Although the underprediction bias for the highest concentrations was eliminated, more of the lower concentrations were underpredicted. The overall model performance was reduced, such as the correlation or the number of pairs within a factor of two.
In this example the performance statistics are unchanged because the lowest 100 m are relatively well mixed and therefore variations in the concentration layer depth have little effect.
In this example, although the performance statistics are still not quite as good as the original calculation, the substantial number of underpredicted values have been reduced and the peak concentration predictions are quite close to the measured values.