Assessment of reanalysis flux products based on eddy covariance observations over the Tibetan Plateau

Surface sensible heat flux (SH) and latent heat flux (LH) are important variables linking the energy and hydrological cycles over the Tibetan Plateau (TP). Through comparison with eddy covariance observations of turbulent fluxes at two sites from 2013 to 2015 and at eight sites in August 2014 over the TP, five reanalysis flux products were evaluated for monthly mean biases and daily variations after bilinear interpolation. The results showed the JRA-55 and ERA-Interim reanalyses had optimal performances regarding SH, while NCEP2 had reasonable performance regarding LH. The SH in the reanalysis data generally reached maximum negative biases in winter and minimum positive biases in summer, whereas the seasonal bias of LH was found opposite to that of SH. The diurnal variation of reanalysis fluxes was found consistent with observed data at most sites, and the diurnal SH and LH maxima were all overestimated except SH in the JRA-55 at Nagri. The SH bias and the bias of the surface wind speed in NCEP2 were highly significantly correlated, suggesting the bias in the surface wind dominates the SH bias over the TP in NCEP2. Large spreads of averaged biases and standard deviation errors were found among the five reanalysis bulk variables. In particular, the maximum-minimum ranges of the biases of surface wind speed and ground-air–specific humidity contrast were > 5 m s−1 and > 15 g kg−1, respectively.