Geographical probe calculation

The Geodetector model is a statistical methodology that can be utilized to identify geographical anisotropy, assess the significance and contribution of individual drivers, measure the strength of interactions between components, and facilitate the detection of potential risks. The concept is grounded in the principles of spatial statistics and spatial autocorrelation.

Geodetector is used to detect the spatial heterogeneity of the dependent variable Y (i.e., NDVI values) and the magnitude of the influence of the independent variable X (i.e., selected natural and anthropogenic factors) on the spatial heterogeneity of Y, measured by the q value²⁵.

Here, h is the number of categorical terms of the independent variable X, h = 1, …, L, with a value range of [0, 1], and the larger the value, the stronger the influence of X on the spatial differentiation of Y; Nh and N are the number of cells in category h and in the whole region, respectively; σh2 and σ2 are the variance of category h and of the whole region Y, respectively; SSW and SST are the sum of the variance of category L and the total variance of the whole region, respectively.

The corresponding X and Y attribute values were then extracted from the raster data and subsequently utilized in the Geodetector model for computation and analysis. The natural and anthropogenic factors are shown in Table ​Table2,2, and the driving factors, such as precipitation, air temperature, surface temperature, potential vapor dispersion, soil moisture, elevation, GDP, and population density, were classified into seven categories by the natural breakpoint method. Soil types were classified into 12 categories, vegetation types into eight categories, and land use types into six categories according to the Chinese national criteria for broad categories⁵⁰.

Classification of natural and anthropogenic factors based on geographic detectors.

This method can identify the interaction between different independent variables X and calculate whether the influence of the two factors on Y is related or independent when they act together. The value can be represented by q(X1 ∩ X2), as shown in Table ​Table33.

Model driving force size criterion of interval and interaction.