search

joe045 / wave_sensors_one_ocean_expedition_2023

Processing of the wave data obtained during the One Ocean Expedition
0 stars 0 forks source link

Effect of corrections #2

Closed joe045 closed 1 year ago

joe045 commented 1 year ago

Hei @jerabaul29 and @lahole,

Now that the data are better, I had a look at the effect of two corrections, because they require a lot of post-processing.

  1. Using the extra IMUs compared to the main IMU
  2. Correcting the roll, pitch, and yaw values.

In my opinion, the effect of the corrections is only significant for ug1, not for ug2 and radar. I have only looked at the data from October 2021 and may need to look at more data, but I don't want to spend too much time on it.

Would it be an idea to simplify the system to only use main IMU and no corrections? Or should I continue to use the system with the corrections, since they are already done?

effect_corrections.pdf

jerabaul29 commented 1 year ago

I think this shows that the IMU1 is harder to use, likely because being at a very "exposed" location where there is a lot of motion etc, so there are also a lot of large amplitude effects to correct for, which is harder. I think focusing on ug2 and / or radar is the best then, according to your figures. I think if I understand correctly that using ug2 and radar with correction is what works best and should be used? :) . But all of this is only my 2 cents, and I may be misunderstanding.

lahole commented 1 year ago

I think you should continue with the simplified system with ug2/radar and the main IMU.

joe045 commented 1 year ago

For ug2 and radar these corrections does not really improve the results. For ug1 the corrections improve the results a bit, for example around 2021-10-17 .

If I continue with the simplified system it will probably reduce the post-processing by 50%.

jerabaul29 commented 1 year ago

I think both alternatives are fine as long as you show and discuss and explain what you do :) . I suppose the post processing CPU cost is not that high, so it is not so much cost in including it? (or were you thinking about "working hour costs"? :) ).

lahole commented 1 year ago

I think @joe045 meant working hours cost. I think she should now focus on completing the thesis, so I agree with her.

joe045 commented 1 year ago

For me the working hours are already put down, so it does not matter what I choose. I was thinking more of making it easier for future work. It makes more sense to present the simplest method when it has similar results as the more demanding method.

Here are some numbers of the mean bias for SWH for the period in the figures above. The bias is very small for ug2 and radar, but considerable for ug1 (0.5m).

corrections.pdf