Results

Fig. 5 shows modeled and observed discharge to downstream from 1999 to 2002. Generally, the stream flow is fairly well reproduced, except the peak and low flow in 1999. It is surprising because the investigation of mountain torrent and spring ravines was performed in that year. One possible reason is that the mountain torrent and spring did not rush into the main stem as expected but were absorbed by the plain. The widespread existence in the plain of sand loam and sandy soil with high water permeability supports that hypothesis. The purpose of this study is not mainly to reproduce the hydrograph, but a good reproduction indicates that the study is providing a reasonable water budget.

Table 1 shows the water budget for the Tuoshigan-Kumalake River main stem from 1999 to 2002. The flow to sub-areas Q_sb (1,613×10⁶ m³/year) was obtained from observations, including water diversion (54% of Q_sb) to the two sub-areas in the study area and water transfer (46% of Q_sb) to outside the study area. The shallow water table and numerous artesian springs contributed to large water regeneration, and besides, the planting of paddy field caused large drainage. The return flow (Q_rf) and tributary inflow (T_r) were 440×10⁶ m³/year, near half of the water diversion into the study area (870×10⁶ m³/year). Due to the low precipitation and high evaporation rate, it is expected that the precipitation affected the water budget little and the evaporation from open water surface was substantial. Most notable is the storage change in aquifer ( S_a), which is larger than the tributary inflow. Detailed investigation of the monthly variation of aquifer storage change shows that large volumes of river water are stored in the alluvial plain in the flood season (July to September) and the river-groundwater interaction is relatively small during the low flow season. This indicates that the river loss to the aquifer is consumed in the alluvial plain or regenerates as surface water.

The mean annual water budget for the ditch systems in the two sub-areas is shown in Table 2. Artesian spring flow from outside or inside study area (D_is) is as large as the water diversion from the riverway (D_ir), indicating that surface water-groundwater interaction is intensive in the study area. Nearly 60% of total water diversion is ditch losses (D_e and D_g), and the remaining 40% reaches the farmland. This is possible because of the high water permeability of the underlying sandy soil and earth ditches without liner. Large ditch loss recharges the groundwater because water table is shallow and this contributes to lifting up the water level in the aquifer. The utilization of ditch liner to prevent water loss will increase the water transfer efficiency, but it may also lower the water table. Groundwater discharge of artesian spring, which originates from groundwater recharge, such as ditch penetration and river seepage, is freely used by the local population. The methods to prevent ditch water loss will inevitably disturb the artesian springs. It is clear that much additional work will be required before a complete understanding of the magnitude of the disturbance can be reached.

The mean annual water budget for the irrigation area in 1999 to 2002 in the two sub-areas is shown in Table 3. Ditch diversion and groundwater recharge compose the major water recharge in the irrigation area. Evaporation from the irrigation area, the water consumption, amounts to about half of the total water supply. It is notable that the groundwater exchange with the non-irrigation area (X) is as large as the water consumption. The flood irrigation system is believed to contribute to the large lateral groundwater flow. Fourteen percent of the water supply is drained out of irrigation area, and the proportion is 23% in Wensu sub-area, where paddy field is cultivated.

Table 4 shows the mean annual water budget for the non-irrigation area in the two sub-areas from 1999 to 2002. It is notable that the precipitation over the non-irrigation area is much less than the total water consumption, i.e. the sum of evaporation from native plants land and waste land. Actually, precipitation stands for only 14% of the total water supply. The remaining 86% of the water supply comes from groundwater recharge (G_ni) and groundwater from the irrigation area (X), meaning that the natural ecosystem of the arid area relies on the managed agricultural ecosystem for water supply. The storage change is small in Wushi but relatively large and negative in Wensu sub-area, indicating that there might be a water table drawdown in this sub-area. Unfortunately for Wensu sub-area, data from only one groundwater gauge well from the years 2000 to 2002 was available to validate drawdown. These show a mean 18 cm drop during the three years.

The results indicate that both irrigation area and non-irrigation area are supported by the water from river way. The water to irrigation area is from river way through a canal system directly. However, the water to non-irrigation area comes from canal loss and groundwater from irrigation area. This manifests that water supply of natural plants relies on the water from agricultural ecosystem in the hyper-arid environment. Tight water connection between irrigation area and non-irrigation area suggests that natural ecosystem needs to be integrated in a comprehensive agricultural management in arid environment.