Spatial autocorrelation clustering method for remote correlation mode

Abstract

The invention provides a spatial autocorrelation clustering method for a remote correlation mode. According to the method, the correlation degree of each space grid unit and an adjacent unit is considered; the method is based on local Moran index definition. A correlation coefficient original numerical value is adopted, and centralization processing is not carried out. Therefore, a local Moran index calculation formula is improved, a new spatial autocorrelation local index LISAAC is obtained, the detection of a significant positive or negative remote correlation aggregation range is realized,and meanwhile, the recognition of abnormal values (i.e., non-significant or negative grids appear in a significant positive value region and non-significant or positive grids appear in a significant negative value region) is realized. Experimental results show that spatial clustering of different types of remote correlation can be realized according to the standardization property of the remote correlation coefficient, and the results are convenient for transverse comparison of remote correlation degrees in different seasons.

Description

technical field [0001] The invention relates to the field of meteorology and climatology, and more specifically, to a space autocorrelation clustering method oriented to a teleconnection model. Background technique [0002] The relationship between large-scale meteorological factors and precipitation anomalies around the world has always been paid attention to, and the strength and mode of this teleconnection can be quantified by using the relevant meteorological factor indicators and the correlation coefficients of monthly or seasonal precipitation in various places. However, in actual calculation, the precipitation in adjacent areas is often not independent, but there is a strong correlation. In order to detect this kind of spatial autocorrelation, the local index of spatial autocorrelation (LISA) is often used for spatial autocorrelation analysis to detect whether there is information about the spatial pattern of variables. However, for the teleconnection model measured ...

AI Technical Summary

Problems solved by technology

However, for the teleconnection model measured by the correlation coefficient, the coefficient itself has standardized properties, and the commonly used LISA indicators such as the Moran index, according to the definition described in Chapter 4 of "Econometric Geography", the calculation process will be Variables are standardized, so the results obtained can only reflect the relative distribution of high and low values ​​at the same time and indicators, and it is difficult to make horizontal comparisons of calculation results at different times or indicators

Images