| This study reports the example which carried out dam examination filling based on the models of total volume prediction by snowmelt discharge and the comparison and integration of these models、 and based on the reliability information on the predictive value using time series data like the depth of snow cover and the temperature、 snowmelt runoff、 etc. updated every day. The construction of dam is complete after having confirmed the safety around the dam body and reservoir area by examination filling. There are a lot of examples of planning and executing the dam examination filling by making use of the snowmelt discharge in the snowy watershed where the proportion of total volume of it is large. The test impounding of dams is conducted to verify that newly constructed dams can be managed without problems. The reservoir water level is increased and decreased according to a plan and its behavior is monitored for a certain period to confirm the safety of the dam body、 outlet facilities and ground around the reservoir. Test impounding of the Rumoi Dam in Japan、 which was completed in 2010、 was planned for the three months from March to May using snowmelt runoff single year impounding). The estimation results for the scheduled year (2006、 normal water level) indicated that impounding would start on March 6、 and the surcharge and minimum water levels would be reached on April 23 and May 26、 respectively. In reality、 while impounding did start on March 6、 the surcharge level was reached on May 3、 eleven days later than scheduled、 since the total volume of snowmelt was small. In addition、 the minimum water level was reached on May 22、 five days earlier than scheduled. In this way、 it was anticipated that the water level necessary for testing would not be reached、 and the information of the total volume inflow prediction value was very important. The models of the correlation between the precipitation and the total volume of snowmelt discharge using previous data、 the snow-survey method and the heat balance method were applied to predict the total volume inflow value. And these models were compared and integrated. For example、 the estimation by snow-survey method results vary greatly、 the use of values (snow depth、 etc.) measured in the basin is promising as a means of improving the reliability of accuracy. Although the estimation value is very variable when limited observation data is used、 and it is difficult to select observation stations that are highly correlated to the amount of water contained in the basin、 based on past survey results. Therefore、 the snow depth estimation with remote sensing technology was examined. It is effective measure in case of years with small snow accumulation、 not only the snowsurvey method but also the heat balance method. This study shows that it is possible to improve the prediction value comparing and integrating the three models mentioned above、 using time series updated every day data in addition. |
