Analysis of Relationship Between Ionospheric and Solar Parameters Using Graphical Models

0542098 - ÚFA 2022 RIV US eng J - Journal Article
Podolská, Kateřina - Koucká Knížová, Petra - Chum, Jaroslav - Kozubek, Michal - Burešová, Dalia
Analysis of Relationship Between Ionospheric and Solar Parameters Using Graphical Models.
Journal of Geophysical Research-Space Physics. Roč. 126, č. 5 (2021), č. článku e2020JA029063. ISSN 2169-9380. E-ISSN 2169-9402
R&D Projects: GA ČR(CZ) GA18-01969S; GA ČR(CZ) GA18-01625S
Grant - others:ESA - The European Space Agency(XE) ESA 4000126709/18/NL/IA VERA
Institutional support: RVO:68378289
Keywords : atmospheric waves * graphical models of conditional * independences * ionospheric variability
OECD category: Meteorology and atmospheric sciences
Impact factor: 3.111, year: 2021
Method of publishing: Limited access
https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2020JA029063

For the investigation of time variations of critical frequency (foF2), solar radiation flux at 10.7 cm wavelength (F10.7 index) and geomagnetic activity index Kp, we use conditional independence graphical models which describe and transparently represent the structure of relationships in the time series. We employ multivariate statistic methods applied to daily observational data obtained from midlatitude ionosondes within the period 1994–2009 (23rd Solar Cycle). It is demonstrated that conditional independence graphical models represent a robust method of multivariate statistical analysis, useful for finding a relation between one of the main ionospheric parameters and space weather conditions. This method appears to be more appropriate than correlation analysis between foF2 and the main geomagnetic and solar indices, especially for long‐term data for which the model characteristics may change or time series can be interrupted. We compare the results obtained by the graphical models method with cross‐correlation analysis. We found that the method is suitable for the analysis of the dependence between foF2 values and time‐shifted F10.7 time series. In particular, we clearly identified +0 day shift in all cases, and +1‐, +2–3‐, and +4–5‐day shifts in European, American, and East Asia sector, respectively. The conditional independence graphs method can be applied even in the case when classical parametric methods are not convenient.
Permanent Link: http://hdl.handle.net/11104/0319589