ALM-MOSART has been run over the 1972-2004 period with the IMCLM45 compset, i.e., CLM4.5+MOSART+I2000. The validation is done mainly through comparison with the observed streamflow data from the Global Runoff Data Center (GRDC). 9 out of world largest 50 rivers are arbitrarily selected, hence the corresponding GRDC stations nearby the river mouths, as shown in the figure. The red line is for the simulated streamflow by ALM-MOSART; blue dashed line for the simulated streamflow by CLM4-MOSART; and black dashed line is for the GRDC observed streamflow. The CLM4-MOSART is the version used in Li et al. [2015]. The difference between ALM-MOSART and CLM4-MOSART is mainly due to 1) the difference of soil hydrology between ALM and CLM4.0; 2) the different methods of treating the Qgwl runoff term. In ALM-MOSART, all Qgwl values, no matter positive or negative, are directly sent to the ocean without channel routing. In CLM4-MOSART, positive Qgwl values are sent to the hillslope routing, then channel routing; while the negative Qgwl values are treated ad hoc so that the total runoff is not negative before sending to the hillslope and channel routing.

Qmonthly_CLM4_ALM_MOSART_Obs.pdf

Major conclusions:

MOSART reproduces the streamflow reasonably well at various locations coupled with ALM;

Overall, it seems that ALM-MOSART performs better than CLM4-MOSART.