Study On Atmospheric Rivers Using Graph Theory

Moisture advection and transport in the atmosphere is the most active process in the global water cycle.The characteristics of water vapor transport affect the formation and distribution of precipitation,and exert a far-reaching influence on water resources and ecological environment all over the world.However,there is a lack of quantitative tools to describe the pattern of water vapor transport from the perspective of hydrological cycle.To address the problem,this thesis proposes a conceptual model,namely,the water vapor transport network,to use the mathematical graph structure to describe the complex water vapor transport pattern.The research of the water vapor transport network is based on tracking the Lagrangian trajectory of each water vapor parcel.In order to establish clear correspondence between water vapor evaporation and precipitation along each trajectory,we proposed the concept of unit trajectory and the concept of water vapor quantity along it,as well as the method to calculate the actual contribution of water vapor to precipitation on each trajectory and to identify the source locations of water vapor,thus,providing clear linkages between water input and output.Using ERA-Interim reanalysis data,we calculated the unit trajectories related to the precipitation over Sanjiangyuan region and strong rainfall events all around the world.The macroscopical pattern of water vapor transport is represented by the collection of massive trajectories.In this thesis,a semi-structured scatter points aggregation method is therefore proposed to convert all trajectories into a unified structure of water vapor transport network that maintains the water mass conservation during the conversion.The water vaper network can represent the actual physical relationship among evaporation,precipitation and transportation of water vapor.Using the proposed method,we established the water vapor transport network flowing into the Sanjiangyuan region,and analyzed the water vapor sources and transport characteristics for this region.We found that the structure of the network is highly related with geomorphic characteristics,and the contribution rate of local evaporation to precipitation is about 9.15%,basically consistent with the relevant conclusions from previous studies.Moreover,the global water vapor transport network related to worldwide strong rainfall events also demonstrates that this network model is suitable for quantitative analysis of macroscopical patterns and spatiotemporal variability of water vapor transport.We also studied the large-scale regions that are relatively independent in the atmospheric water cycle system,termed atmospheric basins.We concluded that within the theoretical framework of water vapor transport networks,the extraction of atmospheric basins can be done with a graph partitioning method in graph theory.The division of global atmospheric basins yields a result that reflects the differences in water vapor transport mechanisms for various regions and is meteorologically interpretable.The model of water vapor transport network proposed in this study has a clear physical meaning,holds quantitative relationships of water mass,and is also extensible.Powerful mathematical tools of graph theory can help to realize diverse applications,providing a new tool for the study of atmospheric water vapor transport.