APPLICATION

In the present study, the DBHM has been applied to the Yellow River basin to investigate its applicability to a region with large variations in climate and land cover. The simulations were implemented by hourly time step from 1981 to 2000. Basic data representing topography, land cover, soil type were prepared in 10 km mesh grids derived from HYDRO1K (HYDRO1k Elevation Derivative Database; URL: http://lpdaac.usgs.gov/gtopo30/hydro/), GLCC and FAO, respectively.

The Yellow river is the second-longest river basin in China. It originates from the Tibetan plateau, drains through the northern semiarid region, crosses the loess plateau, passes through the eastern plain, and finally discharges into the Bohai Gulf. The whole basin has an area of 794,712 km². The Yellow River basin and river way in China and the river basin and river network derived from ten kilometers DEM are shown in Fig. 4.

Figure 4: The Yellow River basin and river way (Left); DEM derived river basin and river network (Right).

59#59

The meteorological data was obtained from the China Meteorological Administration (CMA) and GSSDD. Observation data at 287 stations was interpolated to ten kilometers grids with thin plate splines. The simulated hydrographs at 9 locations were compared with the river discharge observations for evaluation of the model in the years when the gauge records were available.

The discharge simulation is evaluated in terms of the error variances for the stream flow values. Let x_i and f_i (i = 1,..., n) denote time series of observations and simulated discharge. The averages are 60#60 = 61#61x_i/n and 62#62 = 61#61f_i/n. The mean square error between model-simulated f_i and the observed x_i is:

e² = 29#2963#6364#64f_i - x_i65#65 (23)

Set 66#66 = 67#67 and 68#68 = 69#69. The mean square error can be written in a fraction equation (Murphy, 1988):

70#70 + 71#71 + 72#72 = 1 (24)

e_m² = 73#7374#74 - 75#7576#76 (25)

e_a² = 77#7778#78 - 79#7980#80 (26)

e_p² = 278#7879#7981#811 - 82#8263#63(f_i - 74#74)(x_i - 75#75)83#83 (27)

The three terms in the equation (24) are related to overall bias error, amplitude error (through the ratio of the variances) and phase error (through the correlation). The root mean square error, rmse = 84#84, the relative root mean square error, rrmse = rmse/60#60, and the mean square skill score, MSSS = 1 - e²/85#85, are also calculated.

Table 1 shows the MSSS, rmse and rrmse values and the percentage of error related to the mean, amplitude, and phase of the simulated values for discharge in the Yellow River basin.

Table 1: Comparison of model-simulated and observed discharge

Stations	MSSS	rmse	rrmse	Percentage of error related to
				mean	amplitude	phase
Tangnaihai	-	-	-	-
Lanzhou	-	-	-	-
Toudaoguai	-	-	-	-
Sanmenxia	-	-	-	-
Huayuankou	-	-	-	-
Lijin	-	-	-	-

The results show that the model can perform better in upper stream of the Yellow river basin than in down stream. Fig. 5 gives the simulated and observed daily hydrographs comparison at an upper stream gauge Tangnaihai station and the river outlet gauge Lijin station. Even though the simulation shows less fluctuation than the observations, it can capture the major features of the hydrograph at the upper stream station. At river outlet gauge Lijin station, the simulated discharge is much larger than the observation. The model had success with the simulation of the flood peak which occurred in August 1996. But the simulated river discharge in the low water period cannot agree with the observed hydrograph. The discharge records show that the Yellow River stopped flowing for 438 days at Lijin station during the three years period from 1995 to 1997. The observed discharge was less 100 m³/s in 57% of days during the same time period. A major reason for the disagreement is artificial water regulation, such as reservoirs and water diversions. According to International Commission on Large Dams (ICOLD, 1984, 1988), there were 207 reservoirs in the Yellow River basin. Water withdrawals from Yellow River in 1993 were about 44.95 billion m³ (United Nations, 1997), which was about 13% of the total precipitation water in the basin of that year. The monthly simulated water balance is given in Fig. 6. It is found that evaporation is the major component of the water balance which agrees with the observations that show the annual evaporation is about 88% of precipitation. These results indicate that DBHM has good performance on hydrological simulation of large river basin. Further research on artificial water regulation would improve the simulation in down stream.

Figure 5: Comparison of daily hydrographs at Tangnaihai station (Left) and Lijin station (Right).

86#86

Figure 6: Monthly water balance of the Yellow River Basin.

87#87