STA logger location accuracy
It is a common question; “How accurate is the STA logger’s location?”. The response is usually a regurgitation of the technical specifications of the GPS module which states an average autonomous horizontal accuracy of <2.5 metres. However, we recognise and fully admit that this response can be somewhat unsatisfying for those looking for real world answers. So we’ve set out to remedy this by performing some field tests.
What follows is a brief test of the STA loggers location accuracy when compared to other location devices. There are a range of factors that influence a receivers accuracy so the tests conducted here are only as good as the conditions that they are performed under. Over time we hope to perform more tests to demonstrate the STA loggers accuracy in different scenarios.
The technical specifications
First off, what does the documentation say?
For those that aren’t familiar with satellite systems, GPS stands for Global Position System and is the US name for their satellite constellation. Technically, a “GPS” is a “GNSS”, or Global Navigation Satellite System. The term GPS just stuck from the US’ version and is colloquially used by most people today. The STA logger receives signal from GPS (USA), GALILEO (EU) and SBAS (where available) meaning it has more options to acquire signal than just a GPS. Nevertheless, we will use the generic term “GPS” to mean “GNSS” even though the STA logger receives more than just the GPS signal.
The unit records at 1 Hz by default (1 point per second). From a cold start, the unit takes on average 26 seconds to get an adequate signal, however, once started in the morning, it will reacquire a lock from a hot start within 1 second.
All of this is secondary information. What really matters is the horizontal accuracy which, as stated above, is on average <2.5 m.
Lets put it to the test!
Location Accuracy Testing
Without getting into the weeds about it, lets just say that most location measurements come with a degree of uncertainty and take a lot of work to get the uncertainty value as small as possible. This is what the field of surveying is all about. There are survey locations plotted around the landscape, but they are best for stationary tests where you watch the location from a receiver wobble around the ‘known’ location. We may update this blog with a stationary test in the future, but what we are really interested in is a test of the location accuracy while the device is moving through the landscape, just like how a weed spraying operator might be moving.
For this test we will compare the STA logger to a conventional mobile phone receiver and an RTK (real-time kinematic) receiver. The RTK device will be considered our “truth”, though in reality it has a horizontal accuracy in the range of 3-5 cm (possibly higher for our test, but lets just consider it as ‘very accurate’). The mobile phone is considered equivalent to what many others may be using in the field. Despite what you might think, modern mobile phone GPS’ are very good and are usually able to get ~3 m horizontal accuracy in good conditions. So roughly equivalent to the STA logger and what you would expect from consumer grade GPS devices.
- Date: 13/01/2021
- Location: Victoria, Australia
- Cloud coverage: 10-20% high cloud cover
- Obstructions: Parkland with very little tree cover
- Method: All equipment was given 10 minutes to gain a satellite fix. All units were walked along a footpath. The pattern was: walk down the centre of a path, return half way, deviate down a side path, return, then return to starting point. Total distance is approximately 587 m over 7 minutes (~5 km/h).
- Mapping Application: SW Maps, version 2.7.2
- Corrections Application: Lefebure NTRIP Client (last update 11/03/2020)
- Corrections: AUSCORS NTRIP Broadcaster
- Data stream: WOTG00AUS0 (baseline approximately 20 km)
- Coordinate system: GDA2020 MGA Zone 55
- Device: Google Pixel 3
- Application: ArcGIS Collector, version 20.2.2
- Coordinate system: GCS WGS 1984
The path was walked as described above with all 3 devices tracking the path through their respective software. Upon completion, data was consolidated onto one map, reprojected where required and visually compared to one another with the RTK result being considered the most accurate.
Average distances from the RTK line were measured laterally (perpendicular to the path) from 20 locations along the path at an equal interval. We understand that this isn’t a true measure of the horizontal accuracy as it isn’t measuring the distance from point to point. But it was useful enough for the purposes of this first publication. To differentiate it, we’ll call it “lateral accuracy”.
- The lateral accuracy of the STA logger was, on average 0.85 m and never exceeded 2.27 m from the RTK line.
- The lateral accuracy of the mobile phone was, on average 1.42 m and never exceeded 5.18 m from the RTK line.
Lets remember that the RTK line has it’s own level of variability but it was negligible (<0.1 m).
The STA logger out performed the mobile phone in both maximum distance and average distance from the RTK line. It is apparent from the plot above that the STA logger was more consistent than the mobile phone which tended to dart from side to side and occasionally veer away from the path. The STA logger, tended to follow the same pattern as the RTK line, even though it was offset from the line by about 80 cm on average. The obvious outcome is that RTK is very good, but this requires expensive hardware, specialist knowledge, set up time, and in our configuration, mobile phone network coverage. RTK just simply isn’t feasible in most situations which is why the STA logger was designed as it is – to be as accurate as possible while maintaining automation and simplicity.
This testing was conducted without the assistance of a Satellite Based Augmentation System (SBAS) which is available in many areas and is expected to improve accuracy even further.
We hope to perform more tests with more rigorous methods and a greater variety of conditions in the future.