GeoErnest Response to “Ask Me Something: GNSS Lever Arm Estimation”

Hi, GeoErnest !

Thanks for the public platform you offer here!
I ‘ve got two questions. The first one is about lever arm calibration. I’ve read “7.5 Tips for Maximum Accuracy”. You recommended that “ Hire a land surveyor to determine the GNSS lever arms” and “ Don’t use your post-processing Kalman filtering software to estimate it.” Directly measuring the lever arm is certainly a good choice, but almost all the post-processing softwares (e.g. POSPac from Applanix) offer the function to calibrate the lever arm using the actual data from the GNSS-Inertial system. I am a little confused about the relation between these two ways. Is it necessary to hire a land surveyor to achieve the desired accuracy? Or the post-processing Kalman filer itself can just satisfy a relative lower demand. Someone tells me that the post-processing software can refine the measured lever arm. But in my opinion, the post-processing lever arm calibration function may just be used to detect obviously wrong lever arm parameter settings. I was wondering if you could help me clarify it. It may be a great help to me.

The second question is about the POS data logging and processing. It is required that continuing logging data for 5 to 10minutes after the fight (as stated in the POSAV Training Guide). What is the post-flight data for? Is it a necessary part for the whole mission?

Thank you in advance for your help!


Hi Y.L.,

Thank you for coming to my website!

I will try my best to answer your questions. Yes, some GNSS-INS mapping software allows you to estimate the GNSS lever arms for the sensor. You can use this function, but be careful. Use the lever-arm estimation in the context of your project requirements.

I am also assuming you are employing direct georeferencing techniques with no Aerial Triangulation.

GNSS lever arm measurements are only as good as your GNSS accuracy. 
This means, they are at best 2-5 cm horizontal accuracy, and vertical accuracy is about 2 x horizontal accuracy.

What does this mean? It means the higher your accuracy requirement, the error influence of the GNSS estimation will be larger. My personal threshold is if you require better than 10 cm ground accuracy, then you should consider using land survey techniques to measure the GNSS lever arm.

For example:
– If your project requires 15 cm ground accuracy, then the GNSS lever arm estimation will fall within the noise. You could use the software to estimate.

– If your project requires 10 cm ground accuracy, then the GNSS lever arm estimation will fall within the noise. You can use the software to estimate, but you may begin to see error influences.

– If your project requires 5 cm ground accuracy (such as vertical LIDAR), then the you should use land survey techniques to measure your lever arms because your lever arm estimation error will influence your final accuracy.

Another example:
– If your GNSS lever arm has a 3 cm constant measurement error along one axis, then you will see a systematic shift of your data in that same direction. You will notice the error if your accuracy requirements are 5 cm, but you may not see the error if your accuracy requirements are 15 cm.

Yes, you can use the software to estimate obvious GNSS lever arms errors! This is a software benefit to reduce human input errors!

If you have a set of GNSS lever arms measured by land survey (and you trust them), then input them into the software, and lock them. At this point, there is no need to use GNSS lever arm estimation.

The 5-10 minutes of pre and post flight data logging is primarily to gather high-quality GNSS data. The GNSS-INS Kalman filter needs more data and redundancy to minimize positional errors.

Yes, you should log 5-10 minutes of data before and after. The reason why you gather data before AND after the mission is because the software will try to establish the highest position accuracy immediately. Thus, it will process “forward”, then independently process “backwards”, and combine them to get the best solution from the two processes.

Why is 5-10 minutes suggested?
If you tell some pilots to log for 10 minutes, they will log for five minutes.
If you tell them five min, then they will log for three minutes.

In reality, 10 minutes is probably too long, and 3-5 minutes is sufficient. Just be careful what you tell your pilots :)

About GeoErnest
This is GeoErnest. Visit

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: