Rain storm spatial structure derived from radar observation of vertically integrated liquid water.

摘要

This work presents a study of the spatial structure of rain storms with a focus on the vertically integrated liquid (VIL) water content as observed by weather radar. The study shows two-dimensional VIL (kg/m 2) fields observed at 10-minute time increments during rainfall events recorded by the Tulsa, Oklahoma WSR-88D radar (KINX). The event features include a duration of approximately 6 hours, 36 radar reflectivity volume scan observations, and a relatively low magnitude average intensity of 1–2 mm/hr. Using VIL derived from radar volume scans at a horizontal resolution of 4 km, the spatial correlation function for each observation time was developed. Additionally, a summary of the spatial correlation function for the entire event, representing the average of all the time steps is presented. This case study shows the highest correlation magnitudes occurred in the earlier time of the event duration. The critical correlation distance (defined here by the threshold 1/e) of approximately 4 to 8 km was found when considering temporal averages over all time steps and a VIL magnitude threshold of 0.025 kg/m2 (2.5mm precipitable water).; This work was conducted to support the implementation of real-time data assimilation methods used for updating initial conditions in Quantitative Precipitation Forecast (QPF) models from weather radar data. Specifically, the model HydroQPF (French et al. 2001), developed from the work of Georgakakos and Bras (1984), Lee and Georgakakos (1990), Seo and Smith (1992), and later extended by French and Krajewski (1994) to utilize a formulation for real-time state updating (Georgakakos 2000) from radar reflectivity observations. Optimal updating requires the definition of state covariance for the state domain; and the findings of this study explain the implementation of the update filter formulation in HydroQPF.