In this study, an image processing method is proposed for classifying the sea ice condition as open water, thin ice, first-year ice, and old ice using the downward-looking camera images obtained from the ship-borne optical camera in the Arctic Ocean during late summer and early autumn. Additionally, we discuss the versatility and validity of this method in the major regions of the Arctic Ocean for observing the sea ice condition in the future. Open water is identified based on the average level of the red channel, whereas the thin ice and thick (first-year or old) ice types are identified based on the average hue. Further, the differences between the red and blue levels and between the green and blue levels can be used to distinguish between the first-year ice and old ice. This method exhibits a high concordance rate for old ice and a low concordance rate for first-year ice and thin ice.
The historical flood that occurred in western Japan in July 2018 claimed over two hundred lives. The river embankment breach caused severe flood damages particularly in Okayama, leading to complete destruction of 2100 houses. According to the engineers who appeared in the news, the flood was likely triggered by fluvial backwaters between Takahashi and Oda River. To understand whether other mechanisms might have been related to this flood event, we investigated the interconnection between tides in Seto Inland Sea and water levels in Takahashi River. This concern originated from two facts: (1) the river mouth of Takahashi River has been expanded because of the extensive river/port development, and (2) the inner part of Seto Inland Sea is subject to high tides due to the two tidal waves from both sides of the sea. We performed a numerical analysis to simulate the strong turbulence induced by the collision of the flood tide and river flow around a submerged weir constructed near the river mouth. The computational result demonstrates that a moderate tidal flow triggers the phase shift of the hydraulic regime from supercritical to subcritical flow, leading to substantial increase in river water levels. The survey we conducted one month after the disaster confirmed that the flood watermark found adjacent to the weir was not remarkably different between the leeward and windward of the structure, which indicates that a supercritical flow did not occur during the flood peak. A hydrological analysis reveals that the timing of peak water levels during the 2018 flood corresponded to the phase of the flood tide. Thus, it is likely that the upstream water level excessively rose because the flood tide seemed to disturb the river outflow. The study findings suggest that the phase of estuary tides should also be considered for river flood risk management.