Modeling the 20 November 2003 Ionosphere Storm with GRACE

The Local Area Augmentation System (LAAS) provides differential GPS corrections for Category I precision approach to aviation users within tens of kilometers of the LAAS Ground Facility (LGF). To ensure integrity, any circumstance that may lead to hazardously misleading information (HMI) being transmitted to the user must be identified. If the probability of the situation exceeds the allocated integrity risk, its maximum user errors must be bounded by one or more user-computed protection levels. Over short baselines, the differential ionosphere error between a user and LGF can pose one such threat. This work examines how the spatial rate of change of the ionosphere over baselines of tens of kilometers can be modeled to provide insight into hazardous conditions. We develop a static spatial model of the ionosphere based on a technique developed for understanding the impact on WAAS of the 31 October 2003 localized nighttime ionosphere enhancement. In this work we apply the method to the 20 November 2003 ionosphere storm, a day on which GPS observations of the disturbed ionosphere were previously used to populate the space of possible ionosphere threats to LAAS users. This 3-D model assigns vertical profiles to latitude and longitude regions. The horizontal “enhancement” and “background” ionosphere regions are identified based on measurements made by the Continuously Operating Reference Stations (CORS). Since observations of vertical density variations are limited with GPS receivers on the ground, space-based GPS data from on board one of the GRACE satellites passing through the disturbed ionosphere around 19:40 UT is used to test a range of vertical electron density profiles. The 500 km orbital altitude of the GRACE satellites effectively limits the contribution to the total electron content of the topside and plasmasphere. After finding the electron density model that minimizes the mean squared error, we then integrate through the lines of sight of a set of CORS receivers located in Ohio and separated by baselines as short as 50-75 km. Pairs of these stations mimic user-LGF pairs and were used in the literature to estimate the possible spatial decorrelation over LAAS baselines. We integrate through the CORS lines of sight (LOS) to compute the differential ionosphere over tens-of-kilometer baselines. We find that a 3-D static model of the ionosphere cannot reproduce the 350-400 mm/km spatial rates of change in ionosphere error that have been verified with data. However, we find that allowing the ionosphere anomaly to sweep westward at 300 m/s, as has been estimated from data, can in fact reproduce rates of change between neighboring CORS stations on the order of 400 mm/km. Further refinements are needed to optimize the model-predicted delay over the three-hour timespan of the terrestrial observations, though the gross features are similar. This work confirms that the relative velocity of the ionosphere structure and the lines of sight are an important factor in apparent gradients in the ionosphere and also helps to further validate the estimates made from the CORS observations of short baseline spatial rates of change in ionosphere delay.