Response of microscopical hyperspectral data to past climatic variable

Tree growth is ordinarily affected by changes in environmental conditions, and the annual sequence of favorable and unfavorable climates is regularly recorded in wide and narrow tree rings. Therefore, data extracted from tree rings can be adopted to reconstruct the past climate. In dendroclimatology, the commonly used proxy indices to obtain paleoclimatic parameters include tree-ring width, blue intensity, density, and isotopic series. Notwithstanding, reconstructing the past local climatic parameters with high accuracy is a challenge. Here, we demonstrate how data from microscopic-hyperspectral scanning of the dendrochronological samples, together with the tree-ring width data, facilitates the precise estimation of one paleoclimatic parameter, the mean deviation of monthly air temperature. In a proof-of-concept analysis experiment, samples were collected from two sites located in the middle part of China, measured with a hyperspectral scanning system, model achieved an averaged root mean square error (RMSE) of 0.524℃ on the mean deviation of monthly air temperature from March to July, which demonstrates significant improvements in accuracy compared with the classical single-indexed tree-ring width method, especially when the number of tree-ring samples is limited. Further, our results suggest that climatic data stored in the tree-ring spectrum could be divulged by analyzing their surface hyperspectral reflectance. We hope our work will be a genesis for more intricate models in dendroclimatology that use microscopic-hyperspectral scanning as a standard tool, which promises enhanced accuracy for past climate reconstruction. Besides tree rings, the capacity of the microscopic-hyperspectral scanning technique can be exploited on other samples collected in paleoclimatology, including core samples from rocks, sediment, ice, or surface samples on shells, boreholes, and corals, to derive valuable information from their surface spectral patterns.