%0 Dataset
%T A dataset of monthly soil freezing depth distribution in the Pan Arctic from 2004 to 2023 hydrological years
%J National Cryosphere Desert Data Center
%I National Cryosphere Desert Data Center(www.ncdc.ac.cn)
%U http://www.ncdc.ac.cn/portal/metadata/86dcd91b-7090-48fe-9bfc-50b19b5f1c9a
%W NCDC
%R 10.12072/ncdc.arctic-change.db7144.2026
%A Chen Liyuan
%A Zhu Wenquan
%K Soil freezing depth;heat transfer;Stefan equation;machine learning;Pan Arctic region
%X Under the Arctic and winter amplification of warming, soil freeze depth (SFD) serves as a sensitive indicator of changes in pan-Arctic frozen soils. However, existing SFD mapping schemes fail to capture both spatiotemporal heterogeneity and physical constraints, and they mainly focus on annual maximum values. Consequently, high-precision datasets of monthly SFD in the pan-Arctic region are still lacking. We developed a monthly SFD mapping scheme (MSFDmap) by coupling the simplified Stefan equation with a random forest regression model, thereby integrating both physical constraints and spatiotemporal heterogeneity. The MSFDmap was implemented using 2123 site-month observations from 60 pan-Arctic sites over 20 years (water years 2004–2023), producing a monthly SFD dataset for the cold season (October–May) in the pan-Arctic region. In permafrost-underlain regions, data values represent the potential SFD—generally exceeding the active layer thickness—determined by assuming all soil heat loss is consumed for freezing. Validation against site-month observations shows high accuracy, with a root mean square error (RMSE) of 19.21 cm and a coefficient of determination (R2) of 0.91. Compared with data produced by existing schemes, RMSE decreases by 24–55 %, and R2 increases by 8–65 %.  Site-month SFD series derived from the dataset shows a highly consistent trend with observed series, with a Pearson correlation coefficient (r) of 0.99 and RMSE of 9.13 cm at the site-average level; 83% of sites exhibited strong consistency between the two trends (r ≥ 0.8). Spatially, the dataset presents the expected latitudinal and altitudinal gradients of monthly SFD, with the pattern consistency (r) relative to ERA5-Land soil-temperature-based estimates is 0.6. Overall, the dataset effectively captures the spatiotemporal dynamics of monthly SFD in the pan-Arctic region and outperforms data produced by existing schemes, providing a valuable data basis for frozen soil research.