Rather than dealing with question of appropriateness of skew and kurtosis, it is more important to investigate what brought about the bimodal distribution. Can you break up the data into two individual distributions? Perhaps one distribution came from data collection from one particular shift in a factory and another form a night shift, for example and there might have systematic data collection errors (or biases) associated with one over the other?

The sand samples are old sand sediments, i think this bimodale behavioir is due to sampling conditions, because in deltaic sediment we should be aware from the normal and inverse grading sediment, when we took the sample we dig about 30 cm, so the sample is not homogenes.

Are this curve bimodale or unimodale? I am little bit confuse.