Skip to main content
. 2021 Aug 12;11:16384. doi: 10.1038/s41598-021-92927-0

Table 2.

The assimilation parameters which have the most impact on the assimilation and the values used for this reanalysis. They are set in fortran namelists within the input.nml file.

Parameter Value Description
filter_kind 1 Ensemble Adjustment Kalman Filter (EAKF)13
ens_size 80 Number of ensemble members
num_groups 1 Number of groups into which members are divided
inf_flavor 5 Enhanced, spatially and temporally varying, adaptive, prior, covariance inflation14.
inf_lower_bound 0.0 Inflation mean lower bound
inf_upper_bound 100.0 Inflation mean upper bound (is never reached)
inf_sd_lower_bound 0.6 Inflation standard deviation lower bound
inf_sd_max_change 1.05 Inflation standard deviation maximum change in an assimilation cycle
inf_damping 0.9 Inflation damping (acts opposite of inflation to enhance temporal adaptiveness)15
horiz_dist_only .false. Include the vertical distance in calculations of the distance between observations and state variables
cutoff 0.15 Half width (in radians) of the spatial localization function16
vert_normalization_scale_height 1.5 Scaling factor relating vertical distance to horizontal in localization calculations. 1.5 scale heights = 1 radian in the horizontal
sampling_error_correction .true. Apply a correction to account for the limited ensemble size17
input_qc_threshold 3.0 Reject observations having a quality control value larger than this
outlier_threshold 3.0 Reject observations which are more than this number of total spread standard deviations different from the ensemble mean