A Multiscale GR-EOF Approach Reveals Climate Mode Biases in Sea Level Rise Estimates

Satellite altimetry has revealed a persistent non-uniform global sea level rise over the past three decades, but trend estimates remain sensitive to internal climate variability on interannual to decadal timescales. Here, we develop an improved global-to-regional double EOF (empirical orthogonal function) (GR-EOF) framework to quantify the various climate modes induced sea level variability from 1993 to 2025. The leading global EOF modes capture the long-term trend, the annual cycle, the El Niño–Southern Oscillation (ENSO), and the Pacific Decadal Oscillation (PDO), together explaining ∼70.3% of total variance. During pronounced climate episodes, these modes collectively modulate the global-mean sea-level trend by up to 0.7 mm yr⁻¹ in 2011–2015. After removing these global climate-mode-related components, regional EOF analyses reveal the distinct roles of the Indian Ocean Dipole and the North Atlantic Oscillation. Correcting for these regional modes reduces the estimated sea-level trend from ∼8.1 to ∼5.7 mm yr⁻¹ for the Indian Ocean, and from ∼5.7 to ∼4.2 mm yr⁻¹ in the Atlantic. The Indian Ocean trend estimates show strong regional sensitivity to internal variability. On the other hand, in the Atlantic Pacific, driven influences appear less sensitive, likely reflecting stronger basin-scale ocean–atmosphere coupling and dynamics in the Indo–Pacific. The GR-EOF analytical framework provides a systematic approach for separating and quantifying forced signals from natural variability, thereby improving the robustness of global and regional sea-level trend estimates.