Abstract:The influence of bedrock subsurface flow on storm runoff generation was investigated in Japan in two regions in Japan underlain by three bedrock types. One region, with approximately 2500 m of relief, is located in the Japan Alps, central Japan (In a region), where six small forest-covered watersheds underlain by granite and Mesozoic shale were studied. Two of these watersheds were instrumented to monitor stream discharge and the other four are springs occurring at the bedrock exposure. The other study area is in northern Kyoto prefecture, western Japan (Oe region). Here, spring distribution and seasonal flow were monitored in two steep basins underlain by serpentinite rocks. Distinctly different runoff responses were observed: the granite watershed exhibited a large and rapid runoff peak that coincided with the rainfall peak (type 1); and the shale and serpentinite basins exhibited small initial runoff peaks followed by a maximum peak discharge five to ten times greater with a convex recession limb (type 2). Runoff response from bedrock springs had characteristics similar to type 1 hydrographs; however, discharge increased only when the antecedent precipitation index was large and the runoff peak was delayed between 10 h and 1 week after rain events. The specific discharge of the springs decreased with increase in altitude in the Oe region, especially in dry periods. This decline in discharge with elevation suggests that a deep subsurface flow system through bedrock fissures contributes to the storm water flow in serpentinate basins. When comparing runoff response and spring discharge, slow or double peak runoff response may be a good indicator of the influence of bedrock subsurface outflow on storm runoff generation in steep mountainous regions.

Base flow from 66 tributaries in watersheds underlain by serpentinite , granite, and Paleozoic shale were measured in Oe yama region, Kyoto Prefecture, Western Japan. The base flow measurements sufficiently after rainfall were conducted during the snowmelt season (April) and summer dry season (August). Spatial variation of specific discharge was greater in the serpentinite basins than those in the granite and the Paleozoic shale basins . In addition, there are many tributaries without discharge in serpentinite basins, even though they have large drainage areas and well-defined valleys . On the contrary, the specific discharges in the granite and the Paleozoic shale basins are relatively constant among tributaries. The average specific discharges in the serpentinite basin are smaller in August than that in April , whereas little difference between base flow discharges of April and August were observed in both the granite and the Paleozoic shale basins . These findings suggest that the seasonal variation of groundwater table level in bedrock is larger in serpentinite mountainous region than that of the granite and Paleozoic shale areas. Hence, we expect that lithology has a crucial effect on spatial variation , quantity and seasonal change of base flow.

No abstract

No abstract

Abstract:Eagleson's classical regional flood frequency model is investigated. Our intention was not to improve the model, but to reveal previously unidentified important and dominant hydrological processes in it. The change of the coefficient of variation (CV) of annual maximum discharge with catchment area can be viewed as representing the spatial variance of floods in a homogeneous region. Several researchers have reported that the CV decreases as the catchment area increases, at least for large areas. On the other hand, Eagleson's classical studies have been known as pioneer efforts that combine the concept of similarity analysis (scaling) with the derived flood frequency approach. As we have shown, the classical model can reproduce the empirical relationship between the mean annual maximum discharge and catchment area, but it cannot reproduce the empirical decreasing CV-catchment area curve. Therefore, we postulate that previously unidentified hydrological processes would be revealed if the classical model were improved to reproduce the decreasing of CV with catchment area. First, we attempted to improve the classical model by introducing a channel network, but this was ineffective. However, the classical model was improved by introducing a two-parameter gamma distribution for rainfall intensity. What is important is not the gamma distribution itself, but those characteristics of spatial variability of rainfall intensity whose CV decreases with increasing catchment area. Introducing the variability of rainfall intensity into the hydrological simulations explains how the CV of rainfall intensity decreases with increasing catchment area. It is difficult to reflect the rainfall-runoff processes in the model while neglecting the characteristics of rainfall intensity from the viewpoint of annual flood discharge variances.