Small reservoirs catch and store water for long periods and they decrease streamflow and increase ground-water recharge. A field monitoring program provided the measured water depth for four years in several reservoirs in the Republican River Basin where there are concerns about their aggregate effects in the basin.
The daily water budget operation for one reservoir was developed. Daily seepage rates were estimated by using precipitation, inflow and evaporation which was assumed equal to grass reference evapotranspiration (ET0), that average 120 to 150 cm/yr, along with the measured stage-storage and stage-surface area relationships. Two computer simulation modules, written in FORTRAN 95, were developed to estimate 1) overflow and gross seepage and 2) potential for ground-water recharge underneath the reservoir. Required daily input data are precipitation, ET0, and inflow from the watershed area. Required reservoir site characteristics include stage-storage and stage-surface area relationships, a standard seepage rate (S0) at 14 different levels in the reservoir, soil-water and plant-growth characteristics and a monthly crop-residue factor.
The gross seepage module calculates water depth that determines daily overflow, the water-surface area for evaporation and the head of water on the 14 levels to cause seepage losses. If a level is not inundated, seepage is zero. If a level is inundated less than 0.3-m, S0 is used. When the water head (hL) on a level exceeds 0.3 m, the seepage rate (SL) is increased by, SL = S0 * (hL/0.3)0.25. This relationship was chosen after testing several exponent values between 0 and 1.
The modules were calibrated on one reservoir and verified on two others in northwestern Kansas. Results showed runoff from the watersheds averaged about 1.2 to 1.6 cm/yr from the average annual precipitation of 46 to 62 cm. The three reservoirs reduced streamflow at the reservoir site by 74 to 97%, but 90 to 95% of the retained runoff was calculated to contribute to ground-water recharge.
Several sensitivity analyses for model inputs were done. Results showed that, the ratio of the average annual inflow volume from the watershed area to the reservoir storage volume was the most sensitive input variable tested.
