This paper discussed the integration of model and GIS. It is shown that the integrated expression and the processes of the object state model

the object simulative model and the object analytic model are key issues for integration. An integration scheme based on metadata and metamodel is suggested. The characteristic of the derivation nets deducted from petri nets was analyzed and a new derivation+ nets was presented to express the evolution history of metadata and the adjustment of metamodel. Then an integration scheme based on derivation+ nets

which is used to express the data process flow

was proposed. As a result

a derivation+ net of the remote sensing classifier of BP neural nets is made to illustrate the strategy of the integration of model and GIS.