Flood Series Research Articles

Possibilities of Using Regional Index-Flood Method and Annual Maximum and Partial Duration Series: A Case Study of Susurluk River Basin, Turkey

Floods are defined as disasters that occur as a result of a natural formation and cause negative effects on society life and economy. Floods that cause the greatest loss of life and property after earthquakes among the natural disasters experienced in our country, unplanned urbanization caused by the increasing population, uncontrolled settlements in river beds, and changing climate have increased its impact and frequency of occurrence over time. Therefore, it is important to estimate the magnitude of floods and the frequency of their occurrence in the most accurate way. This research aims to perform at-site and regional frequency analysis with L-moment methods in the Susurluk Basin. For regional frequency analysis, annual maximum and partial duration flood series were obtained using the data from 22 flow observation stations selected from the basin. In the first stage, the stations in the region were considered as a whole and the stations were found discordant with each other. Hierarchical cluster analysis was performed with the help of various physiological-hydrological characteristics of the stations located in the basin with geographic information systems, and the basin was divided into 2 regions, but homogeneity was still not achieved. Therefore, the basin is divided into 3 regions according to the dendrogram. There was no discordancy in all three regions and homogeneity was achieved. After achieving homogeneity, the most suitable frequency distribution was determined for each region. In addition, L-moments and L-moment ratios, probability distributions, and quantile discharges of the annual maximum and partial duration flood series were estimated for at-site frequency analysis. As a result, in regional flood frequency analysis, generalized logistic distribution provided the best fit in the annual maximum series, while Generalized Pareto and Pearson type 3 provided the best-fit distributions for the first and second regions in the partial duration series, respectively, in addition, generalized normal provided the best fit in the third region. It was concluded that it would be useful to use partial duration flood series as well as annual maximum series in the flood frequency analysis of the Susurluk basin.

Flood scenario spatio-temporal mapping via hydrological and hydrodynamic modelling and a remote sensing dataset: A case study of the Basento river (Southern Italy)

Today, hydrological and hydraulic modelling are essential tools for flood risk management, although these models are still affected by elements of uncertainty that needs to be reduced by optimizing their results. The present research aims to implement an operational mechanism on the Basento river basin in Southern Italy based on the cascading use of a physically based concentrated-parameter hydrological model for the estimation of flood hydrographs, and a two-dimensional hydraulic model for flood mapping. The calibration of the hydrological model uses physical information to reduce the initial range of the set parameter values, and an automated optimisation procedure based on a genetic algorithm to find optimal values of the model parameters by comparing simulated and observed data for the 2013 flood event. To calibrate the hydraulic model, a series of flood maps extracted from multi-temporal SAR images was used. In addition, validation of the hydrological and hydraulic models was carried out on March 2011 flood event. The results show the reliability of the models during both calibration and validation, with the hydrological model achieving a Nash-Sutcliffe Efficiency coefficient between 0.86 and 0.91, and the hydraulic model leading to results with an accuracy close to 70 %. Considering the significance of the results, the developed modelling chain was used to simulate future event scenarios for risk management assessment and could operate as an early warning system.

Goodness-of-fit, identifiability and extrapolation: Can the Two-Component Extreme Value distribution be used in at-site flood frequency analysis?

When fitting three-parameter flood frequency models to annual maximum (AM) flood series, the lack-of-fit can be mitigated by censoring potentially influential low flows (PILFs). An alternative and less-studied approach is to apply mixture probability models with four or more parameters, which trade off greater flexibility to fit AM series against the need to deal with degeneracy which arises when there is insufficient information to identify mixture components. However, the issue of degeneracy and the lack of a robust inference framework present a significant barrier to adoption. This study investigated the potential of the most parsimonious mixture model, the four-parameter Two-Component Extreme Value (TCEV) model. A Bayesian framework using Markov chain Monte Carlo sampling was developed to robustly characterize parameter uncertainty even in the presence of degeneracy. Two new posterior diagnostics based on the strength of the TCEV components were developed to aid identification of degeneracy. Armed with a robust inference framework, the study evaluated the potential of TCEV using a case study based on 31 catchments in eastern Australia with records exceeding 70 years. The evaluation used short to long records to compare TCEV and Log Pearson III fit and extrapolative uncertainty. The Log Pearson approach (LP3-PILF) censors PILFs following the approach described in Australian Rainfall and Runoff. With respect to goodness-of-fit, we found in most cases that TCEV fitted AM flood peaks well without the need to censor or stratify data, and overall, no clear difference emerged between TCEV and LP3-PILF. However, and contrary to expectation, TCEV produced high flow quantile confidence intervals consistently narrower than LP3-PILF even in the presence of degeneracy. While more case studies in different regions are needed to confirm the potential of TCEV, this study is a reminder that goodness-of-fit is a necessary but not sufficient criterion for selecting the probability model that best represents flood frequency.

ENSEMBLE MODELS FOR FORECASTING FLOODS IN GREAT BRITAIN BASED ON SOLAR ACTIVITY

This study aims to uncover hidden dependencies between streams of solar particles emanating from the Sun and flood events caused by precipitation in the UK. The uniqueness of this work lies in applying solar activity data as input parameters for machine learning and correlating them with real data on precipitation intensity and floods in the UK, which serve as output data. The analysis covered 20 significant floods from October 2001 to December 2019. The analysis was performed on a daily basis, taking into account the time shift between solar emissions and their impact on terrestrial weather conditions, which ranged from 0 to 9 days. The study employs correlation analysis to determine the degree of interrelation between time series of solar activity and floods, laying the foundation for a deeper understanding of possible cause-and-effect relationships. Subsequently, using predictive modeling methods, including decision tree algorithms and ensemble classification models, the potential connection between changes in solar activity and terrestrial floods was explored. An ensemble of models was created using the hard voting method, which takes into account various indicators such as proton density, differential proton flux, and ion temperature, with a time shift of up to 9 days. The analysis showed that these key parameters could effectively predict the occurrence of floods with up to 92% accuracy for the specified period. The results underscore the importance of integrating space weather and solar activity into the models for predicting terrestrial weather conditions, particularly in the context of floods in England. This not only allows for a better understanding of the causes of natural disasters but also enhances the effectiveness of emergency planning and reduces potential damages from them. Thus, this work makes a significant contribution to the field of space weather research and its impact on Earth, particularly in the areas of hydrometeorology and flood risk management.

Integration of RS and GIS in assessing flood risk: A Case Study on Istanbul - Esenyurt

Floods, exacerbated by escalating urbanization, pose significant threats to life and property globally. Over the past decade, the Esenyurt district in Istanbul has witnessed a series of floods, highlighting existing flood risks. Rapid population growth in this area and dense urbanization caused by intensive construction increase flood risks. Given these factors, the study focuses on examining the historical impact of urbanization on flood risks, considering spatial and temporal changes. Landsat-8 satellite data, specifically examining NDVI, NDBI, and BU, was employed to detect building imprints and reveal their historical backgrounds for temporal risk calculations. The analysis showed a sudden increase in urbanization rates in 2016, 2017, and 2021. In the flood risk calculations, 2014 data for a return period of 100 years were used and flood inundation depth, economic damages, affected population and depth-damage function were taken into consideration. The results indicate that from 2014 to 2022, increasing urbanization led to a 32.9% increase in the population affected by floods, a 22.3% rise in potential economic damage, and a 13.6% increase in total flood risk. The relationship between flood risks, contemporary urbanization, and its economic dimensions has been evaluated to reduce risks and achieve sustainable cities.

What controls the tail behaviour of flood series: rainfall or runoff generation?

Abstract. Many observed time series of precipitation and streamflow show heavy-tail behaviour. For heavy-tailed distributions, the occurrence of extreme events has a higher probability than for distributions with an exponentially receding tail. If we neglect heavy-tail behaviour we might underestimate the magnitude of rarely observed, high-impact events. Robust estimation of upper-tail behaviour is often hindered by the limited length of observational records. Using long time series and a better understanding of the relevant process controls can help with achieving more robust tail estimations. Here, a simulation-based approach is used to analyse the effect of precipitation and runoff generation characteristics on the upper tail of flood peak distributions. Long, synthetic precipitation time series with different tail behaviour are produced by a stochastic weather generator. These are used to force a conceptual rainfall–runoff model. In addition, catchment characteristics linked to a threshold process in the runoff generation are varied between model runs. We characterize the upper-tail behaviour of the simulated precipitation and discharge time series with the shape parameter of the generalized extreme value (GEV) distribution. Our analysis shows that runoff generation can strongly modulate the tail behaviour of flood peak distributions. In particular, threshold processes in the runoff generation lead to heavier tails. Beyond a certain return period, the influence of catchment processes decreases and the tail of the rainfall distribution asymptotically governs the tail of the flood peak distribution. Beyond which return period this is the case depends on the catchment storage in relation to the mean annual rainfall amount.

Low-Flow Identification in Flood Frequency Analysis: A Case Study for Eastern Australia

Design flood estimation is an essential step in many water engineering design tasks such as the planning and design of infrastructure to reduce flood damage. Flood frequency analysis (FFA) is widely used in estimating design floods when the at-site flood data length is adequate. One of the problems in FFA with an annual maxima (AM) modeling approach is deciding how to handle smaller discharge values (outliers) in the selected AM flood series at a given station. The objective of this paper is to explore how the practice of censoring (which involves adjusting for smaller discharge values in FFA) affects flood quantile estimates in FFA. In this regard, two commonly used probability distributions, log-Pearson type 3 (LP3) and generalized extreme value distribution (GEV), are used. The multiple Grubbs and Beck (MGB) test is used to identify low-flow outliers in the selected AM flood series at 582 Australian stream gauging stations. It is found that censoring is required for 71% of the selected stations in using the MGB test with the LP3 distribution. The differences in flood quantile estimates between LP3 (with MGB test and censoring) and GEV distribution (without censoring) increase as the return period reduces. A modest correlation is found (for South Australian catchments) between censoring and the selected catchment characteristics (correlation coefficient: 0.43), with statistically significant associations for the mean annual rainfall and catchment shape factor. The findings of this study will be useful to practicing hydrologists in Australia and other countries to estimate design floods using AM flood data by FFA. Moreover, it may assist in updating Australian Rainfall and Runoff (national guide).

Factors Influencing the Rheology of Methane Foam for Gas Mobility Control in High-Temperature, Proppant-Fractured Reservoirs

Gas-enhanced oil recovery (EOR) through huff-n-puff (HnP) is an important method of recovering oil from fracture-stimulated reservoirs. HnP productivity is hampered by fracture channeling, leading to early gas breakthroughs and gas losses. To mitigate these issues, foam-generating surfactants have been developed as a method of reducing injected gas phase mobility and increasing oil recovery. This work investigates foam generation and propagation by a proprietary surfactant blend in high-temperature, high-pressure, high-permeability, and high-shear conditions that simulate the environment of a proppant-packed fracture. Bulk foam tests confirmed the aqueous stability and foaming viability of the surfactant at the proposed conditions. Through several series of floods co-injecting methane gas and the surfactant solution through a proppant pack at residual oil saturation, the effects of several injection parameters on apparent foam viscosity were investigated. The foam exhibited an exceptionally high transition foam quality (>95%) and strong shear-thinning behavior. The foam viscosity also linearly decreased with increasing pressure. Another flood series conducted in an oil-free proppant pack showed that swelling of residual oil had no effect on the apparent foam viscosity and was not the reason for the inversely linear pressure dependency. An additional flood series with nitrogen as the injection gas was completed to see if the hydrophobic attraction between the methane and surfactant tail was responsible for the observed pressure trend, but the trend persisted even with nitrogen. In a previous study, the dependence of foam viscosity on pressure was found to be much weaker with a different foaming surfactant under similar conditions. Thus, a better understanding of this important phenomenon requires additional tests with a focus on the effect of pressure on interfacial surfactant adsorption.

The effects of vegetative feedbacks on flood shape, sediment transport, and geomorphic change in a dryland river: Moenkopi Wash, AZ

Since the 1950s, Moenkopi Wash, in Arizona, United States, has been transformed from a relatively wide river at low flow, to a narrow, heavily vegetated river that is less than half of its former width. We analyzed a ~95-years-long instantaneous-discharge record, an extensive sediment-transport record, oblique and aerial photographs, historical channel surveys, and historical stage-discharge rating relations to determine the primary mechanisms responsible for this transformation.Frequent, large floods dominated the early part of the discharge record between 1926 and 1940. A dramatic ~30 % decrease in annual mean discharge, and ~27 % decrease in the mean of the partial-duration flood series occurred in 1941. This decline did not result in widespread channel change but provided an opportunity for vegetation to establish along the channel margins. Widespread channel narrowing began after a second decline in mean and peak discharge in 1956 at which time the channel banks became heavily vegetated. Between 1952 and 2019, the river channel narrowed by 57–59 %. Approximately 2–4 m of bed aggradation occurred at most study sites. As the channel narrowed and became more vegetated, large floods maintained the narrower channel width but did not cause rewidening. Suspended-sediment transport data illustrate the sediment trapping effects of vegetation that lead to channel narrowing; suspended-sand concentrations decline, thereby indicating sediment deposition, as vegetation becomes progressively inundated during floods. Furthermore, dense channel-margin and floodplain vegetation provide increased roughness, resulting in flood-peak attenuation and alteration of hydrograph shape. Vegetation expansion causes positive feedbacks whereby sediment deposition during floods is exacerbated by the roughness and sediment-trapping effects of vegetation leading to further narrowing. These positive feedbacks have reduced sediment delivery to the Little Colorado River downstream. Channel widening is not likely to occur unless there are very large floods that exceed the erosional threshold of the channel-margin vegetation, or unless large-scale vegetation removal efforts are undertaken.

A practice-oriented framework for stationary and nonstationary flood frequency analysis

In flood frequency analysis (FFA), choices of distribution and methods can hinder the reproducibility of results. Besides, changes in climate, land use/cover, and water management can induce nonstationarity. Frameworks to select between stationary FFA (S-FFA) and nonstationary FFA (NS-FFA) are lacking, and NS-FFA tools are limited. Therefore, this paper introduces a systematic and software-supported framework enabling repeatable workflows for both S-FFA and NS-FFA. The framework has three modules to a) process flood series for exploratory data analysis (EDA) and NS-FFA model determination (if needed), b) select the S-FFA or NS-FFA approach underpinned by the EDA, and c) perform FFA including model determination, parameter estimation, uncertainty quantification, and model performance assessment. The framework incorporates various distributions, methods, and metrics, and recent advancements in NS-FFA for model determination and uncertainty quantification and allows for the modeller's intervention while ensuring reproducibility. The software is freely available to the public.

Historical floods in the southeastern Iberian Peninsula since the 16th century: Trends and regional analysis of extreme flood events

Studies related to historical floods have generated much knowledge about flood patterns and frequencies in recent years. Extending the series of floods to a longer range of time allows society to clarify what has happened in the past, and thus know what may happen in the future. From the descriptions of primary and secondary sources, a classification of historical floods has been carried out following a method specifically adapted for the study area. The southeastern Iberian Peninsula is the last region to compile historical floods in the Mediterranean Iberian Peninsula. Therefore, the main objective of the present study is to extend the flood series of historical events for four river basins: the Almanzora, Antas, Aguas and Andarax. First, flood events are classified according to their descriptions, generating a dataset of all flood events from 1500 to the present day for each basin. Second, we analyze the trends in the historical series, identifying four trend periods linked with the availability of records: 1500–1850, 1850–1900, 1900–1955 and 1955–2000. Notably, the trend in recent decades has broken with normality because of the large number of lower magnitude events. Lastly, we compare our dataset with seven historical series from different rivers elsewhere in the Iberian Peninsula. The results show a high correlation with all southeastern basins and less with the northern and Atlantic basins. For the purposes of the analysis, we considered only the extraordinary and catastrophic floods in all basins, that is, the floods of magnitude 3 or 4 (M3 or M4), and we identified two flood gaps in 1500–1540 and 1660–1720; two flood-poor periods in 1790–1845 and 1955–1970; and five flood-rich periods in 1540–1560, 1610–1654, 1720–1745, 1860–1891 and 1970–1990 in relation to the four river basins in the study area and the other seven Iberian basins, the last century is highly biased by the construction of reservoirs in all basins. The analysis of historical floods shows a link between the flood-gap periods and negative phases of NAOi and TSI. From 1970, lower flood magnitudes occurred during periods of highest amounts of information, an increase in tourism areas. Consequently, the last flood-rich period clearly stands out because of the disparity of the data, and the consequences are biased, for example, by surface runoff in urban areas where non-irrigated agricultural areas were traditionally located, resulting in an increase in economic damage.

Mitigating infectious disease risks through non-stationary flood frequency analysis: a case study in Malaysia based on natural disaster reduction strategy.

The occurrence of floods has the potential to escalate the transmission of infectious diseases. To enhance our comprehension of the health impacts of flooding and facilitate effective planning for mitigation strategies, it is necessary to explore the flood risk management. The variability present in hydrological records is an important and neglecting non-stationary patterns in flood data can lead to significant biases in estimating flood quantiles. Consequently, adopting a non-stationary flood frequency analysis appears to be a suitable approach to challenge the assumption of independent and identically distributed observations in the sample. This research employed the generalized extreme value (GEV) distribution to examine annual maximum flood series. To estimate non-stationary models in the flood data, several statistical tests, including the TL-moment method was utilized on the data from ten stream-flow stations in Johor, Malaysia, which revealed that two stations, namely Kahang and Lenggor, exhibited non-stationary behaviour in their annual maximum streamflow. Two non-stationary models efficiently described the data series from these two specific stations, the control of which could reduce outbreak of infectious diseases when used for controlling the development measures of the hydraulic structures. Thus, the application of these models may help prevent biased prediction of flood occurrences leading to lower number of cases infected by disease.

Dynamic variations in thermal regime and surface deformation along the drainage channel for an expanding lake on the Tibetan Plateau

ABSTRACT The outburst of Zonag Lake in 2011 triggered a series of floods in the continuous permafrost region of the hinterland of the Qinghai-Tibet Plateau. This re-distributed the surface water in the basin and caused rapid expansion of the tail lake (Salt Lake). To avoid potential overflow of the expanding Salt Lake, a channel was excavated to drain the lake water into a downstream river. In this study, to investigate the permafrost thermal regime and the surface deformation around the expanding Salt Lake and the channel, in-situ monitoring sections were settled from Salt Lake to the downstream of the channel to obtain the permafrost temperature. Additionally, using small baseline subset interferometric synthetic aperture radar (SBAS-InSAR), the surface deformation around Salt Lake and the channel was measured. The data showed that the ground temperature at the channel was 0.6°C higher than the natural field and the mean subsidence rate around the channel was 1.5 mm/yr higher than that at Salt Lake. These results show that the permafrost temperature in the study area changed considerably with variations in the distance from the lake/channel, and the deformation in the study area was dominated by subsidence.

Different responses of event-based flood to typhoon and non-typhoon rainstorms under land use change in Xixi Basin of southeastern China

There are two major types of rainstorms in the Asia-Pacific region –typhoon induced rainstorm and non-typhoon rainstorm. This study investigated and separated the different flood responses to these two types of rainstorms, based on a scheme that combined the classification of typhoon and non-typhoon rainstorm floods, the simulation from a distributed hydrological model, and a method to forward restore the impact of land use change on flood frequency analysis. For flood events in a typical basin on the southeast coast of China, the flood characteristics in response to typhoon and non-typhoon rainstorms were significantly different in three aspects: (1) The key parameters in the HEC-HMS model were systematically different. The initial loss and wave velocity of the model were apparently larger for the typhoon rainstorm flood than for the non-typhoon rainstorm flood. (2) The impacts of land use change were different, with smaller effect on typhoon induced floods than on non-typhoon floods under the same magnitude of flood peak flow. (3) Classification of two types of floods resulted in more reasonable frequency analysis affected with the land use change because the annual maximum flood series is the mixture of typhoon and non-typhoon rainstorms. Frequency analysis for the forward-restored mixture flood series showed that the land use change with the decrease in the area of forest land and the increase of garden and construction land resulted in the increase of the mean maximum flow from 2266.6 to 2510.19 m3/s, but the values of coefficient of variation were slightly altered, the ratio of skewness coefficient to variation coefficient were unchanged. The results provide new ideas and methods for a deeper understanding of the impact of land use change on flooding in typhoon and non-typhoon co-action regions.

Frequency Analysis of Flood Flow in Markanda Basin of Ghaggar River System in North Western India

Abstract This study aims to estimate the probabilities of occurrence and return periods of peak flood discharges over the Markanda basin of Ghaggar river system in north western India. For this purpose, two most frequently employed probability distribution models, namely Gumbel Extreme Value (GEV) and Log-Pearson Type III (LP-III) have been used to estimate the future flood discharges by means of annual extreme flood series (Qmax) data from 1990 to 2013, available at eight gauge and discharge sites. Two goodness-of-fit tests, i.e., Kolmogorov-Smirnov and Anderson-Darling have been applied to the fitted probability distributions to identify the best-fit model. The Qmax for several return periods, for example 2, 5, 10, 25, 50, 100, and 200 years have been estimated and compared. The return period for the highest Qmax recorded at Jhansa (2670 m3/s) gauge and discharge site has been computed as 9.5 and 8.2 years using the GEV and LP-III distribution model, respectively. The analyses have shown that GEV distribution model has produced the overestimated results in comparison to LP-III distribution model. Also, flood index has been computed for flood regionalization. The flood index values have been found variable at different gauge and discharge sites with an increase in return periods. Finally, the finding of this study will be valuable for water resource engineers in designing the hydraulic structures for the management of recurrent floods.

Flood Incidences as Public Health Challenge in Katsina State, Northern Nigeria

Flood incidences are one of the major environmental hazards that occur yearly in different parts of the world. The year 2022 has witnessed series of catastrophic floods not experienced in recent times in parts of countries such as Afghanistan, Bangladesh, Italy, Pakistan, Niger Republic, Pakistan, South Africa and United State of America. Nigeria like other developing countries in the continent of Africa has also experienced this kind of floods incidences. This paper examines flood incidences as a public health challenge in Katsina state northern Nigeria. Data for the study was generated from a semi-structured questionnaire administered on 100 students. The students are those studying National Diploma in Environmental Health Technology at Hassan Usman Katsina Polytechnic, Katsina, Katsina State. Data on annual flood incidences from 2018 – 2022 was collected from the Katsina State Emergency Management Agency (SEMA) including discussions with the Directors of Information and Search and Rescue. The results have shown that flood incidences within the period under study have become a public health challenge resulting to loss of lives, injuries to victims, destruction of houses and creation of displaced persons among others. These are happening despite the efforts of Katsina State Government to tackle the challenges posed by the floods to public health. It is therefore recommended that on-going flood control projects should continue and more efforts on safeguarding public health from flood incidences be pursued.

Leaky dams augment afforestation to mitigate catchment scale flooding

AbstractDespite calls for large‐scale afforestation to alleviate flooding, the effectiveness of such action remains unclear. Simulations with the SHETRAN hydrological model are therefore carried out for the 335‐km2 Irthing catchment and its 1‐10‐km2 headwater catchments in northwest England to determine: (a) whether forests can reduce flood peak discharges in a large catchment; (b) the proportion of the catchment that requires afforestation to be effective; and (c) the extent to which a combination of afforestation and natural flood management features (leaky dams) improves upon afforestation on its own. Four‐year simulations were run with a range of forest covers and extents of leaky dam installation, the latter modelled as a channel hydraulic resistance. Hydrograph, flood frequency and peak discharge magnitude responses to forest cover simulated (and observed) in the headwater catchments are replicated in simulations at the full scale. Afforestation on its own can reduce the frequency of given flood magnitudes but has a variable impact on individual peak discharge magnitudes. For the Irthing, a 76% forest cover reduces the mean discharge of 20 peaks in a partial duration flood series by 10% relative to the current 21% forest cover but reduces the largest peak by only 2.5%. Accompanying adverse effects include 17.5% reduction in long‐term runoff and loss of agricultural land. By contrast, leaky dams mitigate flood frequencies and peak discharges effectively, over a range of discharge magnitudes, with no reduction of annual runoff. The dam installation required for a 10% mean peak discharge reduction reduces the largest peak by 22%. The flow resistance increases by which the dams are simulated have still to be translated into specific dam designs. Nevertheless, and considering the figures indicatively, dam installation in 20% of the Irthing streams with Strahler orders of 1–3 achieves a 10% peak discharge reduction with only 35% forest cover and 7.5% runoff reduction. The study illustrates the potential for a dense network of leaky dams to augment the impact of afforestation on flood mitigation, especially at the largest discharges, while minimizing adverse impacts on water resources and food security.

Validation of Synthetic Design Hydrographs through 2D hydrodynamic modelling

The procedure for the determination of Synthetic Design Hydrographs (SDHs), proposed in previous works, is validated by comparing the peak discharges obtained by routing a long series of historical floods and the synthetic floods at different stations along a complex river system.At this aim, the 60 km long terminal stretch of the Dora Baltea river (Northern Italy) has been modelled according to fully 2D high resolution hydrodynamic approach. The fluvial branch is of considerable complexity due to a strong contraction induced by the presence of a narrow Roman bridge, which, during the most important flood events, causes the reactivation of a paleochannel and the flooding of a part of the city of Ivrea. The hydraulic model has been calibrated on the basis of the main historical floods. Then, all the historical floods over a period of more than 80 years (1939–2020) and the SDHs derived by the same series have been routed. Historical and synthetic peak discharges at two downstream stations have been then compared in probability plots. The results show that the peak discharge distributions derived by routing the historical floods and the SDHs compare well. This suggests that SDHs construction procedure is reliable and has statistical significance.

Assessment of first-order-moment-based sample reconstruction method for design flood estimation in changing environment

Estimating the design flood under nonstationary conditions is challenging. In this study, a sample reconstruction approach was developed to transform a nonstationary series into a stationary one in a future time window (FTW). In this approach, the first-order moment (EFTW) of an extreme flood series in the FTW was used, and two possible methods of estimating EFTW values in terms of point values and confidence intervals were developed. Three schemes were proposed to analyze the uncertainty of design flood estimation in terms of sample representativeness, uncertainty from EFTW estimation, and both factors, respectively. To investigate the performance of the sample reconstruction approach, synthesis experiments were designed based on the annual peak series of the Little Sugar Creek in the United States. The results showed that the sample reconstruction approach performed well when the high-order moment of the series did not change significantly in the specified FTW. Otherwise, its performance deteriorated. In addition, the uncertainty of design flood estimation caused by sample representativeness was greater than that caused by EFTW estimation.

Time-series generative adversarial networks for flood forecasting

Reliable flood forecasting serves as one of the fundamental tasks in flood management. However, the forecast performance of modern data-driven models depends heavily on the quantity and quality of training data. The lack of field data of flood events highly underpins the development of machine learning (ML) flood forecasting techniques. With consideration of the rareness of flood events and the high dimensionality of flood time series, two latest variants of generative adversarial networks (GANs), Time-series Generative Adversarial Network (TimeGAN) and Real-world Time Series Generative Adversarial Network (RTSGAN), were applied to generate synthetic flood timeseries. In the example analysis of the Xijiang River Basin in southern China, results show that time-series GANs can accurately and efficiently mimic the spatiotemporal correlations between flood series from multiple sites, and RTSGAN outperforms TimeGAN especially when the lengths of time series are long. Additionally, gradient boosting regression tree (GBRT), long-short term memory (LSTM) and quantile regression method integrated LSTM networks (QRLSTM) flood forecasting models were trained with real sequences and an extra amount of synthetic sequences. It shows that the expansion of training datasets in this way is promising in reducing prediction errors of classical machine learning models. In the 3-day-ahead flood forecast, the introduction of synthetic training datasets has little positive effect on both GBRT and LSTM. In 24-hour-ahead flood forecast, as the lead period increases, GBRT-RTSGAN outperforms GBRT evidenced by a 6.7% increase in Nash-Sutcliffe efficiency (NSE) metric, a 56.5% reduction in average absolute relative error (AARE) metric and a 33.1% reduction in root-mean square error (RMSE) metric. For interval forecast, the use of synthetic training datasets helps QRLSTM obtain lower prediction interval normalized averaged width (PINAW), however, at the cost of prediction interval coverage probability (PICP).