Guo-Yuan Lien

The FORMOSAT-7/COSMIC-2 satellites were launched in 2019, which can provide considerably larger amounts of radio occultation (RO) observations than the FORMOSAT-3/COSMIC satellites. The radio signals emitted from the global navigation satellites system (GNSS) are received by these low Earth orbit (LEO) satellites to provide the so-called bending angle accounting for bending of the rays after penetrating through the atmosphere. Deeper RO observations can be retrieved from FORMOSAT-7/COSMIC-2 for use in RO data assimilation to improve forecasts of tropical cyclones. This study used the global model FV3GFS with the finest grid resolution of about 25 km to simulate five selected typhoons over the western North Pacific, including Hagibis in 2019, Maysak and Haishen in 2020, and Kompasu and Rai in 2021. For each case, two experiments were conducted with and without assimilating FORMOSAT-7/COSMIC-2 RO bending angle. The RO data were assimilated by the GSI 4DEnVar data assimilation system for a total period of 4 days (with 6 h assimilation window) before the typhoon genesis time, followed by a forecast length of 120 h. The RO data assimilation improved the typhoon track forecasts on average of 42 runs. However, no significantly positive impacts, in general, were found on the typhoon intensity forecasts, except for Maysak. Analyses for Maysak attributed the improved intensity forecast mainly to the improved analyses for wind, temperature, and moisture in the mid-upper troposphere after data assimilation. Consequently, the RO data largely enhanced the evolving intensity of the typhoon at a more consistent movement as explained by the wavenumber-one vorticity budget analysis. On the other hand, a noted improvement on the wind analysis, but still with degraded temperature analysis above the boundary layer, also improved track forecast at some specific times for Hagibis. The predictability of typhoon track and intensity as marginally improved by use of the large RO data remains very challenging to be well explored.

We present the first ever real‐time numerical weather prediction system with 30‐s update cycles at a 500‐m grid spacing for the prediction of convective precipitation in the subsequent 30 min using a new‐generation multi‐parameter phased array weather radar. The system comprises a regional atmospheric model known as the SCALE and the local ensemble transform Kalman filter (LETKF). To accelerate the SCALE‐LETKF system, data transfer between the two aforementioned components is performed using a memory copy instead of a file I/O. A complete real‐time workflow including domain nesting and observational data transfer is constructed. A real‐time test in July and August 2020 showed that the system is fast enough for a real‐time application of 30‐s forecast‐analysis cycles and 30‐min prediction. The development includes a new thinning method considering the spatially correlated observation errors in the dense radar data. This new thinning method is effective in two past case studies in the summer of 2019.

Non-Gaussian forecast error is a challenge for ensemble-based data assimilation (DA), particularly for more nonlinear convective dynamics. In this study, we investigate the degree of the non-Gaussianity of forecast error distributions at 1 km resolution using a 1000-member ensemble Kalman filter, and how it is affected by the DA update frequency and observation number. Regional numerical weather prediction experiments are performed with the SCALE (Scalable Computing for Advanced Library and Environment) model and the LETKF (local ensemble transform Kalman filter) assimilating phased array radar observations every 30 s. The results show that non-Gaussianity develops rapidly within convective clouds and is sensitive to the DA frequency and the number of assimilated observations. The non-Gaussianity is reduced by up to 40 % when the assimilation window is shortened from 5 min to 30 s, particularly for vertical velocity and radar reflectivity.

Non-Gaussian forecast error is a challenge for ensemble-based data assimilation (DA), particularly for more nonlinear convective dynamics. In this study, we investigate the degree of non-Gaussianity of forecast error distributions at 1-km resolution using a 1000-member ensemble Kalman filter, and how it is affected by the DA update frequency and observation number. Regional numerical weather prediction experiments are performed with the SCALE (Scalable Computing for Advanced Library and Environment) model and the LETKF (Local Ensemble Transform Kalman Filter) assimilating every-30-second phased array radar observations. The results show that non-Gaussianity develops rapidly within convective clouds and is sensitive to the DA frequency and the number of assimilated observations. The non-Gaussianity is reduced by up to 40 % when the assimilation window is shortened from 5 minutes to 30 seconds, particularly for vertical velocity and radar reflectivity.

Non-Gaussian forecast error is a challenge for ensemble-based data assimilation (DA), particularly for more nonlinear convective dynamics. In this study, we investigate the degree of non-Gaussianity of forecast error distributions at 1-km resolution using a 1000-member ensemble Kalman filter, and how it is affected by the DA update frequency and observation number. Regional numerical weather prediction experiments are performed with the SCALE (Scalable Computing for Advanced Library and Environment) model and the LETKF (Local Ensemble Transform Kalman Filter) assimilating every-30-second phased array radar observations. The results show that non-Gaussianity develops rapidly within convective clouds and is sensitive to the DA frequency and the number of assimilated observations. The non-Gaussianity is reduced by up to 40 % when the assimilation window is shortened from 5 minutes to 30 seconds, particularly for vertical velocity and radar reflectivity.

Non-Gaussian forecast error is a challenge for ensemble-based data assimilation (DA), particularly for more nonlinear convective dynamics. In this study, we investigate the degree of non-Gaussianity of forecast error distributions at 1-km resolution using a 1000-member ensemble Kalman filter, and how it is affected by the DA update frequency and observation number. Regional numerical weather prediction experiments are performed with the SCALE (Scalable Computing for Advanced Library and Environment) model and the LETKF (Local Ensemble Transform Kalman Filter) assimilating every-30-second phased array radar observations. The results show that non-Gaussianity develops rapidly within convective clouds and is sensitive to the DA frequency and the number of assimilated observations. The non-Gaussianity is reduced by up to 40 % when the assimilation window is shortened from 5 minutes to 30 seconds, particularly for vertical velocity and radar reflectivity.

The FORMOSAT-7/COSMIC-2 Global Navigation Satellite System (GNSS) Radio Occultation (RO) satellite constellation was launched on June 2019 as a successor of the FORMOSAT-3/COSMIC mission. The Central Weather Bureau (CWB) of Taiwan has received FORMOSAT-7/COSMIC-2 GNSS RO data in real time from Taiwan Analysis Center for COSMIC. With the global numerical prediction system at CWB, a parallel semi-operational experiment assimilating the FORMOSAT-7/COSMIC-2 bending angle data with all other operational observation data has been conducted to evaluate the impact of the FORMOSAT-7/COSMIC-2 data. The first seven-month results show that the quality of the early FORMOSAT-7/COSMIC-2 data has been satisfactory for assimilation. Consistent and significant positive impacts on global forecast skills have been observed since the start of the parallel experiment, with the most significant impact found in the tropical region, reflecting the low-inclination orbital design of the satellites. The impact of the FORMOSAT-7/COSMIC-2 RO data is also estimated using the Ensemble Forecast Sensitivity to Observation Impact (EFSOI) method, showing an average positive impact per observation similar to other existing GNSS RO datasets, while the total impact is impressive by virtue of its large amount. Sensitivity experiments suggest that the quality control processes built in the Gridpoint Statistical Interpolation (GSI) system for RO data work well to achieve a positive impact by the low-level FORMOSAT-7/COSMIC-2 RO data, while more effort on observation error tuning should be focused to obtain an optimal assimilation performance. This study demonstrates the usefulness of the FORMOSAT-7/COSMIC-2 RO data in global numerical weather prediction during the calibration/validation period and leads to the operational use of the data at CWB.

Non-Gaussian forecast error is a challenge for ensemble-based data assimilation (DA), particularly for more nonlinear convective dynamics. In this study, we investigate the degree of non-Gaussianity of forecast error distributions at 1-km resolution using a 1000-member ensemble Kalman filter, and how it is affected by the DA update frequency and observation number. Regional numerical weather prediction experiments are performed with the SCALE (Scalable Computing for Advanced Library and Environment) model and the LETKF (Local Ensemble Transform Kalman Filter) assimilating every-30-second phased array radar observations. The results show that non-Gaussianity develops rapidly within convective clouds and is sensitive to the DA frequency and the number of assimilated observations. The non-Gaussianity is reduced by up to 40 % when the assimilation window is shortened from 5 minutes to 30 seconds, particularly for vertical velocity and radar reflectivity.

Non-Gaussian forecast error is a challenge for ensemble-based data assimilation (DA), particularly for more nonlinear convective dynamics. In this study, we investigate the degree of non-Gaussianity of forecast error distributions at 1-km resolution using a 1000-member ensemble Kalman filter, and how it is affected by the DA update frequency and observation number. Regional numerical weather prediction experiments are performed with the SCALE (Scalable Computing for Advanced Library and Environment) model and the LETKF (Local Ensemble Transform Kalman Filter) assimilating every-30-second phased array radar observations. The results show that non-Gaussianity develops rapidly within convective clouds and is sensitive to the DA frequency and the number of assimilated observations. The non-Gaussianity is reduced by up to 40 % when the assimilation window is shortened from 5 minutes to 30 seconds, particularly for vertical velocity and radar reflectivity.

Non-Gaussian forecast error is a challenge for ensemble-based data assimilation (DA), particularly for more nonlinear convective dynamics. In this study, we investigate the degree of non-Gaussianity of forecast error distributions at 1-km resolution using a 1000-member ensemble Kalman filter, and how it is affected by the DA update frequency and observation number. Regional numerical weather prediction experiments are performed with the SCALE (Scalable Computing for Advanced Library and Environment) model and the LETKF (Local Ensemble Transform Kalman Filter) assimilating every-30-second phased array radar observations. The results show that non-Gaussianity develops rapidly within convective clouds and is sensitive to the DA frequency and the number of assimilated observations. The non-Gaussianity is reduced by up to 40 % when the assimilation window is shortened from 5 minutes to 30 seconds, particularly for vertical velocity and radar reflectivity.