Granularity Structure Analysis And Object-oriented Modeling Simulation Of Xinanjiang Watershed Model

"Pan-spatial Information System and Intelligent Facilities Management" of National Key R&D Program of China has launched in 2016.The program attempts to break through the traditional GIS indirect modeling method using maps as templates,and instead uses multi-granularity spatio-temporal objects to directly describe the real world from micro to macro.Unlike traditional GIS,which mainly uses “space+ geometry+ attribute” for element expression,multi-granularity spatiotemporal object modeling uses an eight expression elements model(spatiotemporal reference,spatiotemporal location,spatial morphology,attribute characteristics,composition structure,behavioral ability,cognitive ability)and five sets of operations model(construction and destruction,decomposition and combination,transformation and evolution,relationship dynamics,learning and decision-making)to achieve.The changes in modeling principles and methods make the existing GIS-based description of geographic phenomena/geographic processes and the series of geographic models developed based on them require proper conversion and reconstruction before they can be applied to full-spatial information systems.Select typical and representative geographic models to model multi-granular spatio-temporal objects,and explore the key technical methods in the process of geographic model transformation and reconstruction to verify the effectiveness of multi-granular spatio-temporal object modeling theory.The applicability of modeling and analysis of complex geographic processes has good reference significance.The hydrological model in geography involves the evolution and interaction of multiple geographic objects.Using multi-granular spatial-temporal object modeling methods to model the geographic objects in the hydrological model can better reflect the evolution characteristics and associations of geographic objects Relationship,and show the superiority of multi-granular spatiotemporal object modeling.Taking the Xinanjiang Watershed Model for the prototype,the author explored to abstract and generalize geographic spatiotemporal objects from eight aspects: spatiotemporal location,geometric shape,attribute characteristics,spatiotemporal reference,attribute characteristics,association relationship,behavioral ability,and cognitive ability,establishing the expression system of spatial-temporal object in thing space,and a knowledge-based granularity hierarchy and classification system.At the same time,the author studies how to abstract the hydrological elements of the watershed and excavate the evolution law of the hydrological process from the perspective of the Scenario Science,so as to realize multi-granularity spatio-temporal object modeling of the hydrological model under Pan-spatial Information System.Finally,the author combines Bayesian networks to explore the evolution rules of watershed objects,studying the reasoning evolution process of multi-granularity spatiotemporal objects,and implementing the calculation model of multi-granularity spatiotemporal objects.The main research contents and innovation are as follows:(1)The thesis applies the multi-granularity spatio-temporal object modeling method to construct a hydrological process scenario based on the theory of scenario science.The author analyzes the hydrological elements mainly including atmospheric water,surface water,soil midstream and groundwater,and constructs granularity hierarchy of watershed hydrological object according to the evolution mechanism and logical relationship of the hydrological process.And explores the interaction relationship and state evolution law of the hydrological objects in the watershed,which lays the foundation for the modeling of multi-granular spatiotemporal objects.(2)On the basis of constructing the granularity hierarchy of hydrological objects in the process of hydrological process scenarios,and taking the watershed as the top-level object,taking the evaporation surface,runoff surface and confluence surface are abstracted as sub-objects,to construct the a multi-granularity spatio-temporal object description framework for the hydrological model under the pan-spatial information system,and to model the watershed object from the aspects of spatiotemporal location,attribute characteristics,and association relationships.The research realizes the application of multi-granularity spatio-temporal object modeling of Pan-spatial Information System in the watershed hydrology field.(3)Combining with the Bayesian network,the author establishes the association relationship of multi-granularity spatio-temporal object in the watershed.According to the basic formula of Bayes' law,the quantitative description of the relationship between objects is studied to realize the reasoning of the evolution process of multi-granularity spatio-temporal objects.According to the prior probability distribution,the river basin flow is predicted,and the calculation model of multi-granularity spatio-temporal objects is established.