The objective function based on CO2 loading results in smaller
values for SSQRE and AARD% due to its smaller magnitude and
range. In the work of (Tong et al., 2012) an AARD% of 1.9% resulted.
If the same objective function is used with the model in this work
the resulting AARD% for the (Tong et al., 2012) data set is 4.3% and
for both data sets together is 3.0%. If a single temperature independent
interaction parameter is introduced between the ionic species
present at the largest concentration, AMPH+ and HCO3
−, the AARD%
of fitting the (Tong et al., 2012) data is reduced to 2.4%. The estimated
error associated with this parameter is large (40%) and it is
questionable if its inclusion can be justified. When fitting the data
using partial pressure as a function of loading including this interaction
parameter results in no improvement in the AARD%. What
is clear is that a 14 parameter model is over-fitting and cannot be
justified.