Save misfits at analysis times
Saving misfits would help in diagnostics at analyses times.
The formula is:
- For U and ER:
\chi_x = \left[\frac{1}{N_e}\sum_k (\mathbf{x}^a_k(t) - \mathbf{x}_0)^T P_{xx}^{-1} (\mathbf{x}^a_k(t) - \mathbf{x}_0)\right]^{1/2}
with\mathbf{x}_0
the average prior. - For MF and SV:
\chi_y = \left[\frac{1}{N_e}\sum_k (\mathbf{y}^o_k(t) - \mathrm{H}_k\mathbf{z}^a_k)^T R_{yy}^{-1} (\mathbf{y}^o_k(t) - \mathrm{H}_k\mathbf{z}^a_k)\right]^{1/2}
where\mathbf{z}
is[\mathbf{u}\mathbf{e}_r]
(SV) or\mathbf{b}
(MF).