Spatial analyses in environmental studies: III. Use of local statistics to reveal spatial variation and spatially varying relationships

It is always a challenge to reveal, quantify or visualize spatial variation properly. This presentation demonstrates the use of local statistics to quantify spatial variation and model spatially varying relationships. Based on... [ view full abstract ]

It is always a challenge to reveal, quantify or visualize spatial variation properly. This presentation demonstrates the use of local statistics to quantify spatial variation and model spatially varying relationships. Based on a total of 6138 topsoil samples in Northern Ireland, neighbourhood statistics of local mean, local standard deviation and local coefficient of variation were calculated. The results showed that high local standard deviation values were found to be associated with high local mean values thereby limiting the usefulness of local standard deviation as an indicator of spatial variation. The strongest spatial variations were observed in areas with strong changes of rock type, e.g., on the western edge of the basalt area along the boundary of the basalt–sandstone areas and the schist area. Within each rock type, spatial variations were relatively weak and this was most clearly demonstrated in the basalt area. Another “local” statistical method of geographically weighted regression (GWR) was applied for the spatial modelling of SOC in Ireland. Based on a total of 1310 samples of SOC data, environmental factors of rainfall, land cover and soil type were investigated and included as the independent variables to establish the GWR model. SOC showed good spatially varying relationships with elevation. The SOC map produced using the GWR model showed clear spatial patterns influenced by environmental factors and the smoothing effect of spatial interpolation was reduced. These results demonstrate the potential uses of local statistics in revealing spatial variation and spatially varying relationships in environmental studies.