Study on the Actual and Optimum Water Use in Ohishi Diversion Weir Command Area (1）

The water level of the irrigation and drainage canal was observed using the water level indicator in the 6ha paddy fields which is selected as the study site and located in the Ohishi diversion weir command area along the Chikugo River. The characteristic of water management in paddy field concerning the weather or growth stage of rice could be clear using the result of the observation. The variation of the water requirement rate depending on the growth stage of the rice or the location of the field could be estimated. The tank model of the paddy field was introduced in the study site. The method of Shuffled Complex Evolution Algorithm (SCE-UA) which is one of the global optimization method was adopted for the tank model parameter estimation. The advantage of this method is that algorithm is relatively easy, the local optimum value can be avoided and it is independent of the initial value. The effectiveness of the SCE-UA method for the parameter identification was proved in this study. The calculated value under the precipitation had good agreement with the observed data, but the calculation in the sunny days didn\u27t fit the observation. There are errors between calculated and observed values under the condition of the manual operation for the water intake. The water level of the paddy field was simulated assuming the future change of water intake in the study site. According to the result of simulation, it proved that it would be difficult that water level of the paddy fields is maintained under the condition of the present water intake of 80%. The present water intake is appropriate in the study site and it is probable that much water intake would be necessary in case of the future decrease and aging of farmworkers.（1）筑後川中流域大石堰掛かり受益地内から選定した約6haの圃区を対象として行なった水田用水路および排水路の水位連続観測から，天候や稲の生育一別による，水管理の特徴を捉えることが出来た．自記水位計による観測で取水実態が把握可能であることが解かった．また，生育期や水田のロケーションに依存した減水深の変化もとらえることができた．（2）対象地区における水田タンクモデルを構築し，パラメータ最適化手法には大域的探索法のひとつであるSCE－UA法を適用した．SCE－UA法は，アルゴリズムが比較的容解であり，局所解に陥りにくくまた初期値への依存性もなく，タンクモデルパラメータ同定法としての有効性を確認した．モデルは，降雨時においては実測値と良い一致が認められたが，晴天時における再現性が不十分であった．人為的な用水操作がなされた場合に計算値と実測値との誤差が生じている．（3）モデルを用いて対象地区における取水量が変化した場合の湛水深シミュレーションを行なった．シミュレーション結果からは，取水量を現状の80％以下に落すと十分な湛水深を保つのは困難となることが予測でき，対象地区の現状の取水量は妥当な量であると考えられる．しかし今後農業従事者の高齢化，少数化が進んだ場合，現状以上の水量が必要となってくる可能性がある