We are getting detections on our Sensorstation (deployment 10860) that are being caused by glitched detections of tagged animals living within our node array. These are unregistered tags that only differ by a few characters from registered tags, and are detected by the same node detecting the true tag ID. For example we have tag ‘7878072D’, a brown pelican, getting detected instead of ‘7852072D’, a lizard that lives at our site. It is incredibly unlikely that this bird flew from the coast to our inland desert site. Only a few of these detections show up in the motus explore page.
I was mostly wondering how I can flag these so that they are not being counted as proper detections.
It looks like this is a case of signal corruption where one or two bits of the code are misrepresented. Usually this is more likely to happen at low light when the transmitter may be underpowered (assuming these are solar tags). This actually is not that uncommon, and if you watch the SS UI live you’ll see all sorts of errant “hits” that aren’t actually valid tags, most of which are from other sources, but some of which may be related to the signals of real tags.
When detection data is sent to CTT’s servers, it’s filtered against the list of tags they have actually provisioned. (CTT would be able to comment on the exact details, but that’s the gist anyway.) If the hits match real tags, then they are passed on to the Motus database.
In this case, not only do those errant hits match real tags, they also match real tags with active deployments. That there are SO many such false detections of tags with active deployment is certainly unusual. This could be because you have a very large number of tags present; the more tags that are present and active, the greater the chance that some of those signals will be corrupted and appear as other tags.
The reason that none of your tags are detected is because they are not registered to this project or have active deployment records. Data from tags without deployments is filtered out on the webpage.
Yup, that all makes sense and was how we’ve been interpreting it. We have quite a few unregistered tags on lizards at our sites and did not expect (or want) them to show up in the motus system.
We were mostly wondering if there was a way we could flag the likely errant detections (Pelicans/shearwaters) in the motus system.
Ok good to know this is all as expected (well, sort of!)
We do have the option to manually flag false positives. You can read about that here: Public Data Filters | Motus Docs. It’s a very blunt tool, and will hide any past or future detection of any of those tags deployments at this station deployment. Maybe it’s possible some of these will be fly-bys at some point! It does only hide them from the website; all detection data regardless of the filter applied is available through the R Package. Let us know what you think.
Out of curiosity, how many tags do you actually have present in the area?
Great to know the process for this! I would be pretty surprised if we were to get oceanic birds at our site, but I suppose it could happen if we get an unusual storm.
We have over 30 tags currently deployed within our node array, though not all of the hybridtags are typically charged up at a time.
Do you think that any of those detections are valid, or are they all errant variants of the non-registered tags you have active? Just wondering if we should flag everything or exlucde some.
If you have a list of any that you think might be valid and should be retained, please let us know.
I think any detections of oceanic birds (Brown Pelican, shearwaters), or birds never reported in Fresno County (pipit, woodcock) are almost certainly errant variants of our deployed tags. Most of our other detections seem to be actual birds/bats, especially as most have been detected by neighboring stations.