Sunday, 12 January 2020
Reliable forecast by numerical weather prediction depends on precise temperature and humidity profiles. This study presents a temperature and humidity retrieval system developed for use with the Chinese new generation polar meteorological satellite FY-3D/HIRAS (Hyperspectral Infrared Radiation Atmospheric Sounding) to retrieve atmospheric temperature and humidity profiles using the variational method. In the technique, the WRF (The Weather Research and Forecast) model is used to generate the first guess, such as atmospheric temperature and humidity as well as other parameters including surface temperature, skin pressure. The RTTOV model is needed to work as a forward observation operator and adjoint model. In the procedure of the minimization of the cost function, Newton nonlinear iteration method was employed. Several key techniques such as quality control of satellite data, background error covariance localization and observation error covariance calculation were solved in this retrieval algorithm. The atmospheric profile retrieval experiments were conducted in Beijing during July, August and September, 2018. The accuracy of the results was evaluated through a comparison with ERA-Interim data. The mean bias (MB) of temperature profiles fluctuated between -0.8 K and 0.9 K, while the root-mean-square error (RMSE) varied between 0.5 K and 2.6 K in the whole atmosphere. In the boundary layer, the 1D-VAR algorithm performed better with the RMSE of 0.88 K, which decreased by 1.64 K compared with the first guess. In the middle troposphere, the retrievals were more dependent on the first guess. With respect to relative humidity, the accuracy of the whole troposphere was improved with the addition of satellite observation, where MB was on the order of -5.68%-2.83%, and the largest RMSE was 9.60% near 775hPa. However, the analysis field obtained by the algorithm were close to the first guess in the stratosphere.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner