IASI channel selection method
With the advent of satellite observation techniques and improvements in data assimilation schemes, the initial state in an NWP (numerical weather prediction) model has become more realistic, which is fast becoming the most vital part in the process. Furthermore, among the many available satellite observations, infrared hyperspectral measurements are known to have the greatest impact on weather forecasting.
In a new study published in Advances in Atmospheric Sciences, an attempt was made to select hyperspectral sounder IASI (Infrared Atmospheric Sounding Interferometer) channels from the 314 channels provided by EUMETSAT (European Organisation for the Exploitation of Meteorological Satellites) for data assimilation in the UK Met Office (UKMO) Unified Model (UM), using a one-dimensional variational analysis (1D-Var). The channel selection was performed by considering the degree of improvement in retrieved atmospheric parameters from 1D-Var over the background atmospheric parameters, using the Channel Score Index (CSI) as a measure of success. In the UM, used operationally by not only the UKMO but also the KMA (Korea Meteorological Administration), IASI measurements have been assimilated since 183 channels were subjectively selected in 2007.
Instead of the currently used 183 channels, 200 newly selected IASI channels, including substantially different H2O and shortwave infrared channels, were used for the UM data assimilation. From the two trial runs using the UKMO UM data assimilation system, it was noted that the new IASI channels gave an overall neutral impact in terms of the NWP index based on parameters such as 500-hPa geopotential. However, experiments resulted in a significant bias reduction in the relative humidity forecasts, in particular over the upper-troposphere layer from 500 hPa to 200 hPa, which was attributed to additional H2O channels in the new IASI channels.