A New Automatic Statistical Microcharcoal Analysis Method Based on Image Processing, Demonstrated in the Weiyuan Section, Northwest China

Abstract

Microcharcoal is a proxy of biomass burning and widely used in paleoenvironment research to reconstruct the fire history, which is influenced by the climate and land cover changes of the past. At present, microcharcoal characteristics (amount, size, shape) are commonly quantified by visual inspection, which is a precise but time-consuming approach. A few computer-assisted methods have been developed, but with an insufficient degree of automation. This paper proposes a new methodology for microcharcoal statistical analysis based on digital image processing by ImageJ software, which improves statistical efficiency by 80–90%, and validation by manual statistical comparison. The method is then applied to reconstruct the fire-related environmental change in the Weiyuan loess section since about 40 thousand years before present (ka BP), northwest China with a semi-arid climate, found that the microcharcoal concentration is low in cold and dry climate and high in warm and humid climate. The two main contributions of this study are: 1) proposal of a new, reliable and high efficient automatic statistical method for microcharcoal analysis; and 2) using the new method in a semi-arid section, revealing the paleofire evolution patterns in the semi-arid region was mainly driven by the biomass rather than the aridity degree found in humid regions.