Linear forecasting of the F_10.7 proxy for solar activity

Abstract

The ability to accurately forecast variations in the solar extreme ultraviolet irradiance is important to many aspects of operational space weather. For example, variations in the Sun's radiative output at these wavelengths drive changes in thermospheric density, which perturbs the trajectories of objects in low Earth orbit. Thus, predicting the conjunction of an operational satellite with orbital debris requires accurate forecasts of solar activity. In this paper we present a simple linear forecasting model for the 10.7 cm radio flux (F10.7), a commonly used proxy for solar activity. Comparisons with simple reference models indicate that this linear model has positive skill for all forecast days that we have considered. We also examine the impact of the F10.7 forecast skill on empirical model predictions of thermospheric density and ionospheric total electron content.

EUV Irradiance Inputs to Thermospheric Density Models: Open Issues and Path Forward

“…But to be considered for operational deployment, the forecast accuracy of these approaches needs to be tested against the performance of the current operational inputs, such as F 10.7 or other proxies. For example, Warren et al (2017) find that the Henney et al (2012) method performs at a similar level to the (much simpler) F 10.7 linear forecast. As our knowledge on the evolution of solar magnetic fields improves so will such models.…”

“…The extrapolations are generally within the 5% density uncertainty requirement (section 2.1) for 2-3 day forecasts (Figure 3j in Emmert et al, 2017), but the 10 day EUV forecasts behave as Brownian motion and hence lead to~(time) 5 increases for in-track errors. More recently, Warren et al (2017) evaluated the performance of a linear extrapolation model for F 10.7 and showed that the forecast skill peaks at 6 days compared to climatology and persistence.…”

The 10.7‐cm radio flux multistep forecasting based on empirical mode decomposition and back propagation neural network

“…The closer to 1, the better (n∈ [ 1,29] ) days. The input signals are forwarded in the order of the input layer, the hidden layer, and the output layer.…”

Thermosphere‐Ionosphere Modeling With Forecastable Inputs: Case Study of the June 2012 High‐Speed Stream Geomagnetic Storm

“…Fortunately, fully physics-based models (Richmond et al, 1992;Roble and Ridley, 1994;Millward et al, 1996;Huba et al, 2000;Ridley et al, 2006) are able to accept such drivers as input and thus can in principle be used for a-few-day lead time forecasts as long as forecasted solar drivers are available. While this research area is still in its early stages, solar wind and solar irradiance forecasts are major focuses of the space weather research effort (Owens et al, 2008;Jian et al, 2015;Henney et al, 2015;Warren et al, 2017). Therefore, we investigate thermospheric-ionospheric forecasts with a lead time of a few days from the perspective that reliable solar forecasts will eventually be available a few days ahead.…”