Sea-ice concentration analysis in model space in IFS cycle Cy49r1

Title
Sea-ice concentration analysis in model space in IFS cycle Cy49r1
Technical memorandum
Date Published
10/2025
Secondary Title
Technical Memoranda
Number
932
Author
Sebastien Massart
Alan Geer
Publisher
ECMWF
Abstract Recent developments, implemented in operations in November 2024, allowed surface-sensitive microwave observations to be assimilated over sea ice in the atmospheric 4D-Var of the European Centre for Medium-range Weather Forecasts (ECMWF). Those developments were based on the framework of the observation-space auxiliary control variables (TOVSCV) that extends the control vector with observation-space variables at observation locations. They provide an estimation of both the sea-ice concentration (SIC) and some additional empirical variables that describe the physical state of the sea ice.
In this paper, we describe an alternative framework that extends the Integrated Forecasting System (IFS) control vector with a model-space variable that represents the sea-ice concentration. We compare the two approaches over two periods of 3 months each, one in Boreal and one in Austral summer, while using the sea ice sensitive radiances for the Advanced Microwave Scanning Radiometer 2 (AMSR2) and the GPM Microwave Imager (GMI).
The two approaches provide an analysis with similar values for the sea ice concentration, with differences between 2-3% in mean value. One important difference between the two approaches is the method used to compute background values of sea ice concentration used for cycling the assimilation system. In the observation-space framework, the background values are derived from the dynamically retrieved surface emissivity in the 10 GHz channel for each observation. In the model-space framework, the analysis sea ice field from the previous analysis update is used as the background. Using this approach, the first-guess departure for the both AMSR2 and GMI is generally less biased, but presents a larger standard deviation. Despite this, the forecast scores are neutral to sightly positive for the model-space approach compared to the observation-space one.
URL https://www.ecmwf.int/en/elibrary/81691-sea-ice-concentration-analysis-model-space-ifs-cycle-cy49r1
DOI 10.21957/4bad006515
Download citation