Data Assimilation

The ability to predict conditions in Earth’s ionosphere and thermosphere is of increasing societal relevance due to the growing dependence on, for example, satellite based communications and navigation (e.g., GPS) systems. The ionosphere and thermosphere are known to vary significantly from day-to-day, and this day-to-day weather is largely driven by processes originating in the lower atmosphere (below ~50 km), especially during periods of low solar and geomagnetic activity.

Graphic of zonal winds
Global normalized root mean square error in zonal wind for wavenumbers 0-6 in different seasons at (a) 1.0, (b) 0.1, (c) 0.01, and (d) 0.001 hPa in WACCMX+DART hindcast experiments. Solid lines indicate the median of all hindcast experiments, and thick dashed line is the best fit error growth curve. The results illustrate that the error growth is faster at higher altitudes, indicating worse forecast skill.

Accurate forecasting of the ionosphere-thermosphere variability thus partially depends on the ability to accurately forecast the component of variability that originates in the lower atmosphere. Making use of recent developments in whole atmosphere modeling, we have performed the first comprehensive investigation of current capabilities to forecast the lower atmospheric drivers of ionosphere-thermosphere day-to-day variability. The capabilities of the Whole Atmosphere Community Climate Model with thermosphere ionosphere eXtension, including data assimilation (WACCMX+DART), was used to evaluate the capability of the model to forecast conditions at altitudes (~60-120 km) that are relevant for generating the day-to-day variability in the ionosphere and thermosphere. The forecast skill was evaluated based on a set of hindcast experiments that were initialized on the 1st and 15th of each month of 2009 and 2010. The results demonstrate that the forecast skill decreases with increasing altitude, and also decreases for smaller spatial scales. Furthermore, it is found that, on average, the primary drivers of spatial and temporal variability in the ionosphere can be forecast several days in advance. The forecast skill of the ionosphere is typically thought to be less than ~24 hours; however, our investigations using WACCMX+DART illustrate that the forecast skill of the ionosphere can potentially be increased by incorporating forecasts of the lower atmosphere drivers.