Development of a Bayesian-copula-based frequency analysis method for hydrological risk assessment – The Naryn River in Central Asia

Liu, Y.R.; Li, Y.P.; Ma, Yuan; Jia, Qimeng; Su, Y.Y.

doi:10.1016/j.jhydrol.2019.124349

Cited by 62 publications

(23 citation statements)

References 51 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…BNN is conducted to model the relationship between selected inputs and monthly runoff, in which, calculating the posterior probability distribution of neural network's parameters w (weights and biases) is the main task, as below (Khan and Coulibaly, 2006):

P (()|, w, D) = \frac{P (()|, D, w) P ((), w)}{P ((), D)}

where P ( w|D ) is the posterior probability distribution of w ; P ( D ) is a normalization factor, representing the marginal distribution of observed data D ; P ( D|w ) is the data set likelihood function; P ( w ) is the prior distribution of w . The prior distribution is assumed to belong to Gaussian distribution, due to analytical simplicity, and the widely use of this assumption for Bayesian framework in previous studies (Ren et al ., 2018; Liu et al ., 2020). Then, the posterior distribution of w can be obtained as (Khan and Coulibaly, 2006):

P (()|, w, D) = \frac{1}{Z_{S} ((),, α, β)} \exp ((), - β E_{D} - α E_{w})

Z_{s} ((),, α, β) = \int \exp ((), - β E_{D} - α E_{w}) italicdW

E_{D} ((), w) = \frac{1}{2} \sum_{i = 1}^{N} {(y_{i} (x_{i}, w) - t_{i})}^{2}

E_{w} ((), w) = \frac{1}{2} \sum_{i}^{W} w_{i}^{2}

where W is the total number of weights and biases in the network; N is the sample size of the training set; y ( x ; w ) represents a network function governing the mean of the distribution, and x is a given input vector corresponding to the observed data t ; α and β are hyperparameters that control other parameters, which can be optimized as part of the training process.…”

Section: Development Of Mbha Methodsmentioning

confidence: 99%

An integrated multi‐GCMs Bayesian‐neural‐network hydrological analysis method for quantifying climate change impact on runoff of the Amu Darya River basin

Liu

et al. 2021

Intl Journal of Climatology

View full text Add to dashboard Cite

As one of the most pressing issues in the world, climate change has already caused evident impacts on natural and human systems (e.g., hydrological cycle, eco‐environment and socio‐economy) in recent decades. In this study, an integrated multi‐GCMs Bayesian‐neural‐network hydrological analysis (MBHA) method is developed for quantifying climate change impacts on runoff. MBHA incorporates multiple global climate models (multi‐GCMs), hydrological model (HBV‐light), and Bayesian neural network (BNN) within a general framework. MBHA can provide the reliable prediction for runoff as well as reflect the impact of climate change on data scarcity catchments. MBHA is applied to the Amu Darya River basin in Central Asia. Climate data are derived from multiple GCMs (i.e., GFDL‐ESM2G, HadGEM2‐AO and NorESM1‐M) under RCP4.5 and RCP8.5. Several findings can be summarized: (1) during 2021–2100, both precipitation and temperature would increase, with more precipitation falling as rain instead of snow; (2) by 2100, glacier areas are predicted to reduce by 62.3% (RCP4.5) and 71.9% (RCP8.5); (3) under RCP8.5, monthly runoff would increase by 11.2% in 2021–2060 and reduce by 5.0% in 2061–2100; this is because the glaciers would rapidly disappear with the rising temperature after 2060. The findings suggest that the shrinked glacier and the reduced runoff threaten the water availability especially in summer seasons as well as affect the agricultural irrigation in the downstream of the Amu Darya River.

show abstract

P (()|, w, D) = \frac{P (()|, D, w) P ((), w)}{P ((), D)}

P (()|, w, D) = \frac{1}{Z_{S} ((),, α, β)} \exp ((), - β E_{D} - α E_{w})

Z_{s} ((),, α, β) = \int \exp ((), - β E_{D} - α E_{w}) italicdW

E_{D} ((), w) = \frac{1}{2} \sum_{i = 1}^{N} {(y_{i} (x_{i}, w) - t_{i})}^{2}

E_{w} ((), w) = \frac{1}{2} \sum_{i}^{W} w_{i}^{2}

Section: Development Of Mbha Methodsmentioning

confidence: 99%

An integrated multi‐GCMs Bayesian‐neural‐network hydrological analysis method for quantifying climate change impact on runoff of the Amu Darya River basin

Liu

et al. 2021

Intl Journal of Climatology

View full text Add to dashboard Cite

show abstract

“…Through the combination of two or more hydrological variables, mainly Qp and V , provide more reliability for hydraulic structure designing, water reservoir management, flood risk assessment and support flood mitigation in the studied basins (e.g. Balistrocchi et al, 2017; Callau Poduje et al, 2014; Jiang et al, 2019; Liu et al, 2019; Reddy & Ganguli, 2012; Requena et al, 2013; Salvadori & De Michele, 2004; Zhang & Singh, 2006; Zhou et al, 2019).…”

Section: Discussionmentioning

confidence: 99%

“…Multivariate FFA constitutes a powerful tool allowing the dimensioning of suitable installations based on a more accurate flood risk assessment (e.g. Balistrocchi, Orlandini, Ranzi, & Bacchi, 2017; Callau Poduje, Belli, & Haberlandt, 2014; Jiang, Yang, & Tatano, 2019; Liu, Li, Ma, Jia, & Su, 2019; Zhou, Liu, Jin, & Hu, 2019).…”

Section: Introductionmentioning

confidence: 99%

Complete multivariate flood frequency analysis, applied to northern Algeria

Karahaçane

Meddi²,

Chebana³

et al. 2020

J Flood Risk Management

View full text Add to dashboard Cite

Extreme hydrologic events are commonly described by several dependent characteristics, such as duration, volume and peak flow for floods. Traditionally in Algeria and North Africa, flood frequency analysis (FFA) is conducted as a univariate approach focusing separately on each single of flood characteristics. On the other hand, elsewhere, multivariate FFA studies have been conducted focusing on some FFA steps (especially modelling). The current study aims to consider complete multivariate FFA at-site case studies in northern Algeria using 11 hydrometric stations. It is also among the first studies dealing with multivariate FFA in a complete way by considering all the required steps of the analysis (multivariate outliers detection, multivariate assumptions testing and copula fitting) and on datasets from Algeria. Multivariate stationarity, homogeneity and independence assumptions have been well verified before modelling. The Weibull distribution is mostly selected as margin distribution for the duration, volume and peak flow series. Frank, Clayton and Gumbel copulas are commonly selected to describe the dependence structure on the three flood pairs of variables. These findings should be interesting in water management and flood risk assessment in these regions. Combining these flood characteristics enables the design of more efficient hydraulic structures.

show abstract

“…The snowmelt runoff created because of temperature rise (rise by between 4 and 5 °C) in the future, and because of precipitation has risen by 7% in the period up until 2100. Liu et al (2020) [10] showed that as little as 8% of the river basin glaciers in Central Asia might remain under scenarios A2 and B2, by the end of the 21st century, so most of the river basin glaciers are predicted to be lost by them.…”

Section: Introductionmentioning

confidence: 99%

Climate Change Scenarios and Effects on Snow-Melt Runoff

2020

View full text Add to dashboard Cite

Climate change is an important environmental issue, as progression of melting glaciers and snow cover is sensitive to climate alteration. The aim of this research was to model climate alterations forecasts, and to assess potential changes in snow cover and snow-melt runoff under the different climate change scenarios in the case study of the Zayandeh-rud River Basin. Three cluster models for climate change (NorESM1-M, IPSL-CM5A-LR and CSIRO-MK3.6.0) were applied under RCP 8.5, 4.5 and 2.6 scenarios, to examine climate influences on precipitation and temperature in the basin. Temperature and precipitation were determined for all three scenarios for four periods of 2021-2030, 2031-2040, 2041-2050 and 2051-2060. MODIS (MOD10A1) was also applied to examine snow cover using temperature and precipitation data. The relationship between snow-covered area, temperature and precipitation was used to forecast future snow cover. For modeling future snow melt runoff, a hydrologic model of SRM was used including input data of precipitation, temperature and snow cover. The results indicated that all three RCP scenarios lead to an increase in temperature, and reduction in precipitation and snow cover. Investigation in snowmelt runoff throughout the observation period (November 1970 to May 2006) showed that most of annual runoff is derived from snow melting. Maximum snowmelt runoff is generated in winter. The share of melt water in the autumn and spring runoff is estimated at 35 and 53%, respectively. The results of this study can assist water manager in making better decisions for future water supply.

show abstract

Development of a Bayesian-copula-based frequency analysis method for hydrological risk assessment – The Naryn River in Central Asia

Cited by 62 publications

References 51 publications

An integrated multi‐GCMs Bayesian‐neural‐network hydrological analysis method for quantifying climate change impact on runoff of the Amu Darya River basin

An integrated multi‐GCMs Bayesian‐neural‐network hydrological analysis method for quantifying climate change impact on runoff of the Amu Darya River basin

Complete multivariate flood frequency analysis, applied to northern Algeria

Climate Change Scenarios and Effects on Snow-Melt Runoff

Contact Info

Product

Resources

About