A Bootstrap-Based Approach for Improving Measurements by Retarding Potential Analyzers

Retarding potential analyzers are frequently flown on small satellites as in situ ion probes, from which can be derived a number of ion plasma parameters from a current-voltage relationship (I-V curve). The traditional method of analyzing retarding potential analyzer data produces inaccuracies in derived estimates when there is significant noise present in the instrument measurements. In this study we investigate the dependencies between parameters that produce uncertainties in noisy I-V curves. It is found that multiple combinations of ion velocity and spacecraft floating potential can produce I-V curves that lie within the noise envelope, which renders it difficult for a traditional curve fitting technique to objectively and accurately estimate parameters from a noisy I-V curve. In this paper we propose BATFORD-a bootstrap resampling-based technique to improve the accuracies of parameter estimates. It is particularly useful when signal-to-noise ratios are low. The algorithm is tested against a traditional curve fitting method for a simulated data set comprising I-V curves for the middle- and low-latitude ionosphere at low Earth orbit altitudes around 450 km, where O+ is the predominant species. BATFORD is found to provide more robust and reliable estimates assuming generalized noise distribution characteristics. As further validation, the algorithm is applied to satellite data from an orbit with deep plasma bubbles and hence low signal levels.