Kriging space interpolation method based on multi-scale wavelet support vector machine optimization

Abstract

The invention discloses a Kriging space interpolation method based on multi-scale wavelet support vector machine optimization which comprises the following steps: using a discrete variation function formula to calculate the experimental variation function values of all sampling point pairs; using a multi-scale wavelet least squares support vector machine to fit the experimental variation function values to obtain a theoretical variation function model; establishing Kriging space interpolation equations; according to the theoretical variation function model, calculating the Kriging weight coefficient Lambda i; according to the Kriging weight coefficient Lambda i, calculating the regional variation amount estimation value of a to-be-estimated position point; based on the regional variation amount estimation value and the actual value of the to-be-estimated position point, using the mean absolute error (MAE) and the root mean square error (RMSE) to conduct interpolation accuracy evaluation. According to the invention, the optimized Krigine space interpolation method achieves a higher accuracy; therefore, it is possible to evaluate the attribute information of a to-be-estimated position point more accurately and to increase the capability of geographical data to interpret geographical space phenomenon.

Description

technical field [0001] The invention relates to the field of geographic space interpolation, in particular to a Kriging space interpolation method based on multi-scale wavelet support vector machine optimization. Background technique [0002] Geospatial interpolation is to estimate the attribute information of geospatial location points to be estimated based on the attribute information of known sample points. The essence of estimating the value of other arbitrary points or arbitrary partitions is to fit and generate a functional relationship that is as close as possible to the geographic spatial distribution characteristics through modeling. There are a wide variety of geographic spatial interpolation methods, commonly used include Thiessen polygon method, inverse distance weighting method, moving fitting method, linear interpolation method, spline function method, trend analysis method, kriging interpolation method, etc. Among them, the kriging interpolation method is als...

AI Technical Summary

Problems solved by technology

The traditional method can only use the existing theoretical variogram model to fit limited geographical space samples, but the existing theoretical model has a fixed shape and cannot reflect the spatial variation trend of the actual sample

Secondly, the spatial change trend often has the characteristics of multi-scale, and the choice of scale and the treatment of scale effects also need to be considered. The experimental variogram fitted only by traditional methods will ignore the scale effect of spatial change

Images