Spatial-temporal Modeling Of Watershed Interval Flood Warning Based On Dynamic Bayesian Networks

While Rivers and lakes bring huge water resources to the human life,its flood problem becomes a huge threat to human in the different regions of the world.Countries around the world are threatened by floods every year to varying degrees and suffered heavy losses,and China is no exception.The vast majority of floods are caused by sudden rainstorms.The severity and urgency of the flood disaster is obvious,and the flood forecasting is the primary link to effectively resist the flood.Through the flood warning,we can take the active and effective emergency measures before the flood is coming.There are many hydrological models for flood forecasting and early warning,but they need to obtain a lot of basic data and model parameters.Despite the technological advances and the updating of data acquisition methods,some data are still difficult to obtain due to the complexity,randomness and uncertainty of the flood process.Temporal GIS and spatio-temporal process are the research hotspots which have important theoretical and practical value.Dynamic Bayesian network has become a hotspot in the field of data mining for artificial intelligence.This paper mainly studies the flood time and space process of interval river basin.Through the establishment of the event graph model of the relationship between the attributes of the geographic entity,the dynamic Bayesian network statistical analysis and the function of probability prediction are used to forecast the future flood.This paper first analyzes the essence and law of the process of time and space,and clarifies the status of time,space,geographical entity,time and space process and event in the process of flood time and space;Then,the data of flood monitoring network are expressed,analyzed and pretreated;Secondly,according to the analysis of the causes,changes and trends of the geographical objects in the flood warning research,the Markov and homogeneity of the spatial and temporal changes of the geographical objects are put forward.From the perspective of application-oriented,Spatial-Temporal Attribute Graph Model in dynamic Bayesian networks;Finally,this model is applied to the sub-basin of the Huangzhuang to Shayang section of the middle and lower reaches of the Hanjiang River,and the watershed extraction and analysis are carried out in the study area,The watershed extraction and analysis of the study area can be obtained by checking the historical flood data of Shayang hydrological station of warning water level,to assess the flood warning level of Shayang;Then,the event data of Huangzhuang to Shayang watershed are applied to the flood space-time Property graphic model based on dynamic Bayesian network,and the Shayang flood warning is obtained.The model of geographic entity attribute graph based on dynamic Bayesian network is verified is available and effective.It has theoretical and practical significance for watershed flood warning under the influence of uncertain and complex factors.