Impact of Satellite Data Latency on Global Weather Forecasts

This project investigates the impact of satellite data latency on NOAA’s global prediction systems by conducting Observing System Experiments (OSEs) using NOAA’s Global Data Assimilation System (GDAS). The NEMSfv3GFS system is used for the forecast model.

For this study three experimental cases are run. For each case, the time window for satellite data assimilation into the GDAS is progressively reduced from the standard (-3 hrs/+3 hrs) about the cycle start time: (-3 hrs/+2 hrs) for the first, (-3 hrs/+1 hrs) for the second, and (-3 hrs/0 hrs) for the third, as a proxy for 1-hr, 2-hr, and 3-hr data latency, respectfully. The period of interest is late-June through mid-September, during the 2018 season.

The system is cycled every 6 hours, with 10-day forecasts launched every 24 hours at 00 UTC. Resolution of the NEMSfv3GFS model is at C384 (~25 km). NEMSfv3GFS utilizes a 4-Dimensional Ensemble Variational (4DEnVar) system to generate analyses at C192 (~50 km).

Results are compared to a Control, into which all existing observations are assimilated. Results will include standard metrics such as systematic bias and root-mean-squared errors, and an anomaly correlation score for atmospheric variables. Results may also include further analysis of high-impact case studies, such as tropical cyclone activity that can be evaluated in terms of track and intensity errors.