Assessment and simulation of global terrestrial latent heat flux by merging CMIP5 climate models and surface eddy covariance observations

The assessment and simulation of global terrestrial latent heat flux (LE) from climate models has flourished as a research focus in the recent decades. In this study, we evaluated LE simulations from 45 general circulation models (GCMs) in the Coupled Model Intercomparison Project Phase 5 (CMIP5) by a comparison with eddy covariance (EC) observations from 240 globally distributed sites from 2000 to 2009 and improved global terrestrial LE estimations for different land cover types by merging seven accurate CMIP5 models based on a Bayesian model averaging (BMA) method. The comparison results show that 45 GCMs illustrate substantial differences in the monthly LE, and CESM1-CAM5 has the best performance with the highest predictive skill and a Taylor skill score (S) ranging within 0.51-0.75 for different land cover types. The cross-validation results illustrate that the BMA method has improved the accuracy of the CMIP5 GCM's LE simulation by decreasing the averaged root-mean-square error (RMSE) by more than 3W/m2 when compared to the simple model averaging (SMA) method and individual GCMs. The inter-annual variation in the BMA-based global terrestrial LE shows an increasing trend with a linear slope of 0.018W/m2 yr-1 (p<0.05) during 1970-2005. This variation may be directly attributed to the inter-annual variations in air temperature (Ta), surface incident solar radiation (Rs) and precipitation (P). However, our study highlights a large difference from previous studies in a continuous increasing trend after 1998, which may be caused by differences in the variations in the input variables for different models on these time scales. Our study also emphasizes the necessity of merging GCMs and more observations to reduce the uncertainties of the individual model and provides corrected-modeling evidence for an accelerated global water cycle under climate change.