Estimating error statistics for Chambon-la-Forêt observatory definitive data - INSU - Institut national des sciences de l'Univers Accéder directement au contenu
Article Dans Une Revue Annales Geophysicae Année : 2017

Estimating error statistics for Chambon-la-Forêt observatory definitive data

Résumé

We propose a new algorithm for calibrating definitive observatory data with the goal of providing users with estimates of the data error standard deviations (SDs). The algorithm has been implemented and tested using Chambon-la-Forêt observatory (CLF) data. The calibration process uses all available data. It is set as a large, weakly non-linear, inverse problem that ultimately provides estimates of baseline values in three orthogonal directions, together with their expected standard deviations. For this inverse problem, absolute data error statistics are estimated from two series of absolute measurements made within a day. Similarly, variome-ter data error statistics are derived by comparing variometer data time series between different pairs of instruments over few years. The comparisons of these time series led us to use an autoregressive process of order 1 (AR1 process) as a prior for the baselines. Therefore the obtained baselines do not vary smoothly in time. They have relatively small SDs, well below 300 pT when absolute data are recorded twice a week-i.e. within the daily to weekly measures recommended by INTERMAGNET. The algorithm was tested against the process traditionally used to derive baselines at CLF observatory , suggesting that statistics are less favourable when this latter process is used. Finally, two sets of definitive data were calibrated using the new algorithm. Their comparison shows that the definitive data SDs are less than 400 pT and may be slightly overestimated by our process: an indication that more work is required to have proper estimates of absolute data error statistics. For magnetic field modelling, the results show that even on isolated sites like CLF observatory, there are very localised signals over a large span of temporal frequencies that can be as large as 1 nT. The SDs reported here encompass signals of a few hundred metres and less than a day wavelengths.
Fichier principal
Vignette du fichier
angeo-35-939-2017(1).pdf (5.49 Mo) Télécharger le fichier
Origine : Fichiers éditeurs autorisés sur une archive ouverte
Loading...

Dates et versions

insu-02305910 , version 1 (18-10-2019)

Identifiants

Citer

Vincent Lesur, Benoît Heumez, Abdelkader Telali, Xavier Lalanne, Anatoly Soloviev. Estimating error statistics for Chambon-la-Forêt observatory definitive data. Annales Geophysicae, 2017, 35 (4), pp.939-952. ⟨10.5194/angeo-35-939-2017⟩. ⟨insu-02305910⟩
47 Consultations
69 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More