Sungkoo Bae

GLAS PAD Calibration Using Laser Reference Sensor Data

Launched in January 2003, the Ice, Cloud and land Elevation Satellite (ICESat) has conducted several periods of science operations with the Geoscience Laser Altimeter System (GLAS). The second extended for 55 days, between September and November 2003. During this period, data from the GLAS Laser Reference Sensor (LRS) was used to correct for apparent, possibly thermally induced, motion of the Instrument Star Tracker (IST). The use of this calibrated IST data in precision attitude determination (PAD) yields significant improvement in estimates of the laser-pointing direction.

Laser Reference Sensor Alignment Estimation Using Reference Signal Observations

The Laser Reference Sensor is the central instrument in the Ice, Cloud, and land Elevation Satellite laser pointing knowledge system, simultaneously observing the altimetry laser, stars, and a reference signal in a single instrument coordinate frame. The reference signal is intended to provide direct observations of the alignment between the Laser Reference Sensor and the Instrument Star Tracker. The reference signal failed early in the mission and a method was developed to partially replace it by comparing two attitude time series: an attitude filter time series for the Instrument Star Tracker and a pure-gyro time series for the Laser Reference Sensor. Only the Instrument Star Tracker and gyros are used in the replacement method, with the gyros tracking the Laser Reference Sensor attitude in order to make the relative motion of the Instrument Star Tracker observable.

Biased Star Tracker Measurements of Forty-Nine Stars from Flight Data

DOI: 10.2514/1.49412 Approximately 1% of the 10,000 stars observed by the ice, cloud, and land elevation satellite star trackers are believed to have position measurement biases caused by near-neighbor stars. The biases are in the tracker measurements, not the star catalogs. Empirical biases were derived for 49 stars. A survey was performed to detect and characterize biased stars by treating each observed star as a target, predicting the tracker measurements of the target, andthen comparing the observations andpredictions. The distribution of prediction accuracies for unbiased stars had a mean of 1.46 arcseconds and a standard deviation of 0.61 arcseconds. Ninety percent of the sky was covered and five million passes of 10,472 stars were processed. Stars were classified using a Mahalanobis distance parameter,whichscaledpositionresidualsbypredictionuncertainties.StarswithlargeMahalanobisdistanceswere then studied individually.