I am looking at residual diagnostic terms in a tidal application. I use a very strong low pass filter and the results, especially for u_accel are small compared to the tidal values. I get a very small negative bias on the u_accel term which becomes important in the residual. The bias is present in the full tidal values so I don't think it is due to the filter. It's easy to see if you just take the mean of u_accel over a lot of tidal cycles. If I back calculate accel from du/dt and low pass filter them both they track each other but not exactly so the bias changes slightly over time. If the order of u_accel is 10e-4 then the residual will be 10e-6 and the bias 10e-7 so it is 3 oom less than the signal itself. There are similar biases in the other uv diagnostic terms to they do all balance. Has anyone else seen this or know where it may come from. I am still running 2.2.
Thanks Wayne