Accuracy and availability of data

[11]  The final accuracy of the data depends on many factors. The first of them is the quality of the baselines, that is, the calibration curves that are used to correct the slow drift in time of the vector magnetometer in order to produce definitive data. Baselines for the Borok observatory are obtained for the H (horizontal), D (declination) and Z (downward vertical) components by fitting a spline curve to the correction values deduced from the absolute measurements. Each year, the spline curve is calculated using data from December of the previous year to January of the following year, in order to avoid discontinuities from one year to the other.

Figure 5
[12]  As an example, baselines for the year 2006 are represented in Figure 5. More than hundred absolute measurements were made in 2006 by BGO observers (each one represented by a coloured square), making it possible to finely follow the magnetometer drift. For each component, the quality of the absolute measurements may be assessed by calculating the standard deviation of the differences between the measurements and the baseline curve. The obtained standard deviations are 0.44 nT for H, 0.39 nT for Z and 11.6 arcsec for D, which are well within INTERMAGNET requirements. Calculated baseline curves have an amplitude of about 8 nT for the X and Z components, and about 2 arcmin for the declination. These variations are caused by slow temperature variations in the magnetometer room, which can reach several degrees per month. They are not a problem as long as regular and frequent absolute measurements are made, which has been the case ever since the installation of the observatory.

Figure 6
[13]  Another way to assess the quality of the measurements is to look at the scalar residual, that is, the difference between the field modulus directly measured by the scalar magnetometer and the modulus calculated from the vector measurement (after correction of the drift by the baseline curve). As can be seen on Figure 6, the scalar residual of hourly mean values remained less than 2 nT during the year 2006. As a result, the uncertainty on definitive values is estimated at pm 3 nT in 2006, based upon the scalar residual and the statistical distribution of absolute measurements. The same uncertainty is found for the 2004, 2005 and 2007 data. This is well within INTERMAGNET's requirement of 5 nT accuracy on definitive data.

[14]  It is also worth noting that there has been very few data gaps since the opening of the observatory: only 0.54  % of the hourly data are missing between April 2004 and December 2007, that is, less than two days per year on average.

[15]  Preliminary data are made publicly available within 72 h on INTERMAGNET's website ( Definitive data for each civil year are prepared within a couple of months after the end of the year. They can be retrieved from INTERMAGNET's website or from the website of the Bureau Central de Magnétisme Terrestre (BCMT) in IPGP ( After a final cross-check by specialists from other institutions participating in INTERMAGNET, definitive data are published on a DVD (previously a CD-ROM) together with the definitive data from the whole INTERMAGNET network. A report on the data processing and the events happening every year is published in the BCMT yearbook [Courtillot and Chulliat, 2008].


Powered by TeXWeb (Win32, v.2.0).