Impact of Filtering Methods in the Assessment of COSMIC Electron Density Profiles Over the Brazilian Region: Excluding Profiles With Negative Values Is the Best Strategy?

Abstract

Abstract Global Navigation Satellite Systems radio occultation (RO) is a valuable and relevant source of information from the atmosphere. Many efforts have been made to provide methods of validation of RO profiles, however, no clear methodology for filtering the profiles from RO can be easily found. In this study, we present strategies to filter RO electron density profiles over a low latitude region. Different methods are applied considering minimum values, manual filtering, range, and discrepancies with reference to the Chapman profile. The assessment is performed by means of ionosonde data for two years (2014–2015). The results show that assuming a minimum electron density limit slightly smaller than zero consistently with electron density estimation errors under very low actual values can provide significant improvements. In this study, the use of a − el/ limit led to better rates of retained profiles (65% vs. 32%) and a reduction in root mean square (RMS) (12% vs. 7%) compared to the exclusion of all profiles with negative values. When considering only electron density values above 100 km in altitude, there is still a significant loss of about 20% in the number of profiles with the same reference values (84% vs. 63%), with a similar performance in the RMS. The best performance is obtained with the strategies: range of occurrence (200–450 km) and the outliers identification (5 standard deviations limit), leading to a RMS reduction of about 26% and 19%, and a RMS reduction of 34% and 25%, respectively, while keeping 91% and 82% of the original profiles. Plain Language Summary Radio occultation (RO) is a method used for sounding the Earth's atmosphere. In the path between the Global Navigation Satellite Systems and the Low Earth Orbit satellite, the transmitted signal is influenced by the components of the atmosphere. With the RO it is possible to retrieve important parameters from the atmosphere and better understand its behavior. RO data is important because its data present a homogeneous distribution and can be used to complement several models, for example, the prediction of natural hazards. However, before using these profiles, it is important to validate them and develop methods for filtering, in order to avoid unreliable data. In this work, we present several ways of filtering RO ionospheric profiles considering a challenging region in low latitude, and data close to solar maximum. The data sets with the filtered profiles are assessed using ionosonde data as reference. For the scenarios considered, the best filtering strategies obtained are based on a range of occurrence and identification of outliers. The use of a negative minimum electron density limit performed better than a zero tolerance to negative values. Key Points Proposal of filtering strategies for COSMIC profiles evaluated using 2‐year data set close to solar maximum in low latitude Strategies assessed by profile exclusion rates and root mean square reduction for and , compared to ionosonde data Filtering based on a range of occurrence shown to be more effective compared to traditional filtering methods