Search

Recently, deep learning (DL) models, especially those based on long short-term memory (LSTM), have demonstrated their superior ability in resolving sequential data problems. This study investigated the performance of six models that belong to the supervised learning category to evaluate the performance of DL models in terms of streamflow forecasting. They include a feed-forward neural network (FFNN), a convolutional neural network (CNN), and four LSTM-based models. Two standard models with just one hidden layer—LSTM and gated recurrent unit (GRU)—are used against two more complex models—the stacked LSTM (StackedLSTM) model and the Bidirectional LSTM (BiLSTM) model. The Red River basin—the largest river basin in the north of Vietnam—was adopted as a case study because of its geographic relevance since Hanoi city—the capital of Vietnam—is located downstream of the Red River. Besides, the input data of these models are the observed data at seven hydrological stations on the three main river branches of the Red River system. This study indicates that the four LSTM-based models exhibited considerably better performance and maintained stability than the FFNN and CNN models. However, the complexity of the StackedLSTM and BiLSTM models is not accompanied by performance improvement because the results of the comparison illustrate that their respective performance is not higher than the two standard models—LSTM and GRU. The findings of this study present that LSTM-based models can reach impressive forecasts even in the presence of upstream dams and reservoirs. For the streamflow-forecasting problem, the LSTM and GRU models with a simple architecture (one hidden layer) are sufficient to produce highly reliable forecasts while minimizing the computation time.

Climate change currently affects the resilience and aquatic ecosystem. Climate change alters rainfall patterns which have a great impact on river flow. Annual flooding is an important hydrological characteristic of the Mekong River Basin (MRB) and it drives the high productivity of the ecosystem and biodiversity in the Tonle Sap floodplain and the Mekong Delta. This study aims to assess the impacts of climate change on river flow in the MRB and flood inundation in the Lower Mekong Basin (LMB). The changing impacts were assessed by a two-dimensional rainfall-runoff and inundation model (RRI model). The present climate (1979–2003) and future projected climate (2075–2099) datasets from MRI-AGCM3.2H and MRI-AGCM3.2S models were applied with a linear scaling bias correction method before input into the RRI model. The results of climate change suggested that flood magnitude in the LMB will be severer than the present climate by the end of the twenty-first century. The increment of precipitation between 6.6 and 14.2% could lead to increase extreme flow (Q5) 13–30%, peak inundation area 19–43%, and peak inundation volume 24–55% in the LMB for ranging of Representative Concentration Pathways (RCP) and sea surface temperature (SST) scenarios while there is no significant change on peak flood timing.

Human activity is a major source of high-frequency seismic noise. Long-term ambient seismic noise levels and their influencing factors are investigated. The diurnal seismic noise level in 5–15 Hz display high correlation with human activities including traffic and industrial operations that are related to economic conditions. The temporal noise-level variations are consistent among three components. Analysis with seismic noises in three consecutive months of each year enables us to estimate the noise levels without seasonal effects. The daytime seismic noise-level changes in major cities of 11 countries are assessed using the 3 month records for decades. The annual seismic noise levels present strong correlations with gross domestic product (GDP), particularly with manufacturing and industrial GDP. The seismic noise levels increase quickly with GDP in low-GDP regions but slowly in high-GDP regions. This is because high-GDP regions already have large volumes of existing noise-inducing sources and because added sources contribute weakly. The seismic noise levels increased by 14%–111% for 5–23 yr depending on the economic conditions. The correlation between ambient seismic noise level and economy growth is a global feature. The high-frequency noise level may be a proxy to present the economic condition. Economic growth affects the Earth environment in a wide range of aspects.

Debris flow triggered by shallow landslides in hillslope catchments is a main geological phenomenon driving landscape changes, and represents an important natural hazard. Numerous studies have assessed sediment transport and deposition by debris flows in hillslope catchments. Thus, the objective of this study is a development of two-dimensional debris flow model to estimate sediment transport and deposition in hillslope catchments. To simulate debris flow, we implemented a vertically integrated shallow-water governing equation based on the Voellmy rheological model and a simple entrainment model. In addition, we applied a quadtree grid structure to support adaptive mesh refinement, where the mesh for the simulation was automatically generated as the debris flow proceeded. Finally, a well-balanced numerical scheme for wet–dry transition treatment was implemented and implicit discretization of the general source terms, including the rheological term, was included for numerical stability. The developed model was verified based on a debris flow triggered by the 2011 Mt. Umyeon landslides in the Republic of Korea. Sediment transport was successfully generated and the sediment deposition area generally matched the field survey data well. Overall, the simulated sediment volume was in good agreement with the survey results, with an error below 1%.

The study aimed to assess the hydrological impacts of the basin development scenarios—which are BS (baseline scenario), DFS (definite future scenario), and FFS (foreseeable future scenario)—by the Mekong River Commission (MRC) on the hydrological regime between the Mekong River and Tonle Sap Lake. The proposed CAESAR-LISFLOOD system was applied to analyze flood inundation patterns of the Tonle Sap Lake using satellite images, as well as to compare the simulated inundation maps. The flow duration variability was also analyzed at the outlet point of the Mekong River basin based on the analysis results of CAESAR-LISFLOOD, according to the basin development. The results show that the CAESAR-LISFLOOD model produces acceptable flooding areas (8607 km2) compared with the satellite observations (9615 km2), and the flood inundation areas decrease gradually from 3.69 (103 km2) in the BS scenario to 3.54 (103 km2) in the FFS scenario. However, it was found that the annual average flow duration variability scenario at the Kratie point, presented in the MRC, does not significantly affect the flow duration downstream from the Kratie point.

Spatiotemporal precipitation data is one of the essential components in modeling hydrological problems. Although the estimation of these data has achieved remarkable accuracy owning to the recent advances in remote-sensing technology, gaps remain between satellite-based precipitation and observed data due to the dependence of precipitation on the spatiotemporal distribution and the specific characteristics of the area. This paper presents an efficient approach based on a combination of the convolutional neural network and the autoencoder architecture, called the convolutional autoencoder (ConvAE) neural network, to correct the pixel-by-pixel bias for satellite-based products. The two daily gridded precipitation datasets with a spatial resolution of 0.25° employed are Asian Precipitation-Highly Resolved Observational Data Integration towards Evaluation (APHRODITE) as the observed data and Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks-Climate Data Record (PERSIANN-CDR) as the satellite-based data. Furthermore, the Mekong River basin was selected as a case study, because it is one of the largest river basins, spanning six countries, most of which are developing countries. In addition to the ConvAE model, another bias correction method based on the standard deviation method was also introduced. The performance of the bias correction methods was evaluated in terms of the probability distribution, temporal correlation, and spatial correlation of precipitation. Compared with the standard deviation method, the ConvAE model demonstrated superior and stable performance in most comparisons conducted. Additionally, the ConvAE model also exhibited impressive performance in capturing extreme rainfall events, distribution trends, and described spatial relationships between adjacent grid cells well. The findings of this study highlight the potential of the ConvAE model to resolve the precipitation bias correction problem. Thus, the ConvAE model could be applied to other satellite-based products, higher-resolution precipitation data, or other issues related to gridded data.

Physically based approaches for the regional assessment of slope stability using DEM topography usually consist of one-dimensional descriptions and include many simplifying assumptions with respect to more realistic, three-dimensional analyses. We investigated a new application of the well-known, publicly available software TRIGRS (Transient Rainfall Infiltration and Grid-Based Regional Slope-Stability Analysis) in combination with Scoops3D, to analyze three-dimensional slope stability throughout a digital landscape in a time-dependent fashion, typically non implemented in three-dimensional models. TRIGRS was used to simulate the dynamic hydraulic conditions within the slopes induced by a rainstorm. Scoops3D then used the resulting pore water pressure for three-dimensional stability assessment. We applied this approach to the July 2011 landslide event in Mt. Umyeon, South Korea, and results were compared with the landslide initiation locations reported for this rainfall event. Soil depth in the study area was described by three different simple models. Stability maps, obtained by the one-dimensional (TRIGRS only) and three-dimensional (TRIGRS and Scoops3D), time-dependent approaches, were compared to observations to assess the timing and locations of unstable sites by means of a synthetic index, previously specifically developed for dealing with point landslide locations. We highlight the performance of the three-dimensional approach with respect to the one-dimensional method represented by TRIGRS alone, and the consistency of the time-dependence of the results obtained using the combined approach with observations.

An accurate landslide-susceptibility assessment is fundamental for preventing landslides and minimizing damage. In this study, a new time-variant slope-stability (TiVaSS) model for landslide prediction is developed. A three-dimensional (3D) subsurface flow model is coupled with the infinite slope-stability model to consider the effect of horizontal water movement in the subsurface. A 3D Richards' equation is solved numerically for the subsurface flow. To overcome the massive computational requirements of the 3D subsurface flow module, partially implicit temporal discretization and the simplification of first-order spatial discretization are proposed and applied in TiVaSS. A graphical user interface and two-dimensional data visualization are supported in TiVaSS. The model is applied to a 2011źMt. Umyeon landslide in the Republic of Korea, and its overall performance is satisfactory. A new time-variant slope-stability model for predicting rainfall-induced landslides is developed.The three-dimensional subsurface unsaturated flow and an infinite slope-stability model are combined.A graphical user interface including input and output visualization is supported.

Many researchers have evaluated the influence of vegetation cover on slope stability. However, due to the extensive variety of site conditions and vegetation types, different studies have often provided inconsistent results, especially when evaluating in different regions. Therefore, additional studies need to be conducted to identify the positive impacts of vegetation cover for slope stabilization. This study used the Transient Rainfall Infiltration and Grid-based Regional Slope-stability Model (TRIGRS) to predict the occurrence of landslides in a watershed in Jinbu-Myeon, Pyeongchang-gun, Korea. The influence of vegetation cover was assessed by spatially and temporally comparing the predicted landslides corresponding to multiple trials of cohesion values (which include the role of root cohesion) and real observed landslide scars to back-calculate the contribution of vegetation cover to slope stabilization. The lower bound of cohesion was defined based on the fact that there are no unstable cells in the raster stability map at initial conditions, and the modified success rate was used to evaluate the model performance. In the next step, the most reliable value representing the contribution of vegetation cover in the study area was applied for landslide assessment. The analyzed results showed that the role of vegetation cover could be replaced by increasing the soil cohesion by 3.8 kPa. Without considering the influence of vegetation cover, a large area of the studied watershed is unconditionally unstable in the initial condition. However, when tree root cohesion is taken into account, the model produces more realistic results with about 76.7% of observed unstable cells and 78.6% of observed stable cells being well predicted.

Flood impacts threaten the socioeconomic conditions of peoples’ lives in the Mekong River Basin. In this study, the rainfall-runoff-inundation (RRI) model, capable of simulating rainfall ru...

In Korea, landslide damage areas have increased significantly from the 1990s to the 2000s owing to increases in rainfall intensity and the number of rainy days in addition to indiscriminate land development. This study was conducted to predict shallow landslide-prone areas by using the transient rainfall infiltration and grid-based regional slope-stability model (TRIGRS). A landslide event that occurred on July 27, 2011, in Mt. Umyeon, Seoul, was modeled, and the stability results were compared with 140 observed landslide points in both time and location to evaluate the performance of TRIGRS. Simulated outcomes from five different raster cells of 5, 10, 15, 20, and 25 m with the same background information were compared to identify the ultimate grid scale. The results revealed that in location estimation, smaller grid size resulted in more accurate results. However, the 10 and 15 m grid sizes gave better results in the timing assessment When the best grid size was considered TRIGRS overestimated t...

This study compares the performance of transient rainfall infiltration and grid-based regional slope stability (TRIGRS) model and time-variant slope stability (TiVaSS) model in the prediction of rainfall-induced shallow landslides. TRIGRS employs one-dimensional (1-D) subsurface flow to simulate the infiltration rate, whereas a three-dimensional (3-D) model is utilized in TiVaSS. The former has been widely used in landslide modeling, while the latter was developed only recently. Both programs are used for the spatiotemporal prediction of shallow landslides caused by rainfall. This study uses the July 2011 landslide event that occurred in Mt. Umyeon, Seoul, Korea, for validation. The performance of the two programs is evaluated by comparison with data of the actual landslides in both location and timing by using a landslide ratio for each factor of safety class ( $${\text{LR}}_{\text{class}}$$ index), which was developed for addressing point-like landslide locations. Moreover, the influence of surface flow on landslide initiation is assessed. The results show that the shallow landslides predicted by the two models are highly consistent with those of the observed sliding sites, although the performance of TiVaSS is slightly better. Overland flow affects the buildup of the pressure head and reduces the slope stability, although this influence was not significant in this case. A slight increase in the predicted unstable area from 19.30 to 19.93% was recorded when the overland flow was considered. It is concluded that both models are suitable for application in the study area. However, although it is a well-established model requiring less input data and shorter run times, TRIGRS produces less accurate results.