The initial moisture conditions of a watershed immediately before a rainfall event begins affect the characteristics of the subsequent storm runoff. The authors found a drainage basin in the Soya Hills, northern Japan, where two different types of flood hydrograph are clearly distinguished for each flood event, and investigated the effect of the initial moisture conditions of the watershed and rainfall on the shape of the hydrograph.
The watershed studied (Fig. 1) is underlain by Neogene shale (Wakkanai formation, Wk). With joint planes about 10cm apart and fissure zones approximately 10m apart, Wk is permeable and does not constitute a hydrological basement. As a result of water level observation on an experimental watershed (drainage area A=0.69km2) in a Wk-hill, two types of flood event are recognized.
Type-1 floods (Fig. 2-a) have only one runoff peak corresponding to each rainfall event. This peak appears within 2 hours after the termination of rainfall.
Type-2 floods (Fig. 2-b) have two runoff peaks for one rainfall event. The time lag between the primary runoff peak and the termination of rainfall is similar to that of Type-1 flood events, while the secondary runoff peak occurs 7 to 26 hours after the primary one. This time lag varies for each event (Table 2).
The runoff ratio of Type-2 flood events are more than 10times greater than Type-1 flood events (Fig. 4). For Type-2 flood events, 90% of the total storm runoff is produced by the secondary peak runoff.
Type-1 floods and the primary peak runoff of Type-2 floods have similar runoff ratios (Fig. 5). These runoffs are explained as the direct rainfall runoff and shallow groundwater discharge from moist areas near the channel.
The conditions under which the secondary peak runoff occurs are expressed by the combination of initial discharge (q0) and rainfall (P1) before the termination of primary peak runoff (Fig. 6); the secondary peak runoff appears when either q0 or P1 is sufficiently large, while it is not detected when both qo and P1 are small.
The authors defined the flood-controlling storage (S) as the amount of water storage in a basin directly contributing to storm runoff and attempted to express this value as a function of discharge (q). First, a two-component recession formula for one recession period is established as
q(t)=e-0.0250t-1.78+1/(0.0103t+3.67)2
where t is the time (h) after the beginning of recession and q (t) is discharge (mm/h) at t.
Then, using the theoretical relationship between q and S for each runoff component, we obtain
S(t)=e-0.0250t-1.78+1/(0.0103t+3.67)2
where S(t) is the flood-controlling storage (mm) at t. These two formulae implicitly represent the relationship between q and S (Fig. 7). Using this relationship, conditions under which Type-2 flood events occur are expressed as
S0+P1>32mm
where S0 is the initial water storage and P1 is rainfall before the termination of primary peak runoff (Fig. 8). There exists, however, an example when the secondary peak is not clearly detected when S0<20mm.
The value of S0+P1 is also related to the time lag between two peaks of Type-2 flood events; the time lag TL (h) is expressed as
TL=39.5-0.512 (S0+P1) R2=0.610 (Fig. 9).
This means that the response of secondary peak runoff is faster when the watershed is relatively wet.
