In this study, complicated atmospheric patterns in a rainy season (BAIU) in the southwestern Japan was classified into eight groups, using the Self-Organizing Map (SOM) algorithm, which converts complex nonlinear features into simple two-dimensional relationships. The groups can be basically represented by five meteorological fields;(1) dry air masses, (2) anti-cyclonic flow due to the Pacific high pressure, (3) the existence of the BAIU front between dry and wet regions, (4) the intrusion of a large amount of water vapor, (5) passages of typhoon and low pressure system. One of the groups has notable feature represented by high precipitable water accompanied by strong wind components (Low Level Jet), which is a typical meteorological field that causes disastrous heavy rainfall events in the northern Kyushu. Therefore, it may be expected that the classification of meteorological fields and associated extraction of heavy rainfall phases contribute to determining whether heavy rainfall occurs or not in a target area as well as the enhancement of the accuracy for the rainfall prediction.
Wind induced currents in Lake Ogawara is discussed being based on field experimental data obtained in autumn of 2006. In the experiment, three ADCPs were set on the bottom of the north, the center and the south water areas for two months. The data showed a periodic appearance of intense shear layer in the metalimnion that seemed to be generated by internal seiche. Measurement from three boats equipped with ADCP was conducted for three days in the same water areas under predominant wind direction of west and south. The data suggested the existence of rather stable horizontal currents in the surface mixed layer. These facts suggest that substances entrained from anaerobic hypolimnion by the shear motion are carried by the stable horizontal current in the surface layer to develop uneven condition of water environment.
In this study, complicated atmospheric patterns in a rainy season (BAIU) were classified into 100 patterns, using the Self-Organizing Map (SOM) algorithm. In addition, rainfall probability exceeding 30mm/6h in Northern Kyushu were predicted using the Artificial Neural Networks (ANNs). Considering high-dimensional complicated atmospheric patterns can be replaced by the location of a node specified in the two-dimensional map, the coordinates (two elements) of the node were used for the learning of ANNs. As a result, using the SOM, nodes characterized by the Low-Level Jet and associated ample water vapor distribution were significantly related to heavy rainfall events. Paying attention to heavy rainfall events, predicted rainfall probabilities using ANNs showed clear-cut relationships between meteorological field pattern and heavy rainfall.Therefore, it can be expected that the incorporation of the coordinates of a node in the map into ANNs contributes to the down-sizing of the ANNs structure and helps the understanding of input-and-output relationships.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.