Magnetic Bias Compensation
Magnetic Bias Compensation for SeaTag Geo-Positioning
The magnetic field intensity readings taken by SeaTag-MOD to determine tag latitude may exhibit a bias that can be caused by multiple factors including calibration errors, the magnetic field of the battery section (tags are calibrated without the battery section), changing self-magnetization of the tag (exposure to strong magnetic fields), temperature variations and others. Experiments have shown that for a given set of deployment conditions, this bias holds largely static. That is, about the same magnetic bias will be experienced throughout a tag’s time on-animal.
A simple method to measure, apply and verify the validity of the magnetic bias or offset can be used to improve tag position estimation accuracy. The following example explains.
Measure the magnetic offset
The magnetic bias can be measured or established by comparing the magnetic field intensity reported by the tag for the day of deployment to the predicted magnetic field value for the GPS determined deployment location. The tag measured daily average magnetic field intensity value is taken from the SDPT_MODDAILY packet, specifically the value reported in field magnetic field amplitude avg (nT).
It is a good idea to plot this parameter for the entire time on-animal, as a first quick-look to determine if reasonable trends can be seen or if the data is noisy. The graph shows the magnetic trend for a fish tagged near Guatemala. The tagging day was 3/20/2012, and the magnetic field strength reported in the SDPT_MODDAILY packet for that day was 35160nT. The last day on-animal was 4/5/2012, with a field value of 34140nT reported.
Side note: Notice how the reported magnetic field value jumped up about 300nT on the second day on-animal, and again on the day before the last day on-animal. This may be real animal motion, but it could also be a result of the tag being some of the time off-animal on both 3/20/2012 and 4/5/2012. So, look for deployment and pop-off time associated spikes or outliers in the data, and consider using the adjacent data points for bias computation instead. In this case, the spikes are small enough that we will consider the data of 3/20/2012 and 4/5/2012
Daily average earth magnetic field strength reported by a SeaTag-MOD during and after time on-animal
In this example, the quick-look of the magnetic data looks good. Starting on 3/21/2012 and continuing to the day before pop-off on 4/4/2012 we see a clear trend towards lower field values from 35500nT down to 33680nT, a span of -1820nT. The data looks ‘low noise’ with the possible exception of jumps of around 400nT associated with the first and last day on animal.
Next, we use an on-line geo-magnetic calculator to determine the predicted magnetic field value for the tagging location and time. In this example, the fish was tagged on 3/20/2012, at GPS position 9.07N ; 84.72W. We will do that with the NOAA geomagnetic calculator available here: link
NOAA Geomagnetic calculator results for GPS recorded tagging time and location
To obtain results, we enter the tagging GPS location and date. We select the IGRF11 magnetic field model, that being the model also used by the SeaTrack software. The predicted field strength for the tagging location is 34494nT. This compares to the tag’s reported field strength for 3/20/2012 of 35160nT. To obtain the magnetic bias, subtract the tag measured value from the calculator value: 34494nT – 35160nT = -666nT.
Applying the magnetic offset
The animal’s position track can now be computed with SeaTrack, using the offset of -666nT to improve accuracy. The screen plot shows the SeaTrack settings used for this example. Note the magnetic offset application. Also applied was a three-day position average for some position track smoothing, and rejection of any ‘short days’ that were detected as being 0 minutes or more shorter than the predicted sunrise-to-sunset duration for the location. The later setting can remove positions with potentially inaccurate noon time detections, i.e. longitude outlier removal.
Verifying the Validity of the Magnetic Offset
The validity of the applied magnetic offset can be verified by noting the latitude discrepancy after tag pop-off from the animal. Once the tag returns to the surface, ARGOS position fixes will be available, and can be compared to the tag’s position estimation. In general, the recommendation is to compare the first reliable post pop-off ARGOS position fix to the last tag position estimate prior to pop-off. In this example however, ARGOS data was not available for the pop-off day (4/6/2012). So, we used data from 4/15/2012 when this shedded tag was at the surface and still recording light/magnetic observations while also transmitting messages to ARGOS. The map shows the results, including the GPS tagging location of 3/20/2012 and the ARGOS position fix of 4/15/2012 shown as yellow push-pins. The green circles indicate the light/magnetic position estimates.
Tag position plot with -666nT magnetic offset compensation
The light/magnetic position estimate for 3/20/2012, the tags first day on-fish, shows the tag about 20 nautical miles north and about 75 nautical miles east of the GPS recorded tagging location. In fact, both positions are on the same magnetic field intensity line as forced by the -666nT offset value. (The significant longitude position error for 3/20/2012 is due to an approximately 5-minute noon time estimation error).
Verification of the magnetic offset value is available in the ARGOS fix and magnetic/light position estimate for 4/15/2012. Here, the tag’s light/magnetic position estimate is about 20 nautical miles straight west of the ARGOS position fix. Using the NOAA geomagnetic calculator again, the magnetic field strength for the tag estimated position and the ARGOS fixed position is essentially the same, implying that the -666nT offset compensation computed for the day of tagging still held valid for the end of the track.
Results without magnetic bias compensation
This map compares results without the magnetic offset compensation applied. The tag position estimate for 3/20/2012 is displaced about 100 nautical miles north of the GPS tagging location fix. The discrepancy at the end of track is about 70 nautical miles north. The results are obviously not as good as with magnetic offset compensation applied.
Tag position plot without magnetic offset compensation
The results seen in this example are roughly representative of other SeaTag-MOD data sets we have studied.