Water Exchange Mechanism and Red Tide Prediction in the Tanabe Bay

京都大学防灾研究所（DPRI）与近畿大学渔业实验室（FISH LABO）和白滨渔民协会合作开展了这项研究项目，以预测田边湾的赤潮。DPRI有监测海况的设施，FISH LABO在同一海湾有养鱼设施。这些研究机构与田边湾渔民协会合作，在该项目中建立了赤潮预测系统以及水质、水循环系统和生态系统监测。以下是该项目的主要成果和活动。(1)水交换预测模型：采用改进的POM代码构建了海湾水循环准三维数值模型，其中标量输运方程在z坐标系下采用TVD格式求解，水流动力学方程在sigma坐标系下进行积分。(2)水质和浮游生物大量繁殖：1998-2001年4年间，利用风暴潮观测塔和养鱼筏遥测，对夏季浮游生物细胞和理化、生物水质进行了密集观测。同时，利用观测船对整个海湾的水质和浮游生物细胞进行了剖面测量。(3)采用海岸环境遥测与人工神经网络（ANN）相结合的方法，建立了田边湾内层浮游植物密度实时预测系统。使用的学习系统是反向传播模型。利用1999年和2000年的观测数据和输出数据（硅藻和鞭毛藻的细胞浓度）进行人工神经网络的学习。2001年的预报明确了输入温度、盐度、太阳辐射和气温（溶解氧）数据，对预报系统进行了改进。(4)本项目测量的所有数据均可随光盘分发。

Disaster Prevention Research Institute(DPRI), Kyoto University has run this research project in partnership with Fisheries Laboratory of Kinki University (FISH LABO) and Shirahama Fisherman's Association to predict red tide in the Tanabe Bay. DPRI has monitoring facilities for sea states and FISH LABO has fish farming facilities in the same bay. Collaboration between these research institutes and the fishermen's society in the Tanabe Bay established a red tide prediction system as well as water quality, water circulation system and ecosystem monitoring, in this project. The followings are the main results and activities of the project.(1)Water Exchange Prediction Model : Quasi-three-dimensional numerical model for water circulation in the bay was constructed with improving POM code, where scalar transport equations were solved with TVD scheme in z-coordinates and current dynamics equations were integrated in sigma-coordinate.(2)Water Quality and Plankton Blooming : Intensive observations of plankton cells and physical, chemical and biological water qualities in summer were conducted for four years 1998-2001 by using the telemetry of Storm Surge Observation Tower and fish farming rafts. Profile measurements of water qualities and plankton cells were also conducted by observation ship in whole area of the bay.(3)A real time prediction system of phytoplankton density in the inner part of the Tanabe Bay was established with combination of coastal environmental telemetry and artificial neural network (ANN). The used learning system is the back-propagation model. Observation data and output data (cell concentration of diatoms and dinoflagellates) in 1999 and 2000 was employed for a learning of ANN. Prediction in 2001 made clear that input data of temperature, salinity, solar radiation and air temperature (dissolved oxygen) are included, the prediction system is improved.(4)All data measured in this project can be distributed with CD-ROM.