Extended control variable (XCV): skin temperature background correction for the assimilation of clear-sky microwave radiances

Extended control variable (XCV): skin temperature background correction for the assimilation of clear-sky microwave radiances
Technical memorandum
Date Published
Secondary Title
ECMWF Technical Memoranda
Abstract The microwave radiances are among the observations that have an important contribution when it comes to improving the Integrated Forecasting System (IFS) forecast skill. In clear-sky condition, these radiances have frequencies that make them sensitive to both the surface and atmosphere. Using them in the atmospheric analysis requires surface information, and therefore a specific treatment as the surface and atmospheric analyses are currently not strongly coupled.
For instance, the clear-sky microwave radiances observation operator expects a skin temperature value at the observation location and time, together with the profiles of the atmospheric variables along the viewing path. To get a skin temperature as accurate as possible, it was included in atmospheric analysis control vector (TOVS Control Variable or TOVSCV approach). This inclusion allows to adjust the skin temperature during the atmospheric analysis, but only at the observation time and location.
The SKTXCV approach was developed recently as an alternative of the TOVSCV approach with a significant difference that the skin temperature is optimised in model space. So far, the benefits of this new approach were neutral, but experimentation allowed us to highlight a potential biases in our analysis of the microwave-based skin temperature. Among the potential sources of biases that we discuss in this document, we believe that the largest contribution comes from the the skin temperature background. This bias is addressed, in model space, thanks to the SKTXCV approach, in a so-called persistence approach where the analysis of the previous days is used to predict it.
We ran an experiment over a northern hemisphere winter season with a new SKTXCV configuration based on the persistence approach. We compared to it to a reference SKTXCV experiment and a control experiment using the TOVSCV approach. The persistence configuration tends to use more clear-sky observations, and tends to reduce the first-guess departure to these observations.
The medium-range forecast from the persistence configuration is overall improved compared to the two other configurations. The largest improvement are located over the Arctic and to a lesser extend in the northern extra-topics, and for the forecast up to day 5 to day 7 when compared to observations. For example, the improvement of the anomaly correlation of the geopotential at 500 hPa in the Arctic is between 3% to 4% in the first few days.
The correction of the background for the microwave skin temperature fields generally converges in 10 to 15 days, depending on the region and surface. Yet, there is no safeguard to prevent the correction to diverge in this approach. Considering a nudging constrain on the correction or adding a component relative to the correction in the weak constrain formulation are two paths that could be explored next.
URL https://www.ecmwf.int/en/elibrary/81514-extended-control-variable-xcv-skin-temperature-background-correction-assimilation
DOI 10.21957/4bbf1919d5