What was done
Using "revisit simulations contributed by six of the Atmospheric Model Intercomparison Project 1 participating modeling groups," the authors "examine the impacts of model revisions, particularly the land surface representations, on soil moisture simulations, by comparing the simulations to actual soil moisture observations."

What was learned
In summarizing their findings, the authors state "the revised models do not show any [our italics] systematic improvement in their ability to simulate observed seasonal variations of soil moisture over the regions studied."  And if those words are not clear enough, they also state "there are no [our italics] indications of conceptually more realistic land surface representations producing better soil moisture simulations in the revised climate models."  In addition, they say there is a "tendency toward unrealistic summer drying in several models," which they note is "particularly relevant in view of the summer desiccation projected by GCMs considered in future assessments of climate change."

What it means
Although the authors note that "simpler land-surface parameterization schemes are being replaced by conceptually realistic treatments," as the climate modeling enterprise moves (supposedly) ever forward, they state that "improvements gained by such changes are, however, not very apparent."  Our translation: that which does not exist, i.e., model improvements, are very hard to see.

These findings are truly astounding, even to us.  We would have thought that in an avowed attempt to improve this particular aspect of climate modeling there would have been some improvement; and we surely believe it will yet occur.  But these are the conclusions of those who have studied the subject; and for now (meaning the publication date of the journal article - 16 November 2000), we are forced to take them at their word.  There has been no real progress, only attempted progress.