Abstract: Comparing the computation made by the thermodynamic and hydrologic equations. it is known that an improved Kuo-type Krishnamurti regression scheme is an insufficient application for the mid-and-lower reaches of the Yangtze River Valley. An improved Kuo type parameterization scheme, therefore, is designed in view of the fact that the local cumulus and stratiform clouds interact in coexistence. It is characterized by 1) the criteria for the generation of cumulus convection is regional; 2) Krishnamurti's convective parameters, which are strongly physically significant are obtained directly from realtime data rather than statistic regression, in addition to the easy applicability seen in the Kuo-type; and 3) the processes are considered physically to incorporate cumulu-scale convection (interaction between cumulus and stratiform clouds) into the parameters without the presence of any microphysical equations. The computation shows that the improved scheme has great positive effects on diagnostic analyses.