10.3 Advancements in Convective-Scale Meteorological Data Assimilation: Integrating FY4A AGRI Radiances with the MOTOR-DA System

Wednesday, 31 January 2024: 11:15 AM
Key 9 (Hilton Baltimore Inner Harbor)
Yali Wu, Shenzhen Institute of Meteorological Innovation, Shenzhen, Guangdong, China; and Y. N. Xie, Z. Qin, and J. Chen

The China Meteorological Administration (CMA) Meteorological Object-oriented Tools and Operators Repository - Data Assimilation (MOTOR-DA), an advanced space-and-time multi-scale data assimilation system, has recently been developed and being operational in real time at the Guangdong Meteorological Bureau since April 2023. By assimilating surface and sounding data, the MOTOR-DA system has substantially improved the accuracy of CMA Guangdong regional analysis and forecast using CMA Global Forecast System (CMA-GFS) as background and boundary conditions. What’s more, the regional forecasts initiated from the MOTOR-DA analyses have been demonstrated comparable to those produced by the original operational NWP system, which relies on ECMWF IFS analysis and forecasts for establishing initial and boundary conditions.

In a forthcoming update, the MOTOR-DA system will concentrate on improvement at convective scales by integrating frequently updated remote-sensing observations, such as images from the Fengyun-4 AGRI geostationary satellite. Enabling this capability involves the addition of data pre-processing, quality control, bias correction, and the incorporation of RTTOV v13.0 as the forward/tangent linear/adjoint operator, etc. Besides that, we implement several new features to enhance minimization efficiency and improve analysis quality.

First, we perform a multiscale thinning and superobbing technique prior to incorporating FY4 AGRI radiances into the assimilation process. This strategy ensures improved consistency between the observed and background fields across various grid scales. Moreover, it prevents the background from excessively fitting to dense satellite radiances at the expense of inadequately fitting to sparse conventional observations. Second, we normalize each component of the cost function, effectively mitigating the issue of excessive fitting of the background to densely observed data points. Third, we implement a preconditioning technique for the vertical water vapor profile. This adjustment is crucial to preventing a scenario where the analytical weight on upper levels becomes disproportionately smaller than that on lower levels. It is more robust than the one using relative humidity as the control variable.

Through case studies and three-month experiments, assimilating FY4A AGRI radiances has demonstrated positive impacts on both analysis and forecasts, meaning that operational implementation of the MOTOR-DA satellite radiance assimilation framework is now ready and feasible.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner