Adaptive Observation Strategies based on the Retrospective Optimal Interpolation using the Lorenz-40 variable model

The optimal distribution of observations is important factor to improve the accuracy of the weather forecast. Adaptive observation strategy is the method to reduce the risk of large forecast errors by placing observations in regions where additional observations are expected to improve a forecast of interest (Kim et al., 2004). Several techniques have been proposed to identify optimal sites of observations. There are two types of adaptive observation strategies. One is the ensemble based strategy. The other is the adjoint based strategy.

In this research, we will focus on the adjoint based adaptive observation strategies derived from the Retrospective Optimal Interpolation (ROI). The ROI is a new data assimilation scheme which was derived from the quasi-static variational assimilation (QSVA) algorithm and introduced by Song et al. (2009). The Hessian Singular Vectors (Leutbecher, 2003) and Forecast Sensitivity to Observation (Gelaro et al., 2007) are used as the adaptive observation strategy. To test the accuracy of these methods, experiments are performed with the Lorenz 40-variable model.

Key words: adaptive observation strategy, retrospective optimal interpolation, hessian singular vector, forecast sensitivity to observation, adjoint.