Comparisons of two models were made in the Arkansas and Red River basin for 1990 to 1999. The same atmospheric forcing-, vegetation type- and vegetation fraction data are used for both models. The simulated data of the CLM for 1992 and 1996 are missing because the CLM does not handle the Febuary of leap years properly.
Our routing model was run for 19 small basins (most of the basins are smaller than 1000 square mile). It seems that the VIC and CLM perform quite similarly in terms of simulated runoff. However, both models considerably overestimate the runoff for all the small basins. The overestimation might be due to the following factors: (1)the difference in upstream area between the actual and the simulated one; (2)the resolution of the origianl forcing data and/or soil, plant parameter data might not be high enough for the simulation in small basins; and (3)the difference in driver direction between the actual and the one we used.
Click the following links to see the streamflow plots:
Plot 1 & 2 Plot 3 & 4 Plot 5 & 6 Plot 7 & 8 Plot 9 & 10 Plot 11 & 12 Plot 13 & 14 Plot 15 & 16 Plot 17 & 18 Plot 19
The routing model was run for 5 larger basins (contributing areas are between 20,000 and 62,000 square miles). Similar to the small basins, the VIC and CLM behave similarly. However, both models perform much better than in small basins although the runoff is still oversimulated (I did not look at how the human activties, such as industrial and agricultural water comsumptions and dams, affect the streamflow).
Click the following links to see the streamflow plots:
Calvin OK Gainesville TX Ralston OK Terral OK Tulsa OK
Since VIC 4.1.0 (the most recent unofficial version) contains bugs (according to Ted), in the following comparisons VIC 4.0.5 (the most recent official version) is used on water balance mode, with model time step being 24 hours and snow time step 3 hours.
The comparisons are made for both large basins (contributing area greater than 20,000 aquare miles) and small basins (contributing area smaller than 1,000 square miles). The simulated streamflow is compared with the observed one (not naturalized). The results show that VIC well reproduces the runoff/streamflow and performs significantly better than CLM for both large and small basins over the simulation period. Click the following links to see the streamflow plots.
Little Rock AR Calvin OK Gainesville TX Ralston OK Terral OK Tulsa OK
Amarillo TX Archer City TX Cimarron 1 NM Cimarron 2 NM Dawson NM Eagle Nest NM Electra TX Golondrinas 1 NM Golondrinas 2 NM Henrietta TX Kelton TX La Cueva NM Logan NM Quanah TX Truscott TX Wayside TX
Little Rock AR Ralston OK Tulsa OK
Archer City TX Cimarron 1 NM Golondrinas 2 NM La Cueva NM Truscott TX
This part of the work was conducted in collaboration with Dr. Christine Delire , University of Montpellier II, France.
| |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |