Estimating carbon content using two methods for spatial averaging: a case study within complex, native Siberian forests

This paper studies spatial averaging of point-based measurements of carbon content within the landscapes of native Siberian forests. Two methods of spatial averaging were used: (1) pixel classification with learning from satellite images, and (2) GIS-based interpolation from point measurements. For the same comparison area of 734 hectares, satellite imagery estimated that there would be 112.1 Mg of carbon per hectare, whereas GIS-based interpolation suggested 126.77 Mg of carbon per hectare - 11–12% higher. The first method is more direct and it takes into account all of the pixel variety, but it works only for parameters that are seen from a satellite. The GIS-based approach is more mechanical (interpolation) and it is certainly sensitive to the number and arrangement of measurement points, but it is applicable both for forest canopies and what is below, for example, soil parameters. We conclude, therefore, that although the GIS-approach might overestimate the carbon content of forests, it still gives reasonably accurate values.