An application of Granier method for a long-term measurement of sap flux (u) was discussed. A continuous measurement of u was conducted for over one year for the test stand of Japanese red pine (Pinus densiflora Sieb. et Zucc.). The observation results of the temperature difference between a pair of sensors (ΔT) showed that ΔT was a good indicator of diurnal and long-term variation in u. However, daily maximum of ΔT (ΔT_MAX), which is observed at the time of daily minimum of u, had decreasing trend over a long-term period independent of seasonal changes in some environmental factors. If ΔT₀, which is ΔT when u is equal to zero, is regarded as the annual maximum value of ΔT, this trend has a possibility to lead some errors in evaluating u. New determination method of ΔT₀ was discussed in the present study, and concluded that ΔT₀ was determined as ΔT_MAX. Sap flux estimated by Granier method based on the new determination of ΔT₀ corresponded with that by heat pulse method. Furthermore, the value of daily transpiration of the test stand estimated by the Granier method was good agreement with previous results obtained at the same observation forest. The conclusion of the present study is that Granier method is an efficient and reasonable method for long-term measurement of transpiration of a tree, if the fluctuation of ΔT_MAX is considered.

View full abstract

In this paper, a long-term runoff model including a evapotranspiration sub-model was developed to discuss the application of the model, focussing on the relationship between actual evapotranspiration and soil moisture conditions in catchments. At first, the properties of actual evapotranspiration from two experimental catchments, forest and reclaimed upland field catchment in Ehime prefecture, were discussed. It was found from the result that evapotranspiration ratios were high in the wet seasons, although those were quite low in the dry seasons. Then the evapotranspiration sub-model, which represents the relationship between evapotranspiration ratio and soil moisture condition in catchment by a simple equation, was developed and incorporated into the long-term runoff model. Thereafter, daily discharge was calculated for the catchments by the runoff model. The results showed acceptable agreement: between the observed and calculated discharge, between calculated evapotranspiration by the evapotranspiration sub-model and estimated actual evapotranspiration by the short term water balance method. The identified parameters of the runoff model represented some catchment properties. As the results, since the evapotranspiration sub-model was very simple and able to describe the change of evapotranspiration easily, it was expected that it will be possible to evaluate the effects of change in land use on the hydrologic cycle in a catchment by the model.

View full abstract

The Nearest-Neighbor (NN) method, which is originally one of the statistical pattern recognition methods, is also very useful for time series forecasting. In the runoff forecasting using the NN method, some past rainfall-runoff events that are the most similar to the current event are selected, and then the next flow is estimated as the sample average of successive flows to these similar events. In this study, we apply the NN method to the real-time runoff forecasting using lead-time of 1-day, 2-day and 3-day in the Yodo River basin and then investigate some practical application techniques of the NN method. Results show that the accuracy of 1-day ahead forecasting is very good in spite of its simplicity. The relationship between length of data periods and forecasting errors indicates that about 10 years data period is required for reliable forecasting. It is also shown that the NN method can be used for runoff forecasting with rainfall data of a small number gauges.

View full abstract

This paper describes automatic optimization methods for the parameters of our mountainous watershed model. Because this model simulates not only the groundwater runoff but also the groundwater levels at observation well, multi-objective optimization method has to be employed. Real-coded GA and MOCOM-UA, effective global optimization methods suitable for multi-objective optimization, are examined in this paper. After examining the basic performance of the both method by numerical experiments, we tested the applicability of them with observed data of Shiozawa mountainous experimental basin that has an unconfined aquifer.

View full abstract

It is well known that the movement of water in a soil can be described by unsaturated flow equations. The relationship between suction and water content is needed to solve these equations, and this relation can be obtained through indoor experiments. The water retained in pores among particles with radii in the range of 10^-3mm to 1mm can be regarded as comprising most of the free water. The free water forms a pendular ring between two particles under unsaturated conditions due to surface tension acting on the free water. The pendular ring depends on the geometrical configuration of two particles. In this paper, a conceptual model for evaluation of the suction and water retention volume among spheres with different radii is proposed from a microscopic viewpoint. The results of quantitative analysis using this model are presented. In order to calculate the suction and water retention volume in natural soil, the coordination number of a particle must be determined. Assuming that the particle size distribution and porosity are known, the average coordination number can be estimated. The results of the proposed model agree well with the experimental results in the case of relatively low water content. However, as the proposed method focuses on only isolated pendular rings between two particles, it is impossible to extend our method to the higher water content region.

View full abstract

Knowledge of low flow characteristics is important in the Pangani river basin in Tanzania as this region has favourable condition for agriculture as well as water for hydropower generation. During dry seasons when low flows are experienced, coflicting presures on water distribution between irrigation projects upstream of Nyumba ya Mungu dam and hydropower generation occur. Of the two main rivers that provide inflow to the reservoir, the Ruvu river shows higher frequencies of low flows than the Kikuletwa river. Droughts assessed using Q70 threshold indicate larger deficit volumes and drought durations on kikuletwa river than Ruvu river. The peak over threshold, (POT), method provides reasonable results on low flows and drought characteristics in the region. This is coupled with the quantile function plots approach and the mean excess plots for identifying the candidate model for fitting the distributions of the extreme values and the number of those extremes.

View full abstract