Instead of running just a single forecast, the prediction model runs many times from slightly different starting conditions. The complete set of forecasts is called an ensemble forecast, and the individual forecasts within it are called ensemble members. To meteorologists, the key is to generate ensemble members for the ensemble forecast. The singular vector (SV) method can capture the most dynamically unstable perturbations, and they have received a lot of attention in the research and operational communities, especially in producing ensemble members.

However, the SV requires global weather forecast centers to develop a tangent linear model (TLM) and its adjoint model (ADM) for their own weather forecast models, which makes it a "luxury" that is not used by many countries.

Currently, it is mainly used only by the European Centre for Medium-Range Weather Forecasts (ECMWF), the Japan Meteorological Agency, and Australia's Bureau of Meteorology, and fortunately, the operational centers in China are also using it. GRAPES-GEPS is a global ensemble forecasting system developed independently by the China Meteorological Administration. Its initial perturbation method adopts the SV method.

Because of the difficulty of TLM and ADM development, SV calculations start with a simple dry model framework, and then add physical processes one by one. Some scholars have proposed that adding and optimizing the linearized moist physical process in calculating SVs can effectively improve the spatial structure of SVs. At present, the linearized physical process used in GRAPES-GEPS SV is a typical dry physical process.

"In order to obtain the initial perturbation with moister physical information, a moist physical linearization process has been added into GRAPES-GEPS," explains Dr. LIU Juanjuan of the Institute of Atmospheric Physics at the Chinese Academy of Sciences. "For example, a large-scale condensation process."

In her recently published study in Advances in Atmospheric Sciences, she carried out experiments to analyze the structure of the moist SVs from the perspectives of the energy norm, energy spectrum, and vertical structure.

LIU was pleased with the results because the moist SVs can contain more small- and medium-scale information, and improve the short-term weather prediction effectively.

"Our next plan is to add another moist physical linearization process, and examine the performance of the two new moist linearized physical processes along with their impact on the ensemble forecast in GRAPES-GEPS", concludes LIU.

Citation: Wang, J., B. Wang, J. J. Liu, Y. Z. Liu, J. Chen, and Z. H. Huo, 2020: Application and characteristic analysis of the moist singular vector in GRAPES-GEPS. Adv. Atmos. Sci. , 37(11), 1164?1178, https://doi.org/10.1007/s00376-020-0092-9.

Media contact: Ms. LIN Zheng, jennylin@mail.iap.ac.cn