Deep learning has ushered in many breakthroughs in vision‐based detection via convolutional neural networks (CNNs), but the vibration‐based structural damage detection by CNN remains being refined. Thus, this study proposes a simple one‐dimensional CNN that detects tiny local structural stiffness and mass changes, and validates the proposed CNN on actual structures. Three independent acceleration databases are established based on a T‐shaped steel beam, a short steel girder bridge (in test field), and a long steel girder bridge (in service). The raw acceleration data are not pre‐processed and are directly used as the training and validation data. The well‐trained CNN almost perfectly identifies the locations of small local changes in the structural mass and stiffness, demonstrating the high sensitivity of the proposed simple CNN to tiny structural state changes in actual structures. The convolutional kernels and outputs of the convolutional and max pooling layers are visualized and discussed as well.

During the Fukushima Dai-ichi Nuclear Power Plant (NPP) accident, radioactive cesium was released in the environment and deposited on the soils. Depth profiles of radioactive cesium in contaminated soils provide useful information not only for radiation protection and decontamination operations but also for geoscience and radioecology studies. Soil samples were collected using a scraper plate three times between December 2011 and December 2012 at 84 or 85 locations within a 100-km radius of the Fukushima Dai-ichi NPP. In most of the obtained radioactive cesium depth profiles, it was possible to fit the concentration to a function of mass depth as either an exponential or hyperbolic secant function. By using those functions, following three parameters were estimated: (i) relaxation mass depth β (g cm(-2)), (ii) effective relaxation mass depth βeff (g cm(-2)), which is defined for a hyperbolic secant function as the relaxation mass depth of an equivalent exponential function giving the same air kerma rate at 1 m above the ground as the inventory, and (iii) 1/10 depth L1/10 (cm), at which the soil contains 90% of the inventory. The average β value (wet weight) including ones by hyperbolic secant function in December 2012, was 1.29 times higher than that in December 2011. In fact, it was observed that depth profiles at some study sites deviated from the typical exponential distributions over time. These results indicate the gradual downward migration of radioactive cesium in the soils. The L1/10 values in December 2012 were summarized and presented on a map surrounding the Fukushima Dai-ichi NPP, and the average value of L1/10 was 3.01 cm (n = 82) at this time. It was found that radioactive cesium remained within 5 cm of the ground surface at most study sites (71 sites). The sech function can also be used to estimate the downward migration rate V (kg m(-2) y(-1)). The V values in December 2012 (n = 25) were in good agreement with those found by a realistic approach using a diffusion and migration model. Almost all values ranged between 1.7 and 9.6 kg m(-2) y(-1) in this study.

The Fukushima Daiichi nuclear power reactor units that generated large amounts of airborne discharges during the period of March 12–21, 2011 were identified individually by analyzing the combination of measured 134Cs/137Cs depositions on ground surfaces and atmospheric transport and deposition simulations. Because the values of 134Cs/137Cs are different in reactor units owing to fuel burnup differences, the 134Cs/137Cs ratio measured in the environment was used to determine which reactor unit ultimately contaminated a specific area. Atmospheric dispersion model simulations were used for predicting specific areas contaminated by each dominant release. Finally, by comparing the results from both sources, the specific reactor units that yielded the most dominant atmospheric release quantities could be determined. The major source reactor units were Unit 1 in the afternoon of March 12, 2011, Unit 2 during the period from the late night of March 14 to the morning of March 15, 2011. These results corresponded to those assumed in our previous source term estimation studies. Furthermore, new findings suggested that the major source reactors from the evening of March 15, 2011 were Units 2 and 3 and that the dominant source reactor on March 20, 2011 temporally changed from Unit 3 to Unit 2.

Spatial distributions and temporal changes of radioactive fallout released by the Fukushima Dai-ichi Nuclear Power Plant accident have been investigated by two campaigns with three measurement schedules. The inventories (activities per unit area) of the radionuclides deposited onto ground soil were measured using portable gamma-ray spectrometers at nearly 1000 locations (at most) per measurement campaign. Distribution maps of the inventories of (134)Cs, (137)Cs, and (110m)Ag as of March, September, and December 2012 were constructed. No apparent temporal change of the radionuclide inventories was observed from March to December 2012. Weathering effects (e.g., horizontal mobility) were not noticeable during this period. Spatial dependence in the ratios of (134)Cs/(137)Cs and (110m)Ag/(137)Cs were observed in the Tohoku and Kanto regions. The detailed maps of (134)Cs and (137)Cs as of September 2012 and December 2012 were constructed using the relationship between the air dose rate and the inventory.

No abstract