Closed regan-sarwas closed 7 years ago
With closer inspection
select * from CollarFixes as F
left join CollarFiles as C on C.FileId = F.FileId
left join CollarFiles as P on C.parentFileId = P.FileId
--left join CollarDataIridiumMail as I on I.FileId = F.FileId
where fixDate > '2016-01-01' and Lon > -100 and C.Format = 'C' and P.Format = 'I'
order by Lon, Lat
Lon is between 1 and 19.75 (inclusive although only 11 out of 1890 are above 5.00) Lat is between 0 and 90 (inclusive, integer values only)
All the bogus Telonics/Iridium fixes have a percent value (i.e. 66.3%) in the ArgosLatitude column of the CollarDataTelonicsGen4 table. Otherwise they have either null or a numeric value. It is also true they have a non-date value in the GPSFixTime column. There are 2506 of these records (as of this posting). There are less locations, because some of these fixes had lat/long that were out of the permitted range, or an AcquisitionTime that was after the upload date.
The bogus fixes were deleted from the CollarFixes table (with a trigger to fix the Location table). And the CollarFixes_Insert SPROC was fixed to ignore these records going forward.
Close with 71069ad26a04be302da8829cacd185eec2414978
It appears that Iridium collars transmit battery or other system info in the location fields about every 2 weeks. The latitude is typically between 40 and 70, while the Longitude is between 0 and 5.
By their value they are obviously not locations in Alaska, so they are manually hidden as they are discovered. This is problematic, because:
I need to find an automated way to reject these as locations before they are added to the Locations table.