Reconciling Discrepancies Between Observed and Model Predicted Shortwave Cloud Radiative Trends

Abstract Satellite observations over the past two decades indicate increasing absorbed shortwave (SW) radiation at the top-of-atmosphere. Much of this increase results from cloud-driven changes (dSW cld ), yet climate models reproduce only about ~60% of the observed global-mean trend. To investigate this discrepancy, we apply an independent cloud radiative kernel (CRK) framework to estimate dSW cld . CRK-based estimates reproduce the spatial pattern of dSW cld but explain only about one-third of its global-mean trend over 2003–2022, leaving a near-uniform positive residual. In contrast, models show close agreement of dSW cld with CRK-based trends, with no evidence of a uniform residual. We further identify a small positive drift in observed SWCRK that projects onto multi-decadal cloud radiative effect trends and subtle negative reflectance drifts inferred from deep convective cloud albedo diagnostics. These findings suggest the true cloud-driven shortwave trend may lie near the lower bound of observational uncertainty, reducing the apparent model–observation discrepancy.