4.2 Evapotranspiration, soil water storage, and runoff
The seasonal cycle of the regional average latent heat flux is shown in Fig. 4. Generally, the CAS-LSM simulated values of the latent heat flux followed the observations, except with an overall underestimation in the nine representative regions. The general underestimation of the seasonal cycle was apparent, except for Amazon, which generally followed that of the observations (although with an underestimated climatology). Overall, the peaks of the seasonal cycle were smaller in the other eight places relative to those seen in the observations, among which India displayed the worst result.
The differences in the mean soil water content between MAM and SON between the model simulation and GLDAS2 (Rodell et al. 2009) are shown in Fig. 5. While the simulations agreed well with the observations in general (Figs. 5(a–b)), the value magnitudes of the simulations were smaller than that of GLDAS2 in regions of the Amazon, southern and central Africa, North America, southern and eastern Asia, and Australia. This indicated weaker seasonal variations in the simulated soil water content. The hydrology and snow parameterization schemes in the LSM may be largely responsible for these biases.
River discharge to oceans in the CAS-LSM simulations was calculated via the RTM (Branstetter and Famiglietti 1999), which transported grid-cell runoff to oceans via pathways that approximate the paths of the real global river network. Overall, the total simulated river discharge into oceans was scattered around the observations (Fig. 6), with some underestimations at some stations, especially in those that have lower annual discharge. As the total discharge into the world’s oceans was directly related to land precipitation minus evapotranspiration, these underestimations may be due to precipitation bias in the model. In the simulations, the most poorly simulated large rivers were the Congo and Orinoco Rivers, where far too much discharge was simulated for the Congo River (more than 100% discharge than observed) and far too little discharge from the Orinoco River (less than 50% discharge than observed).