engleski

Baseline adaptation based on weighted cubic smoothing splines

The geomagnetic observatory final product – definitive data are obtained by calibrating variometer recordings to the absolute level using adapted baseline values calculated from weekly observations. This way we eliminate the presence of long-term drift in variometer data. The definitive data have dynamic range from several nT up to several tens of μT with resolution of 0.1 nT and sampling frequency 0.017 Hz (standard minute means) or even higher. In cases when we have low-quality observations in conjunction with variable variometer baselines, traditional straightforward fitting techniques may not be the appropriate tool for baseline adaptation. We present a fitting method insensitive to outliers and flexible enough to accurately fit local baseline variation occurring over weeks up to several months. Due to the semi-parametric nature of splines we have almost instant overlapping between temporarily quasi-definitive (QD) baselines and definitive ones that are obtained after the end of calendar year. Robustness to outliers brings the possibility to promptly calculate QD baselines without major revision of the input dataset. Consequently, the proposed technique may find its use in automated processing or in the future when observatories start to operate automated instruments for absolute measurements. With an increase in the number of measured absolute values (once per day, or per hour) it could even be convenient to have a fitting tool that can properly fit short-term baseline variations inside several days or less.

geomagnetic observatories; variometers; baseline adaption; definitive and quasi-definitive data

nije evidentirano