Water that infiltrates bedrock is not recorded at outlet weirs, and studies have reported that in some catchments this deep percolation cannot be ignored. In this study, we used the chloride mass balance method to estimate deep percolation in a 1-ha mountainous catchment in a Tertiary formation. We determined the chloride concentrations of throughfall, stemflow and streamwater during baseflow and stormflow, and groundwater. The analysis of chloride inputs and streamwater output indicated that the annual deep percolation was in the range 410 to 640 mm/yr and the mean annual percolation was 530 mm/year, or 22 % of the annual mean precipitation. The input of throughfall and stemflow and the output during stormflow are necessary to estimate deep percolation using the chloride mass balance method in forest catchments.

View full abstract

Loss of phosphorus (P) from forested ecosystems is usually small. In Japan, the input-output budget of P in forested watersheds has in many cases been shown to have larger inputs than outputs (Input > Output). However, it is possible that P budgets have been inadequately evaluated because output has mainly been calculated using the Periodic method, which ignores changes in P concentrations during storm events. This study sought to clarify the effect of the incorporation of changes in P concentrations during storm events in calculations of the output on the P budget. Stream water was sampled weekly, as well as sequentially at hourly intervals in each of 18 storm events in a small mountainous watershed of a plantation forest of Japanese cypress (Chamaecyparis obtusa). Based on a four-year dataset, the Periodic and the ∑L-∑Q method, which considers changes in P concentrations during storm events, were compared. The result showed that output calculated by the ∑L-∑Q method was on average three times higher than that calculated by the Periodic method. Thus, in dry years, the Periodic method showed Input > Output whereas with the ∑L-∑Q method, Input < Output. Because of the strong affinity for P of soil particles, sediment transport usually causes large P losses during storm events. This results in rapidly increased P concentrations in a stream. Our results showed that the ∑L-∑Q method gave an accurate estimate of P output as it considered increased P concentrations in storm events.

View full abstract

Water retention and hydraulic properties of undisturbed tropical soils collected from many regions of Indonesia were analyzed to estimate infiltration characteristics of the soils. Soil texture was classified based on International Society of Soil Science (ISSS) classification. The van-Genuchten model was used to estimate the relationship between water content and matrix potential at pF=1, pF=2, pF=2.54, pF=4.2. The 165 soil water retention data were used to optimize parameters of the model and to find the air entry value. Green-Ampt and Philip's infiltration models were applied to characterize soil infiltrability of each textural type. The Nash and Sutcliffe's efficiency was used to evaluate numerical simulation of cumulative infiltration of Green-Ampt's infiltration model compared to the results of laboratory experiments. The 165 soil samples were classified and were optimized into 10 ISSS textural types: heavy clay, sandy clay, sandy clay loam, sandy loam, sand, light clay, clay loam, loam, silty clay, and silty clay loam. The results of performance evaluation of Green-Ampt's infiltration model showed that Green-Ampt's infiltration model can describe infiltration characteristics by using soil water retention and hydraulic properties data. The tropical soils based on soil texture exhibit contrasting infiltration characteristics as indicated by infiltration rate, length of wetting front and sorptivity. The characteristics of soil infiltrability are mainly influenced by hydraulic conductivity, initial water content, and matrix potential at the wetting front.

View full abstract

The storage function model developed by Hoshi et al. has been widely used for the rainfall-runoff analysis due to the ease of expressing the nonlinear relationship of rainfall-runoff events with simple equations and its ability to provide relatively easy computation. Many case studies of the model are reported for larger and mountainous river catchments in Hokkaido. On the other hand, the rainfall-runoff process of urbanized small river catchments is quite different from such mountainous catchments, because the water retention capacity of the land has drastically declined due to the increased amount of impervious surface areas and the extension of sewage system.
In this study, new storage function model for urban flood, which considers outflow characteristics of the combined sewer system, is developed. The developed storage function model and other two conventional runoff models, i.e. sequential rational method and Hoshi's storage function model, are applied to the upper river catchment of the Kanda River in Tokyo, which is the typical urbanized area prone to urban floods. The results show that the developed model is able to reproduce the observed hydrograph very well for five objective functions with accurate peak discharge and the total amount of discharge, whereas the performance of other two models are not good enough. In addition, the validity of outflow characteristics through combined sewer system estimated by the developed model is evaluated.

View full abstract

Grid-based distributed hydrological models are grid-size variant; the model parameters have to be readjusted when the resolution of river networks is varied. A simple scaling algorithm has been recently proposed to rescale fine-resolution networks to any coarser spatial resolution. This study evaluates impacts of the proposed scaling algorithm on discharge simulations in different scales of river networks. Discharge simulations by the Muskingum-Cunge routing method using different resolutions of river networks have been applied for the Hayakawa River (the branch of the Fujikawa River) and the Yellow River. It has been found that the simulation results have hardly varied even if the networks have been aggregated to a very coarse scale, because the new scaling algorithm preserves the geomorphologic characteristics of the fine-resolution network, such as river length, elevation gradient, and so on, with sufficient accuracy.

View full abstract