Forecasting soil erosion and sediment yields during flash floods: The disastrous case of Mandra, Greece, 2017

Abstract

Flash floods, among the most destructive natural hazards, are commonly studied as to their catastrophic power in terms of fatalities, infrastructure damages and economic losses. A devastating aftermath of flash floods, which has not received much‐deserved attention in the literature, is the sizeable and permanent soil loss due to soil erosion and sediment yields. This study aims at forecasting soil erosion and sediment yields due to the disastrous storm event that occurred in Mandra town (western Attica, Greece) on 15 November 2017. Gridded hydrometeorological forecasts were conducted at 5‐min and 1‐h time steps by means of the chemical hydrological atmospheric ocean wave system (CHAOS). The forecasts of soil erosion and sediment yields were achieved by a high‐resolution geographic information system (GIS) application of the modified universal soil loss equation (MUSLE) on the basis of the forecasted surface runoff hydrographs. The entire event lasted from the afternoon of 14 November until noon of 15 November, but the bulk of the disastrous storm occurred in the morning of 15 November 2017, flooding two torrent basins. As a result of the extreme flash flood, 2195 tons and 1435 tons of sediment were forecasted to be detached from the body of the soil's surface and transported to the stream networks of the Soures and Agia Aikaterini Torrent basins which cross Mandra. Soil erosion maps were constructed for every hour and the spatial and temporal evolution of soil erosion was shown throughout the event. This study provides concrete insights on the erosion‐prone areas of the study basins, which can inform actions against erosion.

Monthly runoff prediction based on a coupled VMD-SSA-BiLSTM model

“…With the rapid advancement of artificial intelligence technology, numerous deep learning algorithms have emerged, and comprehensive forecasting models based on intelligent methods and numerical weather prediction have been proposed. These models involve various optimization algorithms such as Chaos Optimization Algorithm 1 , bald eagle search optimization algorithm 2 , Particle Swarm Optimization (PSO) 3 , and artificial neural network models 4 , 5 , which have deepened their intersectionality with hydrology.…”