How satellite data corrects our models of the upper atmosphere, the highest layer of the sky

What is it about?

Every satellite in a low Earth orbit skims through air that is so thin it barely registers, yet it is still dense enough to tug a spacecraft slowly off course. Forecasting this drag requires predicting the upper atmosphere, hundreds of kilometres above the weather. The physics models used to describe this layer contain errors, and these errors become more significant when the sun is active. This study tested whether incorporating real satellite observations into a physics model could rectify these errors and enhance forecasts. The team used electron density readings gathered by a constellation of GPS-tracking satellites called COSMIC, which are derived from radio signals bent by the ionosphere as they pass through it. The team then fused these readings hourly into an atmospheric simulation using a statistical technique borrowed from numerical weather prediction. The results were divided cleanly along the solar cycle. When the sun is quiet, the satellite data substantially improves one-hour forecasts, and these improvements carry forward into subsequent cycles. During solar maximum, however, intense solar radiation overpowered the adjustments almost immediately. An interesting result was that correcting the electron density also altered estimates of neutral air mass density, the quantity that determines satellite drag, suggesting applications in orbit prediction that the researchers had not set out to investigate.

Featured Image

Photo by Yuheng Ouyang on Unsplash

Photo by Yuheng Ouyang on Unsplash

Why is it important?

In this work, we assimilate globally abundant radio occultation-derived electron density into a physics-based model. RO-derived electron density is one of the most promising means to test the effect of assimilation on the model forecasted state on a global scale. Accurate forecasting of the upper atmosphere is essential for satellite operations and space weather alerts. However, current models perform poorly, particularly during periods of high solar activity. This study demonstrates that real ionospheric data can meaningfully enhance the accuracy of short-range forecasts and refine air density estimates relevant to orbit prediction. This technique is most effective during solar minimum; however, fundamental model limitations remain unresolved during solar maximum and further work is needed.

Perspectives