The Mixed Sampling Frequency Analysis Of Animal Movement Data

The study of animals’ random walks not only helps to improve the understanding of predation efficiency,but also plays a crucial role in the path search of complex networks.Aiming at the problem of misidentification in distinguishing different types of random walk models,researchers currently move to the study of sampling rate for the process of random walks.The ordinary method of sampling is usually based on a fixed sampling interval,making it easy to understand and convenient to implement in theoretical studies.However,in the real process of sampling,the size of sampling interval is often different,leading to the problem of determining the distribution of the step lengths and the turning angles.To settle this issue,a mixed sampling frequency method which bases on the contour critical characteristics of the random walk path is proposed.The central work of this thesis is conducting the mixed sampling frequency analysis of animal movement data.We realize the sampling simulations of different random walk models by involving the ordinary method and the mixed sampling frequency method,respectively.We also apply the mixed sampling method to the path integration to verify its efficiency.The main contents are listed as follows.(1)Theories of random walks are reviewed in this thesis.These theories include simple random walk,biased random walk,correlated random walk and Levy walk.Non-Markovian random walks including the elephant’ walk,the Alzheimer’ walk and the monkey’ walk are also described herein.(2)Random walk models are applied to movement ecology.We first apply theories of Bayesian statistics to analyzing random walks.Considering the problem of sampling rate,we simulate ordinary sampling method in the application of random walk model.Moreover,we point out the question that the different size of sampling interval can lead to the problem of determining the variable distribution.(3)We study the influence of sampling methods on different random walk models.The ordinary sampling methods for different random walk models are tested with different sampling periods.Then,we propose a new mixed sampling frequency method on the basis of basic characteristic of contour and apply to random walk models.(4)Based on the path integration in the process of random movement,we provide two kinds of path curvature evaluation methods.Through the application of the sampling methods for the study of path integration,we examine the difference between ordinary sampling method and the new mixed sampling frequency method.Results show that the new mixed sampling frequency method provides more accurate information for the path integration.