5.4. Baseline Adoption

A number of methods are used in observatory data processing procedures to adopt baselines from observed baseline values. An adopted baseline may be fitted to the baseline values by hand or by a computer algorithm. The baseline values obtained from the adopted baseline for each day are written to a file in the INTERMAGNET baseline format IBFV2.00, (APPENDIX E4). The format includes a section for comments pertaining to dates and times of baseline adjustments or changes in instrumentation. These files must be transmitted annually to INTERMAGNET along with 1-minute data for the IRDS production.

Statistics of the difference between the adopted and observed baseline can be generated to verify that confidence limits for the adopted baseline meet the INTERMAGNET specifications.

Fig. 5.1 shows observed values (circles) and adopted baseline values (dots) for magnetic field components X, Y, and Z for Belsk observatory during 2003. The time series of F-P total field differences, defined in Section 5.5, corresponds to the time series of adopted baseline values.

Often the baseline changes smoothly due to seasonal changes in temperature or humidity, or from ageing of magnetometer components. The constant segments used to adopt the baseline shown in Fig. 5.1 are just one example for the adoption method. In order to avoid jumps in the calibrated data (at the times between segments of constant baselines), the baseline could be adopted using continuous, linear segments or curves. For the latter, often polynomials or splines of various degrees are used, and can be fitted to the observed baselines by automatic algorithms. Significant jumps in the baseline can still occur (e.g. during maintenance work at the variometer), and these should be reflected by the adopted baseline. Then the computer algorithms for baseline fitting should be used separately to fit the baselines before and after such a jump. The timing of a baseline jump caused by maintenance often coincides with artificially disturbed variometer data. Then the timing of the baseline jump can conveniently be set into the time interval where data from the variometer is not used as it is disturbed during maintenance.

Once the baseline is adopted, the variometer can be calibrated. To check the quality of the variometer data and the adopted baseline, it is helpful to plot the total field differences. For a well-adopted baseline, the total field differences ideally should be small (e.g. smaller than 0.5 to 1 nT), and should not depend on the shape of the baseline too much. If there is an error in the adopted baselines, or in the subsequent calibration procedure, then this will likely show up in total field differences. Also, this will show up when you calculate the differences between the absolute field values determined in the absolute measurements minus the calibrated data from the final data product (e.g. from the IAF files, see APPENDIX C1 ) at the time of absolute measurements.