Investigating the drivers of long-term trends in the upper atmosphere over Rome across four decades

The nature of the long-term changes in the upper atmosphere morphology at mid-latitude remains a subject of debate, particularly regarding whether these changes are purely driven by geomagnetic and solar activities or whether forcing from the lower atmosphere, such as CO2 variations, may play a role. To contribute to this debate, we investigate the nature of the long-term trends of the ionospheric and thermospheric parameters by leveraging on ionosonde data digitally recorded at the Rome Observatory since 1976. The following parameters have been investigated under sunlit conditions (12:00 Local Time): critical frequency of the F1 layer (foF1); critical frequency of the F2 layer (foF2), atomic oxygen concentration at 300 km ([O]); ratio between atomic oxygen and molecular nitrogen concentrations at 300 km altitude ([O]/[N-2]); exospheric temperature (T-ex); thermospheric density at 300 km (rho). The ionospheric parameters are manually scaled from digital ionograms, whereas thermospheric parameters are retrieved using the THERmospheric parameters from IONosonde observations (THERION) method, which utilises ionosonde observations and a physical model of the ionospheric F region. To investigate the influence of the solar and geomagnetic activity on long term variations, we consider the solar radio flux at 10.7 cm (F10.7) and the geomagnetic disturbance index Ap. To identify the various frequency/period components of the time series under consideration and identify the trends, we leverage the high scale/time resolution offered by the Fast Iterative Filtering (FIF) algorithm. A regression analysis of thermosphere/ionosphere parameters against geomagnetic/solar activity indices has then been conducted to investigate the drivers of long-term variability. Our findings reveal that the identified trends are predominantly controlled by external drivers, particularly long-term solar and geomagnetic activity variations. The adopted methodology, based on regression modelling, demonstrates that variability in F10.7 and Ap accounts for nearly all of the observed changes, with the exception of atomic oxygen ([O]), which displays a slightly higher unexplained variability (similar to 7%). The inclusion of CO2 concentration as an additional driver improves the regression model for [O]. However, the effect remains statistically limited, indicating that the impact of CO2 on thermospheric cooling might be of little significance. Further studies with extended time series are necessary to better quantify this relationship and evaluate its importance. These results highlight the predominant influence of solar and geomagnetic activity in determining upper atmosphere long-term trends at mid-latitudes.