Salinity gradient vs. temperature gradient

Figure 1. The salinity gradient and temperature 
gradient observed in an open cup of saturated 
saltwater.

This is the fifth follow-up of my blog article: "A perfect storm in a cup of salt water?" This investigation focused on the relationship between the salinity gradient and the temperature gradient. Is the temperature gradient caused by the salinity gradient, or the other way around? Both arguments seem to make some sense. On the one hand, one can argue that the salinity gradient stops the convection. On the other hand, warmer water tends to dissolve more salt. So we are in a chicken-egg situation.

Let's do an experiment to explore a bit further. I prepared two cups of saturated saltwater. One open and the other sealed. I let them sit overnight and then checked the salinity and temperature distribution the next day using Vernier's salinity sensor and temperature sensor. I did this by moving the salinity sensor and the temperature sensor together up and down in the saltwater. Figure 1 shows the results for the open cup.

Figure 2. The salinity gradient and temperature
gradient observed in a closed cup of saturated
saltwater. Note: The measurement was done
shortly after removing the seal.  Hence the results 
can be regarded as approximately those of the
sealed cup as the gradients will take a longer 
while to establish.

To measure the data for the closed cup, I first removed the seal and then quickly did the measurement. Since the salinity and temperature gradient would take some time to re-adjust after the seal was removed, we can pretty much assume that the results I got approximately reflect what would have been measured if the seal had not been removed. Figure 2 shows the results.

 The comparison of the results shows that the salinity gradient is about the same for the open and closed cup--the bottom is about 1.3 ppt saltier than the top, but the temperature gradients are quite different--the open cup measured about three times as large as the closed cup (0.3°C vs. 0.1°C). 

Due to the evaporative cooling effect, the overall temperature of the open cup is at least 0.5°C lower than the closed one.

What do these results suggest? Is it possible that a weak temperature gradient exists in a closed system that does not have the driving force of evaporative updraft?