next up previous
Next: Discussions Up: Influence of precipitation variability Previous: Model validation and analyses


Results and conclusions

Table 2 summarizes the effects of precipitation and anthropogenic subgrid variability on mean annual water balance components in the Yellow river basin. Without considering precipitation heterogeneity, the runoff contribution was underestimated. The simulated runoff contributions are less than observed ones in mountainous sub-division up TNH. The runoff contribution was well simulated after taking into account precipitation heterogeneity. The runoff contributions always are positive values in the cases without irrigation scheme. This conflicts with the observed negative runoff contributions in arid regions such as sub-division LZ-QTX and QTX-TDG. The negative runoff contribution cannot be simulated by only considering the natural heterogeneity. This constructive model shortcoming can be eliminated by taking into account anthropogenic heterogeneity. With irrigation scheme, the simulated annual runoff contributions in the sub-divisions LZ-QTX and QTX-TDG are -63 mm and -23 mm, corresponding to the observed -61 mm and -56 mm, respectively. The negative runoff contribution was modeled with the irrigation scheme. This also indicates irrigation water withdrawals have changed the pattern of hydrological cycle in the Yellow River Basin.


Table 2: Mean annual runoff (R), evaporation (E) for the various case (mm year-1)
Sub-div. Precipitation Obv. Case 1 Case 2 Case 3
    R E R E R E R E
Up TNH 483 173 310 140 344 174 309 178 305
TNH-LZ 416 88 327 79 336 108 307 83 332
LZ-QTX 316 -61 377 21 295 22 294 -63 379
QTX-TDG 240 -56 296 15 225 15 225 -23 264
TDG-LM 395 18 377 50 345 61 334 32 363
LM-SMX 523 42 481 80 443 114 409 64 459
SMX-HYK 601 86 515 97 504 144 457 46 555

Figure 6 shows the effects of precipitation heterogeneity on runoff simulation. The simulated total runoff of the case 1 in which precipitation is spatially uniformly put over large grid cell is much less than that of the case 2 in which the precipitation heterogeneity is considered. The annual total runoff is 81 mm for case 1 and 101 mm for case 2. The simulated total runoff differences are caused by the surface runoff differences. The annual surface runoff is 20 mm and 43 mm for case 1 and 2, respectively. This indicates the surface runoff simulation high depends on the precipitation heterogeneity.

Figure 7 shows the effects of precipitation heterogeneity and irrigation on annual stream flow along the Yellow River from upstream to downstream. Comparing the case no irrigation without precipitation heterogeneity and the case no irrigation with precipitation heterogeneity, the discharge are underestimated without precipitation heterogeneity. There are no big irrigation districts in the upstream Yellow River. The observed discharge at station TNH and LZ is used to validate the model. The discharge at the upstream station TNH and LZ is well simulated when the precipitation heterogeneity is taken into account. The observed discharge decreases between LZ and TDG station. Without irrigation scheme, the simulated discharge increases in the discharge decreasing zone though the increase is very small. The decreasing discharge along river main stem was simulated when the irrigation was taken into account. The results show that annual discharge at the HYK station decrease 41% because of irrigation. The anthropogenic influence is prominent after LZ station.

Figure 6: Effects of precipitation heterogeneity on runoff simulation.
15#15
Figure 7: Effects of natural and anthropogenic heterogeneity on annual stream flow along the Yellow River from upstream to downstream.
16#16

In figure 8, spatial distributions of water balance components associated with irrigation are shown at 10×10 spatial resolution. The figure 8a shows the irrigation water shortage (%) in each grid cell. The irrigation water shortage is calculated from the irrigation water withdrawals to the irrigation water requirements. The water shortage is small in the grid cells which are near to river main stem or are inside irrigation districts. The figure 8b gives the irrigation water withdrawals distribution (mm month-1) per unit grid cell area. The largest irrigation water withdrawals occur in the grid cells in irrigation districts and with high irrigated fraction. The figure 8c shows that the spatial differences between simulated evaporation with and without irrigation scheme. Evaporation increases in the irrigation districts and grid cells with high irrigated fraction. Within the simulation period, evaporation increases 25 mm year-1 because of irrigation in the Yellow River basin. Runoff spatial differences between simulated evaporation with and without irrigation scheme were shown in the figure 8d. Total runoff decreases because of irrigation, however, larger runoff occurs in the grid cells in irrigation districts because the flood irrigation becomes return flow and contributes to runoff. It is worth mentioning that all the values in figure 8 are mean value over grid cells, and that the values would have been much larger if reported values per unit irrigated area.

Figure 8: Spatial effects of irrigation on water balance components (a) water shortage (b) irrigation water withdrawal (c) evaporation change (d) runoff change.
17#17

Figure 9a shows the simulated surface soil wetness (soil moisture to saturated soil moisture) at the top 2 cm soil layer from the ground without irrigation scheme in the Yellow River basin. The surface soil wetness is lower in the upstream area of the river basin where the annual precipitation is small. The surface soil wetness is higher in the lower stream area where semi-humid area is. Figure 9a shows the simulated surface soil wetness change with irrigation scheme. The surface soil wetness increases because of irrigation water withdrawals, especially in the irrigation districts and high irrigated area. Over the Yellow River Basin and the study period, the surface soil wetness increases 6% because of irrigation. The surface soil wetness increases 11% in the grid cells where the irrigation districts are. The surface soil wetness increases 11% in the grid cells where the irrigated fraction is larger than 30%.

Figure 9: (a) Simulated surface soil wetness (top 2 cm) without irrigation in the Yellow River (b) simulated surface soil wetness change (%) with irrigation.
18#18

The basin averaged change in latent heat flux is 2.0 Wm-2, or 7.8% from 1983 to 2000 over the Yellow River Basin because of irrigation. The latent heat flux increases larger in peak irrigation season from June to August (JJA). The basin averaged change in latent heat flux is 3.3 Wm-2 in JJA. Figure 10 shows peak irrigation season changes in ground surface temperature, canopy temperature, latent heat flux and sensible heat flux for each grid cells in the Yellow River Basin. Ground surface temperature and canopy temperature decrease because of irrigation. Latent heat flux (or evapotranspiration) increases when the irrigation was taken into account. Sensible heat flux will decrease with irrigation. Again, the largest effects can be seen in cells where the irrigation districts are or where the percentage irrigated area is high, i.e. the middle and lower reaches of the Yellow River.

Table 3 shows the changes in energy components averaged over the river basin, the grid cells in the irrigation districts, and the grid cells where the irrigated fraction is larger than 30%. The decreases of ground surface temperature and canopy temperature are small over the basin with values of 0.1 K and 0.06 K, respectively. Averaged over irrigation districts, irrigation causes ground surface temperature and canopy temperature to decrease 0.32 K and 0.23 K, respectively. The ground surface temperature and canopy temperature decrease 0.4 K and 0.31 K respectively over the grid cells where the irrigated fraction is larger than 30%. The maximum change in ground surface temperature and canopy temperature can be found in a grid cell with irrigated fraction 65.5%, where the ground surface temperature and canopy temperature decrease 1.6 K and 1.2 K respectively. The latent heat flux increases over the grid cells in the irrigation districts and the grid cells where the irrigated fraction is larger than 30% are 11.2 Wm-2 and 15.5 Wm-2, or 3.5 and 4.8 times of the averaged increase over the basin. The maximum change in latent heat flux reaches 43.3 Wm-2, or 13.3 times of the mean value. The sensible heat flux decreases over the grid cells in the irrigation districts and the grid cells where the irrigated fraction is larger than 30% are 7.7 Wm-2 and 10.2 Wm-2, or 3.1 and 4.1 times of the averaged decrease over the basin. The maximum change in sensible heat flux reaches 37.8 Wm-2, or 15.1 times of the mean value. The results indicate that irrigation causes lower surface temperature, higher evapotranspiration, larger latent heat flux and smaller heat flux in the Yellow River basin. The lower surface temperatures and higher evapotranspiration resulting from human activities indicate that the near-surface atmosphere will be cooler and moister over irrigated areas than over non-irrigated areas.

Figure 10: Spatial effects of irrigation on energy balance components (a) changes in ground surface temperature dTg (b) changes in canopy temperature dTc (c) changes in latent heat fluxes raet (d) changes in sensible heat fluxes raht in peak irrigation season (JJA).
19#19


Table 3: Changes in energy components in peak irrigation season JJA
Item Avg. ID1 IF2 Max. Min.
Ground surface temperature dTg (K) -0.1 -0.32 -0.4 0.0 -1.6
Canopy temperature dTc (K) -0.06 -0.23 -0.31 0.0 -1.2
Latent heat fluxes raet (Wm-2) 3.3 11.2 15.5 43.3 0.0
Sensible heat fluxes raht (Wm-2) -2.5 -7.7 -10.2 0.0 -37.8
1 Averaged in the grid cells where the irrigation districts are.
2 Averaged in the grid cells where the irrigated fraction is larger than 30%.


next up previous
Next: Discussions Up: Influence of precipitation variability Previous: Model validation and analyses
TANG 2006-03-31