Regional differences in the dominant controls and regulatory mechanisms of surface water and heat fluxes across the climate zones of the Tibetan Plateau

The Tibetan Plateau (TP) is characterized by complex and heterogeneous surface conditions across its multiple climate zones, leading to significant spatial variability in the dominant controls on surface water and energy fluxes. This complexity poses a significant challenge to mechanistic analyses of surface water and energy fluxes and to the evaluation of flux products from remote sensing and reanalysis. In this study, we synthesize multi-year eddy-covariance observations from 19 land–atmosphere interaction stations covering humid, semi-humid, semi-arid and arid regions of the TP and use convergent cross mapping (CCM) and boosted regression trees (BRT) to: (1) systematically identify the principal drivers of latent heat flux (LE) and sensible heat flux (H) and their regional variability; and (2) assess the performance of several commonly used remote-sensing and reanalysis flux products over the TP. We find pronounced climate-zone differences in the dominant controls on LE. In humid and semi-humid zones, net radiation (Rn) is the primary driver; within the semi-humid zone, increasing volumetric water content of the shallow soil (Shallow VWC) shifts LE progressively from a water-limited to energy-limited regimes. In semi-arid and arid zones, LE is jointly regulated by Shallow VWC and vapor pressure deficit (VPD): higher Shallow VWC consistently enhances LE, whereas low VPD favors LE and high VPD strongly suppresses it. These regional differences can be attributed to the differential responses of surface conductance (Gs) to Shallow VWC and VPD. In arid and semi-arid regions, Shallow VWC plays a key regulatory role: increases in Shallow VWC markedly enhance the sensitivity of Gs to VPD and increase the reference conductance (Gs_ref); meanwhile, Gs in these regions is more sensitive to VPD; therefore, under high VPD conditions, Gs declines rapidly, thereby strongly suppressing LE. By contrast, no comparable regulatory effect of Shallow VWC is observed in humid regions, and the sensitivity of Gs to VPD is overall lower than in arid and semi-arid regions. By contrast, controls on H are consistent across climate zones and are dominated by the land–air temperature gradient (Ts–Ta). Evaluation of remote-sensing and reanalysis products indicates critical weaknesses with pronounced regional specificity in their performance. Because the models fail to accurately characterize the roles of Shallow VWC and VPD, LE estimates are least accurate in arid regions; by contrast, biases in H are largest in humid regions, where most products still inadequately represent its modulation by Ts–Ta.