Conceptual Study of Real-time Ionospheric Threat Adaptation Using Space Weather Forecasting for GNSS Augmentation Systems

Abstract

Current GNSS augmentation systems attempt to consider all possible ionospheric events in their computations of worst-case errors in user position solutions. The resulting error bounds are conservative because of this need to cover all possible anomalous conditions that might go undetected. This conservatism can be mitigated by subdividing anomalous conditions into several classes of severity and using different values of threat-model bounds for each class. This is possible if the level of ionospheric activity is classifiable and predictable (at least over periods of tens of minutes to hours) from measurable space weather conditions. This paper presents a new concept of real-time ionospheric threat adaptation that adjusts the ionospheric threat model in real time instead of always using the same ‘worst-case’ threat model. This is done by utilizing a relationship between ionospheric activity and space weather indices. The worst-case threat is defined as a function of the values of space weather indices. Predicted values of space weather indices are used for determining the corresponding threat model. Since space weather prediction itself is not reliable due to prediction errors, an uncertainty model is derived from 15 years of historical data, and the prediction errors are bounded to the required level of integrity of the system being supported. This concept was demonstrated by assessing the performance of real-time ionospheric threat adaptation when applied to the Local Area Augmentation System (LAAS) threat model used for Category I precision approach in the Conterminous United States (CONUS). The disturbance-storm time (Dst) index was selected as a measure of space weather intensity. The relationship between final Dst and the worst ionospheric gradients (or "slopes") identified in CONUS was defined. The ‘predicted Dst bound’ derived by taking into account prediction error statistics was then used to determine the worst slope bounds in real-time. As a result, the upper slope bound of the current threat model (425 mm/km for LAAS) could be lowered about 94% of the time during the 15 years of data (from 1995 to 2009) with an integrity of (1 - 10-7) using the bounded prediction value of Dst for real-time threat determination.