Empirical constraints on the fraction of surface latent heat flux reaching the top of atmosphere as net radiative cooling

Ali Bin Shahid 

The transfer fraction η, defined as the ratio of net cloud radiative effect at the top of atmosphere to surface latent heat flux (η = |CRE_net| / LE), is a fundamental coupling parameter linking surface hydrology to TOA radiation. Despite its relevance to cloud feedback and biophysical forcing, η has not been directly constrained by co-located observations at the spatial scales relevant to climate forcing. This study co-locates 341 FLUXNET eddy covariance surface flux sites (FluxDataKit-v3 pooled with JapanFlux2024) with CERES EBAF Ed4.2 TOA radiation retrievals, and extends the analysis to basin scale for the Amazon, Congo, and SE Asian tropical forest regions. A three-stream decomposition of the surface energy budget is presented, together with a geometric upper bound on η in deep convective regimes. At site level, the global median η is 31.6% [95% CI: 28.3 to 35.8%], with a tropical evergreen broadleaf forest median of 14.7% across twelve sites on five continents (range 5.6 to 31.0% excluding the subtropical-monsoon outlier CN-Din). At basin scale, the Amazon transfer fraction is 20.8% (CRE_net = -19.8 W/m^2, LE = 95.1 W/m^2), with a recycling amplification factor of 1.62x relative to the site-level value; Congo (1.63x) and SE Asia (1.57x) show comparable amplification. The cloud longwave radiative effect scales with convective available potential energy (CAPE) at R^2 = 0.74. The constrained η supports two applications: a diagnostic for cloud feedback in atmospheric models, and an observational TOA counterpart to surface biophysical forcing estimates from land-cover-change studies. These results provide the first dedicated, scale-stratified observational constraint on η at site and basin scales.