GEWEX News, WCRP, International GEWEX Project Office, 6, 4-5, 1996.
This manuscript may be different from the original article in the publication.
River discharge returns water to the ocean which may have been carried in vapour phase by atmospheric winds deeply into continents, and the global water cycle will never close without this process. However, it has not been well focused on in the global climatological studies. Recently, it has been expected that river discharge can be used for validating general circulation model (GCM) simulations. River routing model is required in order to compare the runoff from GCM with observation because there is a lag and accumulating process between runoff generated at GCM grid by land surface parameterization (LSP) and runoff observed at gauging station.
The runoff routing model of Miller et al. (1994) was used by off-line in this report Kanae et al. (1995), for the runoff generated by the modified bucket model of Kondo (1993) embedded in the atmospheric GCM of the Center for Climate System Research (CCSR), University of Tokyo and the National Institute for Environmental Studies (NIES). In the case of this bucket model, some portion of precipitation becomes runoff even before the bucket is ``full''. A numerical experiment was performed for the climatological mean SST with its seasonal change, and the seasonal water cycle in major river basins was investigated in a climatological sense (Oki et al., 1995). River channel network was built both from digital elevation map and from an atlas manually in approximately 5.6o × 5.6o grids, corresponding T21 of the GCM.
Effect of introducing routing model to the Amazon River basin.
[PostScript file is here.]
The velocity of the water in the river channel is a tuning parameter in the runoff routing model, and it was found that the velocity of 0.3 m/s gives good seasonal cycles of river discharge for the Amazon, Ob and Amur river basins (see Figure 1).
Figure 2: Water balance of the Amazon river basin by the AGCM. [PostScript file is here.]
The advantage of this research could be that total water storage in each river basin was compared with other independent estimates by atmospheric water balance (AWB) method (Oki et al., 1995a, b). The total water storage estimated by the AWB method and by the AGCM compared reasonably well (solid lines with open stars in Figs. 3 and 4). Total water storage in river basins estimated by AWB method includes the changes of surface soil moisture, snow accumulation, ground water, water in lakes and river channels, and the estimation is only relative value. Minimum value was set to zero in Fig. 4. It should be really important to examine not only the flux (discharge) from the model but also the state variable (storage) in the model with observation.
Figure 3: Hydrologic cycle of the Amazon river basin estimated by atmospheric water balance. [PostScript file is here.]
Actually, 0.3 m s-1 seems slower than the flow speed in a channel of large rivers. Reminding the idea of bucket model in GCM (Manabe 1969), generated runoff in the bucket model is the water which have no relation any more with evaporation at the grid either by surface runoff or by percolation into deep soil layers. Therefore the effective velocity should be regarded as an integrated mean velocity of whole history that rainwater travels from surface soil layer to the river mouth through various paths. In this sense, the routing model in this study should include the ground water process and not a few fraction of its storage could be occupied by ground water storage.
The river routing model is currently included in a coupled atmosphere-ocean GCM of CCSR/NIES, and it will be investigated how the inclusion of river routing effects the thermohaline circulation of the ocean. Irrigation effect that the runoff water from upstream moistens the soil moisture at downstream grid should be expressed for the fully coupled atmosphere-ocean-river GCM in the near future.
GCM generally uses more robust forcings and boundary conditions compared to traditional rainfall-runoff models in hydrology, and simulating river discharge by GCM can a realization of the hydro-graph estimation of `ungauged river basin', which should have been the one of the ultimate dreams of hydrologists. Therefore, the river routing with LSP in GCMs can be one of the most challenging issues for hydrologists.
Figure 4: South America river basins in a 1 degree by 1 degree mesh. [PostScript file is here.]
For studies on the global water cycle, the river channel newtork ( TRIP ) for major basins of the globe (nearly 300) was recently constructed on a 1 degree by 1 degree global mesh. Figure 4 is an example for South America. This global product can be used in studies such as the International Satellite Land-Suraface Climatology Project ( ISLSCP) Global Soil Wetness Project.
(Last updated at Thursday, 05-Jun-1997 07:59:20 JST, by Taikan OKI)