Global ROTI forecasting with a Bayesian model based in long-tail distributions

This study introduces a Bayesian probabilistic model for forecasting the fluctuations in the Rate of TEC Index (ROTI), which indicate the presence of ionospheric disturbances that can impact Global Navigation Satellite Systems (GNSS) and communication networks. The forecast method divides the Earth...

Descripción completa

Detalles Bibliográficos
Autores: Monte-Moreno, Enric, Yang, Heng, Hernández-Pajares, Manuel
Tipo de recurso: artículo
Estado:Versión publicada
Fecha de publicación:2026
País:España
Institución:Consejo Superior de Investigaciones Científicas (CSIC)
Repositorio:DIGITAL.CSIC. Repositorio Institucional del CSIC
OAI Identifier:oai:digital.csic.es:10261/425077
Acceso en línea:http://hdl.handle.net/10261/425077
https://api.elsevier.com/content/abstract/scopus_id/105029011365
Access Level:acceso abierto
Palabra clave:Bayesian model
Forecast
GNSS
Ionosphere
Long-tail distributions
ROTI
Space weather
Descripción
Sumario:This study introduces a Bayesian probabilistic model for forecasting the fluctuations in the Rate of TEC Index (ROTI), which indicate the presence of ionospheric disturbances that can impact Global Navigation Satellite Systems (GNSS) and communication networks. The forecast method divides the Earth into a grid of 2.5∘ latitude by 5∘ longitude cells to predict when ROTI will exceed thresholds of 0.1, 0.25 and 0.5 TECU/min, with time horizons ranging from 30 min to 6 h. The method is based on the burstiness property of long-tailed distributions and provides as a forecast the median value of activity at each range, of both ROTI amplitude and forecast horizon. Previous proposed ROTI forecasting methods may degrade when faced with missing data points and the irregular, heavy-tailed characteristics of ROTI. In contrast, our model, based on the power-law dynamics observed in the persistent and bursty nature of long-tail distributions, allows for gaps in the measurements and provides a global forecast for regions covered by the network of GNSS stations. The performance of the model has been validated against historical GNSS data across various ionospheric conditions, demonstrating its robustness. The proposed Bayesian probabilistic model demonstrates robust forecasting capabilities, validated across diverse ionospheric conditions using historical GNSS data. It achieves strong performance metrics, with Weighted Kappa values exceeding 40% for prediction horizons up to 120 min and maintaining Mean Precision above 65% across all tested horizons from 30 min to 6 h. By forecasting the probability of ROTI exceeding specific levels, this method helps to identify geographical regions where GNSS reliability may be compromised, thereby aiding in the mitigation of adverse space weather effects on critical navigation and communication systems.