Quantifying the analysis uncertainty for nowcasting application
Abstract. This study proposes a method to quantify the uncertainty of the error in the very high–resolution analysis in near–surface level for nowcasting application. We perturbed the first guess field and observation with Gaussian–distributed perturbations, which have variance equal to that of first guess error. The first guess error not only reflects the spatial characteristic of the difference between the first guess field and observation but also dominates the major uncertainty information of analysis errors. It is important to consider the attenuation of uncertainty dispersion caused by interpolation. Gaussian perturbations are combined with an inflation factor to estimate the attenuation of perturbation dispersion. To assess this method, it was applied to high–resolution analysis and nowcasting in the Beijing–Tianjin–Hebei region for hourly temperature, humidity and wind components. To evaluate the transmission of perturbation information in the nowcasting extrapolation, the ensemble analysis is used to compute ensemble nowcasting. The verifications show that the ensemble analysis has reasonable spread and high reliability, demonstrating effective and accurate quantification of the analysis error uncertainty. Verifications of ensemble nowcasting illustrate that the ensemble spread has effective growth within the nowcast extrapolation up to a lead time of 2 hours, but highly depends on the trend of NWP. The results prove that the propagation of analysis uncertainty representation in nowcasting extrapolation can match to the error increment, beneficial for estimating the near–surface nowcasting uncertainty.