Ensemble forecasting and 4-DVAR analysis experiments

Manuel Pondeca

Florida State University

Abstract: PART: 1
An ensemble forecasting system based on the singular vectors~(SV's) of the Fifth Generation of the Penn-State/NCAR Version-5 Mesoscale Non-hydrostatic Model is developed and tested on synoptic situations featuring extratropical mid-Pacific cyclones. Results are presented for four case studies, each with a five-day duration, and selected within the time period of the experimental project NORPEX between 1200 UTC 4 February and 1200 UTC 24 February 1998.

Compared with control forecasts initiated from the NCEP analyses, the ensemble mean forecasts yield improved skill scores measured in terms of mean-squared errors and correlation coefficients for the 500 hPa geopotential heights. The perturbation sea-level pressure fields of the ensemble mean forecasts are shown to compensate for errors in the intensity and phase of the targeted cyclones.

A few results from a case study of the 4D-Var assimilation of total zenith delay~(TZD) observations from a dense GPS-network and profiler wind data into the MM5~mesoscale model will be reported. The study targets the 12-hour period from 0000~UTC to~1200~UTC~6~December~1997, characterized by the passage of a frontal system that produced intense rainfall over Southern California.