Analysis Of Atmospheric Motion Based On Dynamic Mode Decomposition

It is very important to accurately describe and understand the structural changes and instability mechanism of complex atmospheric fluid movement in the support of meteorological environment prediction and atmospheric science research.However,with the rapid development of computer technology and the continuous improvement of the performance of meteorological observation instruments in recent decades,it is still very difficult to accurately and quantitatively extract the motion structure of complex atmospheric fluid.On the other hand,in the stability analysis of atmospheric flow,it is necessary to greatly simplify the partial differential equations that control the atmospheric motion or calculate the inverse and eigenvalues of the large matrix.As is known to all,the calculation cost of large matrix is huge,and there is no universal method in the stability analysis of atmospheric motion.Through Dynamic Mode Decomposition(DMD)method to analyze the high-dimensional data sequence of atmosphere/ocean,not only can the Mode representing the spatial structure characteristics of the flow field be obtained,but also the Mode growth/decay rate can be obtained through the characteristic value corresponding to the Mode to judge the stability of the flow field.Secondly,a dynamic reduced order model of atmospheric/oceanic flow field evolution can be established to reconstruct and predict the atmospheric/oceanic flow field.In this paper,the DMD method is introduced into the analysis of atmospheric motion data in order to improve the understanding of atmospheric motion characteristics.First,adopt the dmo method and Proper Orthogonal Decomposition(the Proper Orthogonal Decomposition,POD)method for the 200 h Pa air jet flow field in the modal analysis,the modal of space and time were decomposed by two methods are compared,and qualitative contrast the characteristics of two kinds of mode Decomposition method,and method of DMD and POD,a reduced order model of error analysis to obtain the flow field to test the dmo method on the analysis of air flow field effect.Secondly,the improved Sparsity promotion DMD method(Sparsity promoted Dynamic Mode Decomposition(SPDMD)was applied to conduct modal analysis of the 200 h Pa atmospheric jet motion flow field,so as to realize the selection of the optimal Mode of the flow field and conduct stability analysis of the flow field.At the same time,the multi-resolution Dynamic Mode Decomposition(mr DMD)algorithm is applied to analyze the global sea level temperature data set from 1990 to 2016 to realize the extraction of Dynamic characteristics of different frequencies in the long time series.Through experiments and studies,the following conclusions are drawn:(1)through the DMD decomposition method can get the air flow field of different modal characteristic values of the corresponding mode eigenvalue real part reflect the flow field in energy/strength in the time of growth and decay,the imaginary part of characteristic value reflects the morphology of the flow field structure in the aspect of time,through the analysis of the characteristic value can stability assessment of the atmospheric motion mode,and POD method cannot determine the stability of the modal.Moreover,the DMD method can effectively obtain the structural information of each frequency motion mode in multi-scale atmospheric motion,and the POD mode corresponds to the mixed frequency.(2)the dmo method decomposed different modal represents the motion law of the flow field in different spatial scales,zero frequency mode represents the basic flow of air flow field structure and atmospheric movement on this basic flow structure in the flow field can be viewed as dynamic modal superposition of different frequency,as a result,the complex flow problem is converted into a number of main flow mode of evolution over time.A reduced order model of 200 h Pa jet flow field was established based on DMD and POD method.Compared with POD method,DMD method can use fewer modes to reconstruct and predict the flow field with higher accuracy.(3)SPDMD method can select the optimal mode of the flow field and realize the flow field stability analysis;The mr DMD method can accurately identify El Nino and La Nina events with intermittent warming and cooling associated with global weather and climate change,further confirming the ability of dynamic modal methods to analyze complex atmospheric/oceanic flow fields.