This study develops a hydrological model with a distributed groundwater flow component, such that it can predict daily groundwater levels and surface runoff in an upland experimental basin. Using this hydrological model, the authors predict hydrological changes when inhabitants do not use inverting wells. Furthermore, they make clear the effect of infiltration facilities quantitatively with the above hydrological model.

View full abstract

Fukuoka, a megacity in Japan, suffered severe shortage of its normal water supply, for a long period of time in 1978. This crucial experience led to the study of a new system which is adaptable to cover water supply being suffered by shortage of water frequently.
Considering the difficulty of water resources development in Fukuoka's self-watershed, new water source was taken account in neighboring watershed, where Fukuoka has not predominated water right yet.
To solve this problem, the system requests that an occasional conditioning reservoir could be constructed somewhere. Specific water right which is valid only when source water is abundant should be admitted. On these conditions, the system is represented by either with the occasional conditioning reservoir, or without it. It was confirmed that the system was applicable to the circumstance of 1978's drought in Fukuoka, showing effective use of the occasional conditioning reservoir. Prior to apply the system in long time range, a model city, a model water supply system and model time series of water flow were settled to simulate a general circumstances.

View full abstract

To handle the time-dependent and multi-point properties of hydrological data, all data should be treated and analyzed in vector. Pattern recognition theory is powerful and useful for these concepts. For classification, there are several important circumstances, such as (i) the objective functions suited to hydrological analysis, (ii) performance of the global optimum and (iii) optimum number of clusters. In this paper, random K-means algorithm is introduced giving many different initial cluster centers and compared with DHB and ABF algorithms. On the other hand, three objective functions are proposed. The first is to estimate the whole shape as the distance between sample data and cluster center in time. The second is to estimate the peak value having great influence on flood planning. The third is to estimate the gradient of sequence. Then the aggregate objective function is formulated by maximizing three objective functions. Finally, the optimum number of clusters is decided under the condition of constraints subjected to the intra distance represented by aggregate objective function and the inter distance. These threshold values are gained by comparing multi-optimization and AIC techniques.

View full abstract

Two sets of lysimeters (soil packed box with artificial rainfall equipment) with wind tunnels have been installed. In one lysimeter, grass was planted, while the other was kept bared. As for air flow, insolation intensity, wind velocity, air flow temperature, humidity and CO₂ concentration were measured. As for soil layer, the water content and soil temperature were recorded.
Considerable changes in temperature, huridity and CO₂ concentration have been observed in the grassed lysimeter, while less changes have been detected in the bared lysimeter, showing the remarkable effect of plant transpiration and photosynthesis for modifying the environment of atmosphere.
A brief description of the numerical simulation of 'soil-plant-air' system is given to confirm the experimental result.

View full abstract