Autocorrelation Measures For Spatio-temporal Series Characterized Of Long-range Correlation Series:A New Temporally Detrended Spatio-temporal Moran's Index

In research field of geostatistics,spatial autocorrelation is an importantly theoretical method to study association among geographical objects.In recent years,with the enormous development and wide application of 3 S technology and sensor network,a large number of spatio-temporal series have been accumulated,which are characterized by ubiquity of spatio-temporal autocorrelation and heterogeneity.One hopes that some important and hidden information will be mined and utilized by means of various models.However,in aim to build an effective model for this sort of series,the characteristics of spatio-temporal autocorrelation is taken into account.The existing spatio-temporal autocorrelation measurement models are mainly employed to measure autocorrelation of a spatio-temporal series in the presence of stationarity,which is not applicable of the spatio-temporal series characterized by long-range correlation.In this paper,a new temporally detrended spatio-temporal Moran's index will firstly be proposed to measure the autocorrelation of spatio-temporal series characterized by long-range correlation based on the spatio-temporal dependency.Afterwards,one example with artificial data is used to verify reasonability and reliability of model,and another example with real-word precipitation in northern Jiangsu province is taken to analyze spatio-temporal autocorrelation.In the end,significantly statistical test is taken for real-word precipitation.The major research topics are as follows:Firstly,literature review.We reviewed Moran's index theory,analyzed the spatio-temporal Moran's index models carefully and pointed out main disadvantages of them,which are unable to directly measure autocorrelation of spatio-temporal series characterized by long-range correlation and nonstationarity.Secondly,model presentation.Based on spatio-temporal dependency,we put forward a temporally detrended global spatio-temporal Moran's index to measure global autocorrelation of spatio-temporal series characterized by long-range correlation and nonstationarity,and then carried out an investigation on characteristics of model in depth.Thirdly,model analysis.On the basis of this proposed model,we presented a temporally detrended local spatio-temporal Moran's model and a extended Moran scatter plot,and also analyzed and explored characteristics of them thoroughly.Fourthly,example's verification.We utilized artificial series to analyze the characteristics and influencing factors of model,and then verified the model by daily precipitation data of 9 meteorological observation stations in Jiangsu province.In conclusion,based on spatio-temporal autocorrelation theory,this temporally detrended spatio-temporal Moran's model proposed is capable of measuring autocorrelation of spatio-temporal series characterized by long-range correlation and nonstationary series.and this model is adopted in other fields.