DEVELOPMENT OF FORCASTING AND WARNING SYSTEM FOR DEBRIS FLOWS,MT.UNZENDAKE

Debris flows have frequently occurred in the Mizunashi River and caused severe damage in the downstream area at Unzen Volcano by deposition of large amount of sediments。因此，这对预测debris的流动很重要。本研究中获得的结果如下：（1）在两个地点，Debris流量的实地观察和测量已被运送到Unzendake。无线电电流计和超声波水位测量器用于获得表面的速度、深度和放电。Peak discharge was found Q= 185 m ^3/s at the Mizunashi River and Q= 40 m ^3/s at the Nakao River。（2）一个神经网络被用来建立一个debris流的失效模型。1993年6月12日至13日收集了水松河的速度和深度数据。用于学习、调试流Q （t）和10分钟的雨水，用于20 - 70分钟， r （t-20）， r（t-30），...， r （t-70），用作输入单元。已确认的价值观显示与观察者的最接近协议。 ... More 由于水俣河上的其他可靠数据也被捕获了，所以很难通过使用水俣其他事件来验证模型。However，日本建筑部对存款的估计是由日本建筑部衡量的。该模型的适用性可以通过比较debris流量的总数量与计量的存款量进行检查。从预测的水文图中集成了大量的碎片流，显示了与观察到的名单相当好的协议。神经网络是有用的，可以用来进行debris流的runoff分析。（3）在先前的研究中，使用反向传播方法的神经网络（BP）已应用于预测Debris流的情况。我们还发现了这样的模型，可以估计这是一场灾难。在本研究中， LVQ （学习向量量化）是为了提高预测的准确性。LVQ和英国石油公司已应用于Unzen和Sakurajima火山上的碎片流动。比较结果之间的两种方法确认LVQ在预测中有一个优势。英国石油公司的模型被用来在Unzen Volcano找到关键条件。该方法的验证性是由Debris Flow理论的出现标准来证明的。Less（低）

Debris flows have frequently occurred in the Mizunashi River and caused severe damage in the downstream area at Unzen Volcano by deposition of large amount of sediments. Therefore it becomes important to predict the debris flows. The results obtained in this study are as follows :(1) Field observations and measurements of debris flows have been carried out at two locations on Mt.Unzendake. Radio current-meter and ultrasonic water level gauge were used to obtain surface velocity, depth and discharge. Peak discharge was found Q=195m^3/s at the Mizunashi River and Q=40m^3/s at the Nakao River.(2) A neural network is used to make a runoff model of debris flow. The data of velocity and depth of debris flow were collected at the Mizunashi River on 12-13 June, 1993. For learning, the discharge of debris flow Q (t), and ten-minute rainfall 20 to 70 minute ahead, r (t-20), r (t-30), ・・・・・・, r (t-70), are used as the input units. The recognized values show close agreement with the observed ones. … More As on other reliable data of hydrograph has been taken at the Mizunashi River, it is hard to verify the model by using the hydrograph of other events. However, the amounts of deposits were measured by the Ministry of Japan Construction. The applicability of the model can be checked by comparing the total amount of debris flow with the measured amounts of deposits. The volume of debris flow integrated from the predicted hydrographs show fairly good agreement with observed ones. Neural networks are useful for making runoff analyzes of debris flows.(3) In the previous study, the neural networks with back-propagation method (BP) were applied to predict the occurrence of debris flow. It was also found that this model is useful to estimate the critical rainfall. In this study, LVQ (Learning Vector Quantization) is introduced to improve the accuracy of the prediction. The LVQ and BP are applied to the debris flow at Unzen and Sakurajima Volcanoes. Comparison between the results by both methods confirms that LVQ has an advantage in prediction. The BP model is used to find the critical condition at Unzen Volcano. Validity of the method is demonstrated by the theory of the occurrence criteria of debris flow. Less