Tables
Table S1. Error covariance parameters for the GIM inversion. An
exponential function was used to define the prior error covariance
matrix with monthly varying sill variance values, two temporal
correlation lengths (constant in time), and one spatial correlation
length (constant in time). The two temporal correlation length values
were combined with the two model-data mismatch variance values to run
four inversions which were then averaged to produce the estimates in the
study.