The Impact of an Hourly Assimilation Cadence in the NOAA Global Data Assimilation System

Most operational centers cycle their global data assimilation systems at a cadence of 6 hours for global weather prediction, creating analyses 4 times per day. This may not be frequent enough to accurately estimate fast-moving storms, motivating development of a rapidly-updated global assimilation system. One barrier to real-time rapid updates is data latency, or the delay between when an observation is taken and when it arrives at the operational center. Recently, a study by Slivinski et. al. (2022) showed that an overlapping-window approach could successfully be applied to deal with this latency while cycling hourly in a version of the NOAA operational global data assimilation system. This approach was used to reduce the assimilation cadence from 6 hours to 1 hour with some improvement to forecast performance, suggesting that assimilating observations closer to the time they were taken is beneficial. In this study, we eliminate the role of data latency by assuming all observations can be used in the GDAS as soon as they are taken. We compare results from experiments with 1-hourly assimilation windows to those with 6-hourly windows. The difference in the skill of forecasts informed by the same observations can then be used to measure and analyze the impact of reducing the assimilation window.

Slivinski, L. C., D. E. Lippi, J. S. Whitaker, G. Ge, J. R. Carley, C. R. Alexander, and G. P. Compo, 2022: Overlapping Windows in a Global Hourly Data Assimilation System. Mon. Wea. Rev., 150, 1317–1334, https://doi.org/10.1175/MWR-D-21-0214.1.