Abstract: The cloud model atomization process was related to a larger hyper enctropy. Through statistical analysis of the overall trend of the cloud drops, the cloud drops dispersed over the course of hyper entropy increase. By analyzing the dispersion trend of cloud drops in each semantic ranges, it is indicated that the drops represent the core concept dispersed in a low speed. The atomization feature of the cloud model was summarized. The semantic range of the concept represented by the cloud model extended while the hyper entropy increased step by step. The cloud drops spread but the drops nearby the core semantic keep a high density. The atomization feature of the cloud model was used to model the data deviates from the normal distribution. A cloud with a large hyper entropy value represented the concept lack of consensus. For each parameter, the expectation stands for the core semantic value, the entropy represents the semantic range and the hyper entropy shows the degree of consensus of the different semantics ranges. The cloud model knowledge representation application range was extended.