Abstract
At the National Meteorological Center (NMC), a new analysis system was implemented into the operational Global Data Assimilation System on 25 June 1991. This analysis system is referred to as Spectral Statistical Interpolation (SSI) because the spectral coefficients used in the NMC spectral model are analyzed directly using the same basic equations as statistical (optimum) interpolation. The major differences between the SSI analysis system and the conventional optimum interpolation (OI) analysis system previously used operationally at NMC are:
–The analysis variables are closely related to the coefficients of the NMC spectral model.
–Temperature observations are used, not heights as in the previous procedure. As a result, aircraft temperatures are being used for the first time at NMC.
–Nonstandard observations, such as satellite estimates of total precipitable water and ocean-surface wind speeds, can be easily included.
–No data selection is necessary. All observations are used simultaneously.
–The dynamical constraint between the wind and mass fields is more realistic and applied globally.
–Model initialization has been eliminated. The analysis is used directly as the forecast model initial condition.
Extensive pre-implementation testing demonstrated that the SSI consistently produced superior analyses and forecasts when compared to the previous OI system. Improvement in skill is shown not only for the 3–5-day forecasts, but also in one-day aviation forecasts.
