Precipitation governs structural uncertainty and physical attribution of drought trends

<title>Abstract</title> <p> Global drought assessments frequently attribute drying trends to warming-induced evaporative demand represented by potential evapotranspiration (PET), yet the dominant physical mechanism and structural data uncertainty in drought trends remain poorly understood. Using a 240-member Standardized Precipitation Evapotranspiration Index (SPEI) super-ensemble over quasi-global land (50°S–50°N), we find substantial structural uncertainty of estimated drought trends, represented by a large inter-member standard deviation (± 0.0052 yr ⁻ ¹) of quasi-global SPEI trend relative to the ensemble mean (− 0.0029 yr ⁻ ¹). Our uncertainty decomposition analysis shows that precipitation-product discrepancies account for 71.5% of global-mean and 78.1% of grid-level SPEI trend variance during 1981–2024, far exceeding PET-related uncertainty. Three independent physical attribution frameworks consistently show that precipitation changes explain 66.0–74.6% of historical SPEI trends, with PET as a secondary contributor. Ensemble subsets combining gauge- or satellite-derived precipitation with energy-based PET formulations yield non-significant quasi-global drying (11 members) or even wetting trends (73 members), reflecting the sensitivity of apparent trends to precipitation forcing choices. Future CMIP6 projections confirm precipitation’s persistent dominance in drought trends, with PET contributions exhibiting regionally heterogeneous strengthening or weakening. These findings challenge warming-centric interpretations of drought trends and call for explicitly constraining precipitation uncertainty in drought-risk assessments. </p>