Flood simulations 1993

1. Sensitivity to rainfall forcing The streamflow at Elizabeth is simulated reasonably with the rainfall fields obtained using the standard hypsometric method, especially in the case of spring peaks caused by passing frontal storms (Figures 20.7(a) and (b)). During the summer season, all simulations capture well summer peaks and low-flow variability. At Dailey (not shown), the streamflow response is generally overestimated, especially in spring and early summer. Since only dependence with elevation was used to describe orographic effects on rainfall, other factors such as ridge-valley gradients, rainshadow, and so on, were not included in the parameterization of orographic effects, and thus rainfall is overestimated at Dailey.

Although model simulations at 1 km were performed for all watersheds shown in Figure 20.1 and Table 20.1, only one of the hydrographs at Elizabeth was selected to show here (Figure 20.8). The model results and the observations match well independently of the rainfall distribution (not shown). As compared to the 5-km results, all spring peaks, except the maximum peak at the end of April, are better reproduced at 1-km resolution as expected.

2. Structural stability Inspection of the 5- and 1km 1993 simulations for Elizabeth, Enterprise and Dailey reveals that the contribution of subsurface flow is much larger at 1-km resolution, especially in the spring period (e.g. Figures 20.7 and 20.8). Similar to the 1988 simulations at this resolution, the model generated streamflow is comprised mostly of baseflow and interflow. Note, however, that as opposed to what happens in 1988, the differences for the 1993 simulations are larger in early May when heavy rainstorms are more frequent, leveling off in late spring (Figure 20.9). Vegetation controls on soil moisture and subsurface flow response are not critical here, because soil water availability to plants is not a limiting factor for evapotranspiration in 1993.

1993-1 km

600 400 200 0

Observed streamflow Simulated streamflow

1993-1 km

600 400 200 0

Observed streamflow Simulated streamflow

---Simulated subsurface flow

15 Apr 01 May 15 May 01 June 15 June 01 July 15 July 01 Aug 15 Aug 31 Aug

Figure 20.8 Comparison of observed and simulated streamflow hydrographs at Elizabeth, PA, between 09 April and 31 August 1993 at 1-km resolution. The model was forced with distributed rainfall estimated using the modified weighted area method

---Simulated subsurface flow

3 tat

15 Apr 01 May 15 May 01 June 15 June 01 July 15 July 01 Aug 15 Aug 31 Aug

Figure 20.8 Comparison of observed and simulated streamflow hydrographs at Elizabeth, PA, between 09 April and 31 August 1993 at 1-km resolution. The model was forced with distributed rainfall estimated using the modified weighted area method

Dailey

Dailey

Enterprise
Elizabeth
Figure 20.9 Comparison of the time evolution of the ratio of subsurface flow (Qs) to total streamflow (Q) at 1- and 5-km resolutions at Dailey, Enterprise, and Elizabeth in 1993

Surface runoff production is overestimated at 5-km resolution as compared to the 1-km simulations, suggesting that when soils are near saturation (during April 1993), the subsurface flow response (vertical and lateral) at 5km is ''slower'' than that at 1-km resolution. This is consistent with a longer characteristic timescale of subsurface flow: smoother topography at 5-km resolution implies lower hydraulic head gradients, and therefore reduced subsurface flow. Model calibration to replicate observed streamflows would thus lead to an ''optimal'' value of hydraulic conductivity at 5km larger than that at 1 km. On the other hand, consistent with Bindlish and Barros (2000) who found that the optimal magnitude of hydraulic conductivity determined through calibration of a hydrologic model varies inversely with the spatial variability of rainfall, the ''optimal'' value of hydraulic conductivity at both resolutions in the summer would be smaller than that in the spring. Scale dependence of soil hydraulic properties would be therefore imposed artificially to accommodate not only changes in subsurface gradients but also in the type of rainfall forcing (i.e. frontal versus convective storms). Another interesting feature is that the differences between the two simulations quickly become very small

June

July

August

0.1 April

June

July

August

June

July

August

June

July

August

August

Figure 20.10 Intraseasonal variability (monthly timescale) of simulated hydrologie indices at Dailey, Enterprise, and Elizabeth in 1988: (a) volumetric soil moisture content in the root layer (9 ); (b) evaporative fraction (E/P ); and (c) runoff coefficient (Q/P )

August

Figure 20.10 Intraseasonal variability (monthly timescale) of simulated hydrologie indices at Dailey, Enterprise, and Elizabeth in 1988: (a) volumetric soil moisture content in the root layer (9 ); (b) evaporative fraction (E/P ); and (c) runoff coefficient (Q/P )

in late spring and early summer, confirming our hypothesis that when soil water availability is high, évapotranspiration plays a lesser role in controlling the partitioning of surface/subsurface runoff contributions to streamflow as noted earlier in the 1988 simulations.

0 0

Post a comment