Abstract
Continental-scale isoscapes of tree-ring oxygen isotopes (δ(18)O(TR)) are crucial for understanding atmospheric circulation dynamics, interpreting climatic significance, and tracing wood provenance. However, continental-scale δ(18)O(TR) isoscapes remain underdeveloped. We compiled 313 multi-year averaged δ(18)O(TR) records across Asia and generated isoscapes using two machine learning approaches: XGBoost and Random Forest. Results reveal a 'sandwich' pattern: depleted values at high (>50°N) and low (<30°N) latitudes, enriched values at mid-latitudes (30°N-50°N). This pattern closely resembles the distribution of precipitation δ(18)O (δ(18)O(P)). Correlation and commonality analyses confirm δ(18)O(P) as the primary driver of δ(18)O(TR) isoscape patterns across Asia. Continental-scale δ(18)O(TR)-elevation relationships are generally insignificant, except in Indian Summer Monsoon regions showing significant negative correlations (r = -0.69, p < 0.05). These findings suggest that δ(18)O-based paleoaltimetry reconstructions work best in regions with dominant moisture sources such as Indian Summer Monsoon regions. This study provides Asia's first continental-scale δ(18)O(TR) isoscapes, establishing a foundation for atmospheric circulation and dendroprovenancing research.