Search

The potential for extreme climate events to cluster in space and time has driven increased interest in understanding and predicting compound climate risks. Through a case study on floods in the Ohio River Basin, we demonstrated that low-frequency climate variability could drive spatial and temporal clustering of the risk of regional climate extremes. Long records of annual maximum streamflow from 24 USGS gauges were used to explore the regional spatiotemporal patterns of flooding and their associated large-scale climate modes. We found that the dominant time scales of flood risk in this basin were in the interannual (6–7 years), decadal (11–13 years), and secular bands and that different sub-regions within the Ohio River Basin responded differently to large-scale forcing. We showed that the leading modes of streamflow variability were associated with ENSO and secular trends. The low-frequency climate modes translated into epochs of increased and decreased flood risk with multiple extreme floods or the absence of extreme floods, thus informing the nature of compound climate-induced flood risk. A notable finding is that the secular trend was associated with an east-to-west shift in the flood incidence and the associated storm track. This is consistent with some expectations of climate change projections. [ABSTRACT FROM AUTHOR]

Short instrumental streamflow records in the South and East Tibetan Plateau (SETP) limit understanding of the full range and long-term variability in streamflow, which could greatly impact freshwater resources for about one billion people downstream. Here we reconstruct eight centuries (1200−2012 C.E.) of annual streamflow from the Monsoon Asia Drought Atlas in five headwater regions across the SETP. We find two regional patterns, including northern (Yellow, Yangtze, and Lancang-Mekong) and southern (Nu-Salween and Yarlung Zangbo-Brahmaputra) SETP regions showing ten contrasting wet and dry periods, with a dividing line of regional moisture regimes at ~32°−33°N identified. We demonstrate strong temporal nonstationarity in streamflow variability, and reveal much greater high/low mean flow periods in terms of duration and magnitude: mostly pre-instrumental wetter conditions in the Yarlung Zangbo-Brahmaputra and drier conditions in other rivers. By contrast, the frequency of extreme flows during the instrumental periods for the Yangtze, Nu-Salween, and Yarlung Zangbo-Brahmaputra has increased by ~18% relative to the pre-instrumental periods. A new study reconstructs eight-century streamflow over the South and East Tibetan Plateau, showing that observations underestimate the full range of long-term streamflow variability and revealing contrasting regional variability in streamflow between south and north study regions. [ABSTRACT FROM AUTHOR]

A novel Bayesian Hierarchical Network Model (BHNM) for ensemble forecasts of daily streamflow is presented that uses the spatial dependence induced by the river network topology and hydrometeorological variables from the upstream contributing area between station gauges. Model parameters are allowed to vary with time as functions of selected covariates for each day. Using the network structure to incorporate flow information from upstream gauges and precipitation from the immediate contributing area as covariates allows one to model the spatial correlation of flows simultaneously and parsimoniously. An application to daily monsoon period (July–August) streamflow at three gauges in the Narmada basin in central India for the period 1978–2014 is presented. The best set of covariates include daily streamflow from upstream gauges or from the gauge above the upstream gauges depending on travel times and daily precipitation from the area between two stations. The model validation indicates that the model is highly skillful relative to a null‐model of generalized linear regression, which represents the analogous non‐Bayesian forecast. The ensemble spread of BHNM accounts for the forecast uncertainty leading to reliable and skillful streamflow predictions. Key Points: We developed a Bayesian Hierarchical Network Model for ensemble forecasts of daily streamflow with the attendant uncertaintiesThe model provides ensemble forecasts at all the locations on a river network simultaneously, capturing the spatial and temporal correlationThe framework can be applied to any river network and with appropriate covariates [ABSTRACT FROM AUTHOR]

River floods cause extensive losses to economy, ecology, and society throughout the world. They are driven by the space‐time structure of catchment rainfall, which is determined by large‐scale, or even global‐scale, atmospheric processes. The identification of coherent, large‐scale atmospheric circulation structures that determine the moisture transport and convergence associated with rainfall‐induced flooding can help improve its predictability and phenomenology. In this paper, we extend a methodology, used for the analysis of extreme rainfall events, to high streamflow events (HSEs). The approach combines multiple machine learning methods to link HSEs to atmospheric circulation patterns. An application to the German streamflow network using reanalysis data for the period 1960 to 2012 is presented. Daily streamflow from 166 gauges, homogeneously distributed across Germany, are used. Geopotential height fields and integrated vapor transport (IVT) are derived from reanalysis data. An unsupervised neural network, Self Organizing Maps, is applied to geopotential height to identify a finite number of circulation patterns (CPs). Event synchronization between CPs and HSEs is used to establish if they are linked or not. If they are linked, the Event Synchronization method computes the delay between the occurrence of a CP and a HSE. Finally, local logistic regression is used to estimate the probability of occurrence of a HSE, as function of CP and IVT. We demonstrate that our approach is very effective to evaluate HSE probability occurrence across Germany. Key Points: Atmospheric circulation patterns that generate high streamflow events in Germany are identifiedEvent synchronization is used to determine synchronization and delay between atmospheric circulation patterns and high streamflow eventsThe occurrence probability of high streamflow events is conditioned on atmospheric circulation patterns and integrated water vapor transport [ABSTRACT FROM AUTHOR]

The Three Gorges Dam is the world's largest capacity hydropower station located in the Hubei province along the Yangtze River in China, which began operations in 2003. The dam also functions to store and regulate the downstream releases of water in order to provide flood control and navigational support in addition to hydropower generation. Flow regulation is particularly important for alleviating the impacts of low‐ and high‐flow events during the summer rainy season (June, July, and August). The impact of dam operations on summer flows is the focus of this work. Naturalized flows are modelled using a canonical correlation analysis and covariates of subbasin‐scale precipitation resulting in good model skill with an average correlation of 0.92. The model is then used to estimate natural flows in the period after dam operation. A comparison between modelled and gauged streamflow post 2003 is made and the impact of the dam on downstream flow is assessed. Streamflow variability is found to be strongly related to rainfall variability. An analysis of regional streamflow variability across the Yangtze River Basin showed a mode of spatially negatively correlated variability between the upper and lower basin areas. The Three Gorges Dam likely mitigated the occurrence of high‐flow events at Yichang station located near the dam. However, the high flow at the remaining stations in the lower reach is not noticeably alleviated due to the diminishing influence of the dam on distant downstream flows and the impact of the lakes downstream of the dam that act to attenuate flows. Three types of flow regime changes between naturalized and observed flows were defined and used to assess the changes in the occurrence of high‐ and low‐flow events resulting from dam operations. [ABSTRACT FROM AUTHOR]

Streamflow simulation of the headwater catchment of the Yellow River basin (HCYRB) in China is important for water resources management of the Yellow River basin. A statistical‐dynamical model, combining regular vine copulas with an optimization method for structure estimation, is presented with an application for simulating the monthly streamflow with local climate drivers at HCYRB. Local climate drivers for streamflow in every month are analyzed using rank‐based correlation. Precipitation, evaporation, and temperature generally show strong associations with streamflow. Winter streamflows relate to total precipitation of the wet season and total evaporation of October and November, while unfrozen‐month streamflows are correlated with evaporation and precipitation of current month and previous 1 month in the wet season. Both canonical vine and D‐vine copulas are applied to develop different conditional quantile functions for streamflows in different months with their dynamical covariates. The covariates are selected from historical streamflows and climate drivers with appropriate lags using partial correlations. The optimal vine trees are selected using the sequential maximum spanning tree algorithm with the weight based on both dependence and goodness of fit. The model demonstrates higher skill than existing vine‐based models and the seasonal autoregressive integrated moving average model. The enhanced skill of the hybrid statistical‐dynamical model comes from an improved capability of capturing nonlinear correlation and tail dependence of streamflow and climate drivers with the optimization of vine structure selection. The model provides an effective advance to enhance water resources planning and management for HCYRB and the whole basin. Key Points: The associations of streamflow in every month and different climate indices in Headwater Catchment of the Yellow River Basin are analyzedThe sequential maximum spanning tree optimization method is proposed for vine‐based monthly streamflow simulation with multiple predictorsThe model is verified skillful for its capability of capturing nonlinearity and tail dependence with the optimization of vine structures [ABSTRACT FROM AUTHOR]

A Bayesian model that uses the spatial dependence induced by the river network topology, and the leading principal components of regional tree ring chronologies for paleo‐streamflow reconstruction is presented. In any river basin, a convergent, dendritic network of tributaries come together to form the main stem of a river. Consequently, it is natural to think of a spatial Markov process that recognizes this topological structure to develop a spatially consistent basin‐scale streamflow reconstruction model that uses the information in streamflow and tree ring chronology data to inform the reconstructed flows, while maintaining the space‐time correlation structure of flows that is critical for water resource assessments and management. Given historical data from multiple streamflow gauges along a river, their tributaries in a watershed, and regional tree ring chronologies, the model is fit and used to simultaneously reconstruct the full network of paleo‐streamflow at all gauges in the basin progressing upstream to downstream along the river. Our application to 18 streamflow gauges in the Upper Missouri River Basin shows that the mean adjusted R2 for the basin is approximately 0.5 with good overall cross‐validated skill as measured by five different skill metrics. The spatial network structure produced a substantial reduction in the uncertainty associated with paleo‐streamflow as one proceeds downstream in the network aggregating information from upstream gauges and tree ring chronologies. Uncertainty was reduced by more than 50% at six gauges, between 6% and 50% at one gauge, and by less than 5% at the remaining 11 gauges when compared with the traditional principal component regression reconstruction model. Key points: A novel Bayesian network model for streamflow reconstructions using the spatial dependence induced by the river network and regional trees is presentedThe spatial network structure allows a substantial reduction in the uncertainty associated with paleo‐streamflowsThe spatial Markov model improves upon traditional streamflow reconstruction methods [ABSTRACT FROM AUTHOR]

We propose a time series modeling approach based on nonlinear dynamical systems to recover the underlying dynamics and predictability of streamflow and to produce projections with identifiable skill. First, a wavelet spectral analysis is performed on the time series to identify the dominant quasiperiodic bands. The time series is then reconstructed across these bands and summed to obtain a signal time series. This signal is embedded in a D‐dimensional space with an appropriate lag τ to reconstruct the phase space in which the dynamics unfolds. Time‐varying predictability is assessed by quantifying the divergence of trajectories in the phase space with time, using Local Lyapunov Exponents. Ensembles of projections from a current time are generated by block resampling trajectories of desired projection length, from the K‐nearest neighbors of the current vector in the phase space. This modeling approach was applied to the naturalized historical and paleoreconstructed streamflow at Lees Ferry gauge on the Colorado River, which offered three interesting insights. (i) The flows exhibited significant epochal variations in predictability. (ii) The predictability of the flow quantified by Local Lyapunov Exponent is related to the variance of the flow signal and selected climate indices. (iii) Blind projections of flow during epochs identified as highly predictable showed good skill in capturing the distributional and threshold exceedance statistics and poor performance during low predictability epochs. The ability to assess the potential skill of these long lead projections opens opportunities to perceive hydrologic predictability and consequently water management in a new paradigm. Key Points: The dynamics of the multidecadal streamflow signal from long paleo and observed record uncovered by reconstructing the phase spaceLocal Lyapunov Exponents are used to understand temporal variability of predictability potentially enabling predictability‐based managementStreamflow simulated by block resampling of trajectories from neighbors in phase space, with skills consistent with predictability [ABSTRACT FROM AUTHOR]

Low‐flow events can cause significant impacts to river ecosystems and water‐use sectors; as such, it is important to understand their variability and drivers. In this study, we characterise the variability and timing of annual total frequency of low‐streamflow days across a range of headwater streams within the continental United States. To quantify this, we use a metric that counts the annual number of low‐flow days below a given threshold, defined as the cumulative dry days occurrence (CDO). First, we identify three large clusters of stream gauge locations using a Partitioning Around Medoids (PAM) clustering algorithm. In terms of timing, results reveal that for most clusters, the majority of low‐streamflow days occur from the middle of summer until early fall, although several locations in Central and Western United States also experience low‐flow days in cold seasons. Further, we aim to identify the regional climate and larger scale drivers for these low‐streamflow days. Regionally, we find that precipitation deficits largely associate with low‐streamflow days in the Western United States, whereas within the Central and Eastern U.S. clusters, high temperature indicators are also linked to low‐streamflow days. In terms of larger scale, we examine sea surface temperature (SST) anomalies, finding that extreme dry years exhibit a high degree of co‐occurrence with different patterns of warmer SST anomalies across the Pacific and Northern Atlantic Oceans. The linkages identified with regional climate and SSTs offer promise towards regional prediction of changing conditions of low‐streamflow events. [ABSTRACT FROM AUTHOR]

Our understanding of the full range of natural variability in streamflow, including how modern flow compares to the past, is poorly understood for the Upper Indus Basin because of short instrumental gauge records. To help address this challenge, we use Hierarchical Bayesian Regression with partial pooling to develop six centuries long (1394–2008 CE) streamflow reconstructions at three Upper Indus Basin gauges (Doyian, Gilgit, and Kachora), concurrently demonstrating that Hierarchical Bayesian Regression can be used to reconstruct short records with interspersed missing data. At one gauge (Partab Bridge), with a longer instrumental record (47 years), we develop reconstructions using both Bayesian regression and the more conventionally used principal components regression. The reconstructions produced by principal components regression and Bayesian regression at Partab Bridge are nearly identical and yield comparable reconstruction skill statistics, highlighting that the resulting tree ring reconstruction of streamflow is not dependent on the choice of statistical method. Reconstructions at all four reconstructions indicate that flow levels in the 1990s were higher than mean flow for the past six centuries. While streamflow appears most sensitive to accumulated winter (January–March) precipitation and summer (May–September) temperature, with warm summers contributing to high flow through increased melt of snow and glaciers, shifts in winter precipitation and summer temperatures cannot explain the anomalously high flow during the 1990s. Regardless, the sensitivity of streamflow to summer temperatures suggests that projected warming may increase streamflow in coming decades, though long‐term water risk will additionally depend on changes in snowfall and glacial mass balance. Key Points: Tree ring reconstructions of streamflow in the Upper Indus Basin show wetter conditions in the 1990s compared to the last 600 yearsReconstructions are insensitive to the choice of statistical method used (principal components versus Bayesian regression)Streamflow is most sensitive to winter precipitation and summer temperature, but anomalies in these seasons cannot explain recent high flow [ABSTRACT FROM AUTHOR]

The development of paleoclimate streamflow reconstructions in the conterminous United States (CONUS) has provided water resource managers with improved insights into multidecadal and centennial scale variability that cannot be reliably detected using shorter instrumental records. Paleoclimate streamflow reconstructions have largely focused on individual catchments limiting the ability to quantify variability across the CONUS. The Living Blended Drought Atlas (LBDA), a spatially and temporally complete 555 year long paleoclimate record of summer drought across the CONUS, provides an opportunity to reconstruct and characterize streamflow variability at a continental scale. We explore the validity of the first paleoreconstructions of streamflow that span the CONUS informed by the LBDA targeting a set of U.S. Geological Survey streamflow sites. The reconstructions are skillful under cross validation across most of the country, but the variance explained is generally low. Spatial and temporal structures of streamflow variability are analyzed using hierarchical clustering, principal component analysis, and wavelet analyses. Nine spatially coherent clusters are identified. The reconstructions show signals of contemporary droughts such as the Dust Bowl (1930s) and 1950s droughts. Decadal-scale variability was detected in the late 1900s in the western U.S., however, similar modes of temporal variability were rarely present prior to the 1950s. The twentieth century featured longer wet spells and shorter dry spells compared with the preceding 450 years. Streamflows in the Pacific Northwest and Northeast are negatively correlated with the central U.S. suggesting the potential to mitigate some drought impacts by balancing economic activities and insurance pools across these regions during major droughts. [ABSTRACT FROM AUTHOR]

Recent advances in paleoclimatology have revealed dramatic long-term hydroclimatic variations that provide a context for limited historical records. A notable data set derived from a relatively dense network of paleoclimate proxy records in North America is the Living Blended Drought Atlas (LBDA): a gridded tree-ring-based reconstruction of summer Palmer Drought Severity Index. This index has been used to assess North American drought frequency, persistence, and spatial extent over the past two millennia. Here, we explore whether the LBDA can be used to reconstruct annual streamflow. Relative to streamflow reconstructions that use tree rings within the river basin of interest, the use of a gridded proxy poses a novel challenge. The gridded series have high spatial correlation, since they rely on tree rings over a common radius of influence. A novel algorithm for reconstructing streamflow using regularized canonical regression and inputs of local and global covariates is developed and applied over the Missouri River Basin, as a test case. Effectiveness in reconstruction is demonstrated with reconstructions showing periods where streamflow deficits may have been more severe than during recent droughts (e.g., the Civil War, Dust Bowl, and 1950s droughts). The maximum persistence of droughts and floods over the past 500 years far exceeds those observed in the instrumental record and periods of multidecadal variability in the 1500s and 1600s are detected. Challenges for an extension to a national streamflow reconstruction or applications using other gridded paleoclimate data sets such as adequate spatial coverage of streamflow and applicability of annual reconstructions are discussed. [ABSTRACT FROM AUTHOR]

A model to generate stochastic streamflow projections conditioned on quasi-oscillatory climate indices such as Pacific Decadal Oscillation (PDO) and Atlantic Multi-decadal Oscillation (AMO) is presented. Recognizing that each climate index has underlying band-limited components that contribute most of the energy of the signals, we first pursue a wavelet decomposition of the signals to identify and reconstruct these features from annually resolved historical data and proxy based paleoreconstructions of each climate index covering the period from 1650 to 2012. A K-Nearest Neighbor block bootstrap approach is then developed to simulate the total signal of each of these climate index series while preserving its time-frequency structure and marginal distributions. Finally, given the simulated climate signal time series, a K-Nearest Neighbor bootstrap is used to simulate annual streamflow series conditional on the joint state space defined by the simulated climate index for each year. We demonstrate this method by applying it to simulation of streamflow at Lees Ferry gauge on the Colorado River using indices of two large scale climate forcings: Pacific Decadal Oscillation (PDO) and Atlantic Multi-decadal Oscillation (AMO), which are known to modulate the Colorado River Basin (CRB) hydrology at multidecadal time scales. Skill in stochastic simulation of multidecadal projections of flow using this approach is demonstrated. [ABSTRACT FROM AUTHOR]

Summary Historically, flood risk management and flood frequency modeling have been based on assumption of stationarity, i.e., flood probabilities are invariant across years. However, it is now recognized that in many places, extreme floods are associated with specific climate states which may recur with non-uniform probability across years. Conditional on knowledge of the operating climate regime, the probability of a flood of a certain magnitude can be higher or lower in a given year. Here we explore nonstationary flood risk for the streamflow series of the Negro River at the city of Manaus in Brazil by investigating climate teleconnections associated with the interannual variability of the peak flows. We evaluate attributes and the fit of a generalized extreme value (GEV) distribution with nonstationary parameters to the annual peak series of the Negro River stages. The annual peak flood occurs between May and July and its magnitude depends on the Negro River stage at the beginning of the year and on the previous December sea surface temperature (SST) of a region in the tropical Pacific Ocean. A statistically significant monotonic trend is also observed in the peak level series. The indexing of the parameters of a GEV distribution to the NINO3 index and to the observed river stage at the beginning of the year reveals a changing flood hazard for the city, with the joint occurrence of high values associated with La Niña conditions in the previous December and high river stages in January preceding the flood season. The proposed model is shown to be useful for quantifying the changing flood hazard several months in advance for Manaus, thus providing an early flood alert system for the city and may be an important tool for the dynamic flood risk management for the region. [ABSTRACT FROM AUTHOR]

The Masinga Reservoir located in the upper Tana River basin, Kenya, is extremely important in supplying the country's hydropower and protecting downstream ecology. The dam serves as the primary storage reservoir, controlling streamflow through a series of downstream hydroelectric reservoirs. The Masinga dam's operation is crucial in meeting power demands and thus contributing significantly to the country's economy. La Niña-related prolonged droughts of 1999-2001 resulted in severe power shortages in Kenya. Therefore, seasonal streamflow forecasts contingent on climate information are essential to estimate preseason water allocation. Here, the authors utilize reservoir inflow forecasts downscaled from monthly updated precipitation forecasts from ECHAM4.5 forced with constructed analog SSTs and multimodel precipitation forecasts developed from the Ensemble-Based Predictions of Climate Changes and their Impacts (ENSEMBLES) project to improve water allocation during the April-June and October-December seasons for the Masinga Reservoir. Three-month-ahead inflow forecasts developed from ECHAM4.5, multiple GCMs, and climatological ensembles are used in a reservoir model to allocate water for power generation by ensuring climatological probability of meeting the end-of-season target storage required to meet seasonal water demands. Retrospective reservoir analysis shows that inflow forecasts developed from single GCM and multiple GCMs perform better than use of climatological values by reducing the spill and increasing the allocation for hydropower during above-normal inflow years. Similarly, during below-normal inflow years, both of these forecasts could be effectively utilized to meet the end-of-season target storage by restricting releases for power generation. The multimodel forecasts preserve the end-of-season target storage better than the single-model inflow forecasts by reducing uncertainty and the overconfidence of individual model forecasts. [ABSTRACT FROM AUTHOR]

A hierarchical Bayesian regression model is presented for reconstructing the average summer streamfiow at five gauges in the Delaware River basin using eight regional tree-ring chronologies. The model provides estimates of the posterior probability distribution of each reconstructed streamfiow series considering parameter uncertainty. The vectors of regression coefficients are modeled as draws from a common multivariate normal distribution with unknown parameters estimated as part of the analysis. This leads to a multilevel structure. The covariance structure of the streamfiow residuals across sites is explicitly modeled. The resulting partial pooling of information across multiple stations leads to a reduction in parameter uncertainty. The effect of no pooling and full pooling of station information, as end points of the method, is explored. The no- pooling model considers independent estimation of the regression coefficients for each streamfiow gauge with respect to each tree-ring chronology. The full-pooling model considers that the same regression coefficients apply across all streamfiow sites for a particular tree-ring chronology. The cross-site correlation of residuals is modeled in all cases. Performance on metrics typically used by tree-ring reconstruction experts, such as reduction of error, coefficient of efficiency, and coverage rates under credible intervals is comparable to, or better, for the partial-pooling model relative to the no-pooling model, and streamfiow estimation uncertainty is reduced. Long record simulations from reconstructions are used to develop estimates of the probability of duration and severity of droughts in the region. Analysis of monotonic trends in the reconstructed drought events do not reject the null hypothesis of no trend at the 90% significance over 1754-2000. [ABSTRACT FROM AUTHOR]

Summary: Here we analyze the variability of MAM (March–April–May) and JJAS (June–July–August–September) seasonal Satluj River flow into the Bhakra dam in India through Pearson anomaly correlation and composite analyses with antecedent and concurrent seasonal climatic and atmospheric circulation patterns. The MAM seasonal inflow of Bhakra dam is significantly correlated with winter (DJF/FM) precipitation and temperature of the Satluj basin while the correlation with FM was more prominent for precipitation (snow=+0.72, rainfall=+0.60), and temperature (diurnal temperature range (DTR)=−0.76 and maximum temperature (T max)=−0.57). The JJAS inflow was also positively correlated with DJF/FM as well as JJAS precipitation of the Satluj basin while the correlation with basin average FM was the largest (+0.54). These suggested that both MAM and JJAS inflow anomalies are linked with DJF/FM climate over the Western Himalayas and adjoining north and central Indian plains, which were also found to be linked with the fluctuation of equatorial concurrent Sea Surface Temperature anomalies over the western Indian Ocean (max anomaly correlation was>+0.70) and mean sea level pressure over western pole of the Southern Oscillation sea-saw region (max Pearson anomaly correlation was∼+0.60). Low (high) MAM inflow was found to be associated with negative (positive) precipitation anomalies over the basin and north India in DJF and FM while FM precipitation anomaly is more concentrated over the Western Himalayas. In addition, low (high) JJAS inflow is also associated with negative (positive) precipitation anomalies over the basin and north India in DJF and over the Western Himalaya in FM and JJAS. Negative geopotential height anomaly at 500hPa (Z500) over Siberia and northwestern pacific in DJF, and positive Z500 anomaly over the northwest India in FM were noticed in low MAM inflow years. Whereas high inflow in MAM was linked with a negative Z500 anomaly between two positive Z500 anomaly regions – one over eastern Siberia stretched up to northern Pacific and second over the Eastern Europe in DJF, which gets stronger in FM. We also found southwesterly (northeasterly) wind vectors at 850hPa pressure level (uv850) bringing more (less) moisture to the Western Himalayas in DJF and FM in high (low) MAM/JJAS flow years. [Copyright &y& Elsevier]

Seasonal streamflow forecasts based on climate information can guide water managers toward superior reservoir operations, leading to improved water resources management efficiency. Uncertainty, however, is always present in seasonal streamflow forecasts, affecting the forecast value. Thus, a forecast should not be considered complete without a description of its uncertainty, which is critical for climate risk and water resources management. This study investigates the uncertainties of a seasonal streamflow forecast system for Northeastern Brazil based on climate precipitation forecasts and hydrologic modeling. These two sources of uncertainty are treated independently and then compared in order to guide future investments in the forecast system. Sea surface temperature is considered to be the primary source of uncertainty for the seasonal precipitation forecasts, based upon a 10-member climate model ensemble. Parameter uncertainty is considered to be the only source of uncertainty for the hydrologic model. Estimation of parameter uncertainty is estimated by the Shuffled Complex Evolution Metropolis algorithm, which employs a Markov Chain Monte Carlo scheme to provide the posterior distribution of the parameters and form uncertainty bounds on streamflow forecasts. Results indicate that uncertainties associated with the climate forecast are much larger than those from parameter estimation in the hydrologic model. Although model structure has not been considered in the evaluation of hydrologic uncertainties, this study indicates that future efforts to address the predominant source of uncertainty should focus on the climate prediction models. Copyright © 2011 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

In this study, statistical linkages between autumn streamflow in the northeast United States and preceding summer sea surface temperatures are developed to establish predictive potential for climate-informed seasonal streamflow forecasts in this region. Predictor regions with physically plausible teleconnections to local streamflow are identified and evaluated in a multivariate and nonlinear framework using local regression techniques. Three such regions are identified, located in the Bering Sea, the tropical Pacific just west of Mexico, and the tropical Atlantic off the coast of Africa. Asymmetries in each region's univariate local regression result are apparent, and bivariate local regressions are used to attribute these asymmetries to interactions with physical mechanisms associated with the other two regions, and possibly other unaccounted for climatic predictors. A bivariate model including the tropical Pacific and tropical Atlantic regions yields the strongest local regression result, explaining 0.68 of the interannual streamflow variability. An analogous multivariate linear regression analysis is only able to explain 0.20 of the streamflow variability and thus the use of nonlinear methods' results in a marked improvement in streamflow simulation capability. Cross-validation considerably weakens the streamflow forecasts using this model; however, forecast skill may improve with a longer period of record or the inclusion of additional predictors. Copyright © 2010 Royal Meteorological Society [ABSTRACT FROM AUTHOR]

Streamflow disaggregation techniques are used to distribute a single aggregate flow value to multiple sites in both space and time while preserving distributional statistics (i.e., mean, variance, skewness, and maximum and minimum values) from observed data. A number of techniques exist for accomplishing this task through a variety of parametric and nonparametric approaches. However, most of these methods do not perform well for disaggregation to daily time scales. This is generally due to a mismatch between the parametric distributions appropriate for daily flows versus monthly or annual flows, the high dimension of the disaggregation problem, compounded uncertainty in parameter estimation for multistage approaches, and the inability to maintain flow continuity across disaggregation time period boundaries. We present a method that directly simulates daily data at multiple locations from a single annual flow value via K-nearest neighbor (K-NN) resampling of daily flow proportion vectors. The procedure is simple and data driven and captures observed statistics quite well. Furthermore, the generated daily data are continuous and display lag correlation structure consistent with that of the observed data. The utility and effectiveness of this approach is demonstrated for selected sites in the San Juan River Basin, located in southwestern Colorado, and later compared with the disaggregation technique of Prairie et al. (2007) for several locations in the Colorado River Basin. [ABSTRACT FROM AUTHOR]

The article suggests a modified support vector machine (SVM) based prediction model for the improvement of the predictability of the inflow to Shihmen Reservoir in Taiwan. In building the proposed framework, highly correlated climate precursors are identified and adopted to predict water availability and genetic algorithm based parameter determination procedure is enforced to the SVM parameters to know the non-linear patterns in climate systems. Bagging is then applied to create the models.

Gong, Gavin, Lucien Wang, Laura Condon, Alastair Shearman, and Upmanu Lall, 2010. A Simple Framework for Incorporating Seasonal Streamflow Forecasts Into Existing Water Resource Management Practices. Journal of the American Water Resources Association (JAWRA) 46(3):574-585. DOI: 10.1111/j.1752-1688.2010.00435.x Climate-based streamflow forecasting, coupled with an adaptive reservoir operation policy, can potentially improve decisions by water suppliers and watershed stakeholders. However, water suppliers are often wary of straying too far from their current management practices, and prefer forecasts that can be incorporated into existing system modeling tools. This paper presents a simple framework for utilizing streamflow forecasts that works within an existing management structure. Climate predictors are used to develop seasonal inflow forecasts. These are used to specify operating rules that connect to the probability of future (end of season) reservoir states, rather than to the current storage, as is done now. By considering both current storage and anticipated inflow, the likelihood of meeting management goals can be improved. The upper Delaware River Basin in the northeastern United States is used to demonstrate the basic idea. Physically plausible climate-based forecasts of March-April reservoir inflow are developed. Existing simulation tools and rule curves for the system are used to convert the inflow forecasts to reservoir level forecasts. Operating policies are revised during the forecast period to release less water during forecasts of low reservoir level. Hindcast simulations demonstrate reductions of 1.6% in the number of drought emergency days, which is a key performance measure. Forecasts with different levels of skill are examined to explore their utility. [ABSTRACT FROM AUTHOR]

Summary: Several studies have shown that statistics of streamflow time series, in particular empirical moments, scale with physical properties of the drainage basin, such as the catchment area. Those scaling laws have been extensively used to estimate statistics of streamflow series at ungauged sites. The role of climate variability and change has not been considered in such models. Further, most studies are based on classical statistics, where parameter uncertainties are usually neglected or not formally considered. In this paper we develop and apply hierarchical Bayesian models, to both assess regional and at-site trends in time in a spatial scaling framework, and simultaneously provide a rigorous framework for assessing and reducing parameter and model uncertainties. The models are tested with reconstructed natural inflow series from over 40 hydropower sites in Brazil with catchments areas varying from 2588 to 823,555km2. Both annual maximum flood series and monthly streamflow are considered. Cross-validated results show that the Hierarchical Bayesian models are able to skillfully estimate monthly and flood flow probability distribution parameters for sites that were not used in model fitting. The models developed can be used to provide record augmentation at sites that have short records, or to estimate flow at ungauged sites, even in the absence of an assumption of time stationarity. Since model uncertainties are accounted for, the precision of the estimates can be quantified and hypotheses tests for regional and at-site trends can be formally made. A formal inclusion of climate predictors to facilitate seasonal forecasting or climate change scenario development is also feasible. This is indicated, but not developed here. [Copyright &y& Elsevier]

Summary: Efficient management of water and energy is an important goal of sustainable development for any nation. Streamflow forecasts, have been used in complex optimization models to maximize water use efficiency and electrical energy production. In this paper we develop a statistical model for the long term forecasts of hydroenergy inflow into the Brazilian hydropower system, which consists of more than 70 hydropower reservoirs. At present, the planning of reservoir operation and energy production in Brazil is made with no reliable long term (one season or longer lead times) streamflow forecasts. Here we use the NINO3 index and the main modes of the tropical Pacific thermocline structure as climate predictors in order to achieve skillfull forecasts at long leads. Cross-validated results show that about 50% of the total hydroenergy inflow can be predicted with moderate accuracy up to 20 month lead time. [Copyright &y& Elsevier]

Summary: Streamflow simulation and forecasts have been widely used in water resources management, particularly for flood and drought analysis and for the determination of optimal operational rules for reservoir systems used for water supply and energy production. Here we include climate information in a periodic-auto-regressive model in order to provide monthly streamflow forecasts for 54 hydropower sites in Brazil. Large scale climate information is included in the model through the use of climate indices obtained from the sea surface temperature field of the tropical Pacific and sub-tropical Atlantic oceans and the low-level zonal wind field over southeast Brazil. Correlation analysis of climate predictors and streamflow data show that the dependence of the latter on climate variability is seasonal and also a function of the lead time of the forecasts. A ridge regression framework is adopted in order to shrink parameter estimates and improve model outputs. The proposed model is compared with an ordinary linear regression based model with predictors selected by the BIC criterion and with the classical linear periodic-auto-regressive model (PAR), where no climate information is used. Cross-validated results show that the inclusion of climate indexes is able to improve forecast skills up to 3 months lead time. Higher skills are observed for reservoirs with large catchment areas. [Copyright &y& Elsevier]

Seasonal streamflow forecasts contingent on climate information are essential for short-term planning (e.g., water allocation) and for setting up contingency measures during extreme years. However, the water allocated based on the climate forecasts issued at the beginning of the season needs to be revised using the updated climate forecasts throughout the season. In this study, reservoir inflow forecasts downscaled from monthly updated precipitation forecasts from ECHAM4.5 forced with “persisted” SSTs were used to improve both seasonal and intraseasonal water allocation during the October–February season for the Angat reservoir, a multipurpose system, in the Philippines. Monthly updated reservoir inflow forecasts are ingested into a reservoir simulation model to allocate water for multiple uses by ensuring a high probability of meeting the end-of-season target storage that is required to meet the summer (March–May) demand. The forecast-based allocation is combined with the observed inflows during the season to estimate storages, spill, and generated hydropower from the system. The performance of the reservoir is compared under three scenarios: forecasts issued at the beginning of the season, monthly updated forecasts during the season, and use of climatological values. Retrospective reservoir analysis shows that the operation of a reservoir by using monthly updated inflow forecasts reduces the spill considerably by increasing the allocation for hydropower during above-normal-inflow years. During below-normal-inflow years, monthly updated streamflow forecasts could be effectively used for ensuring enough water for the summer season by meeting the end-of-season target storage. These analyses suggest the importance of performing experimental reservoir analyses to understand the potential challenges and opportunities in improving seasonal and intraseasonal water allocation by using real-time climate forecasts. [ABSTRACT FROM AUTHOR]

Results from diagnostic analyses to understand the seasonal evolution of the large-scale climatic state responsible for the development and melt of the winter snowpack, and spring–early summer precipitation in the Churchill Falls region on the Québec-Labrador Peninsula, Canada, are presented in the context of the development of an empirical model for seasonal to annual streamflow forecasting, with a special emphasis on the May–July spring freshet. Teleconnection indices and gridded global measures of atmospheric circulation inferred from the National Centers for Environmental Prediction/National Center for Atmospheric Research reanalysis are used as climatic indicators. Composite and correlation analyses are applied to the climatic indicators conditioned on the spring streamflow for identification of potential predictors. Meridional and zonal atmospheric fluxes over the Atlantic and the Pacific Oceans emanating from regionally persistent sea surface temperature/sea level pressure modes are identified as potential carriers of information. We speculate on the ocean-atmosphere and regional hydrologic mechanisms that may be involved in lending multiseasonal predictability to streamflows in the region. [ABSTRACT FROM AUTHOR]

Seasonal streamflow forecasts contingent on climate information are essential for water resources planning and management as well as for setting up contingency measures during extreme years. In this study, operational streamflow forecasts are developed for a reservoir system in the Philippines using ECHAM4.5 precipitation forecasts (EPF) obtained using persisted sea surface temperature (SST) scenarios. Diagnostic analyses on SST conditions show that the tropical SSTs influence the streamflow during extreme years, whereas the local SSTs (0°–25°N, 115°–130°E) account for streamflow variability during normal years. Given that the EPF, local, and tropical SST conditions are spatially correlated, principal components regression (PCR) is employed to downscale the GCM-predicted precipitation fields and SST anomalies to monthly streamflow forecasts and to update them every month within the season using the updated EPF and SST conditions. These updated forecasts improve the prediction of monthly streamflows within the season in comparison to the skill of the monthly streamflow forecasts issued at the beginning of the season. It is also shown that the streamflow forecasting model developed using EPF under persisted SST conditions performs well upon employing EPF obtained under predicted SSTs as predictor. This has potential implications in the development of operational streamflow forecasts and statistical downscaling, which requires adequate years of retrospective GCM forecasts for recalibration. Finally, the study also shows that predicting the seasonal streamflow using the monthly precipitation forecasts reproduces the observed seasonal total better than the conventional approach of using seasonal precipitation forecasts to predict the seasonal streamflow. [ABSTRACT FROM AUTHOR]

Seasonal forecasting of streamflow provides many benefits to society, by improving our ability to plan and adapt to changing water supplies. A common approach to developing these forecasts is to use statistical methods that link a set of predictors representing climate state as it relates to historical streamflow, and then using this model to project streamflow one or more seasons in advance based on current or a projected climate state. We present an approach for forecasting multivariate time series using independent component analysis (ICA) to transform the multivariate data to a set of univariate time series that are mutually independent, thereby allowing for the much broader class of univariate models to provide seasonal forecasts for each transformed series. Uncertainty is incorporated by bootstrapping the error component of each univariate model so that the probability distribution of the errors is maintained. Although all analyses are performed on univariate time series, the spatial dependence of the streamflow is captured by applying the inverse ICA transform to the predicted univariate series. We demonstrate the technique on a multivariate streamflow data set in Colombia, South America, by comparing the results to a range of other commonly used forecasting methods. The results show that the ICA-based technique is significantly better at representing spatial dependence, while not resulting in any loss of ability in capturing temporal dependence. As such, the ICA-based technique would be expected to yield considerable advantages when used in a probabilistic setting to manage large reservoir systems with multiple inflows or data collection points. [ABSTRACT FROM AUTHOR]

We assess the potential benefits from innovative forecasts of the stream flows that replenish reservoirs in the semi-arid state of Ceará, Brazil. Such forecasts have many potential applications. In Ceará, they matter for both water-allocation and participatory-governance issues that echo global debates. Our qualitative analysis, based upon extensive fieldwork with farmers, agencies, politicians and other key actors in the water sector, stresses that forecast value changes as a society shifts. In the case of Ceará, current constraints on the use of these forecasts are likely to be reduced by shifts in water demand, water allocation in the agricultural Jaguaribe Valley, participatory processes for water allocation between this valley and the capital city of Fortaleza, and risk perception. Such changes in the water sector can also have major distributional impacts. [ABSTRACT FROM AUTHOR]

A semiparametric approach for forecasting streamflow at multiple gaging locations on a river network conditional on climate precursors is developed. The strategy considers statistical forecasts of annual or seasonal streamflow totals at each of the sites and their disaggregation to monthly or higher resolution flows using a k nearest neighbor resampling approach that maintains space-time consistency across the sites and subperiods. An application of the approach to forecasting inflows at six reservoirs in the state of Ceara in northeastern Brazil is presented. The climate precursors used are the NINO3 index for the El Niño-Southern Oscillation and an equatorial Atlantic sea surface temperature index. Forecasts of January through December streamflow are made at three lead times: in January of the same year and in October and July of the preceding year. The skill of the ensemble forecasts generated is evaluated on subsets of the historical data not used for model building. Correlations with the equatorial Atlantic index and with NINO3 translate into useful streamflow forecasts for the next 18 months of reservoir operation and water management. [ABSTRACT FROM AUTHOR]

• A new annual (Nov-Oct) streamflow reconstruction of HCYRB at Tangnaihai Station of nearly two millennia was presented. • The nested principal component regression model is improved by the stepwise best tree-ring subset selection method in terms of AIC, CRSQ , VRSQ , CE, and RE. • The significant high-flow periods of HCYRB are the early 3rd century, circa 300 C.E., early 13th century, 16th century and circa 1900 C.E., while the low-flow periods are the late 5th century and late 15th century. • The reconstruction suggests that a warm climate is more likely accompanied by a high-flow period and low-flow periods are more likely to happen in cold periods. • The results provide adequate data foundation to analyze characteristics of streamflow of HCYRB and long-term optimal operation of Longyangxia over-year regulation reservoir. The headwater catchment of the Yellow River Basin (HCYRB) controls 35% of the streamflow of the Yellow River (YR) which faces increasing water shortages. To better understand streamflow variability in the region we require a better understanding of high and low flow characteristics. This study presents a new annual (Nov-Oct) streamflow reconstruction at the Tangnaihai station in the HCYRB for the last two millennia (159–2016 C.E.) using 12 tree-ring chronologies. The nested principal component regression model combined with the stepwise best subset selection method was proposed to improve the temporal length and model skill of reconstruction. The stepwise best subset selection method was presented to select the best principal components subset, instead of a confidence test, based on k-fold cross-validation error and Akaike's information criteria (AIC). The model assessment results verify that the proposed model exhibits strong reconstruction skills. Besides, the magnitude and duration of both high and low flow periods were analyzed. The results show that (1) the significant high-flow periods are the early 3rd century, circa 300 C.E., early 13th century, 16th century and circa 1900 C.E., while the low-flow periods are the late 5th century and late 15th century; (2) the durations and magnitudes of low-flow periods are longer and larger than high-flow periods and the severities of high-flow periods are greater than low-flow periods. The reconstruction also suggests that a warm climate is more likely accompanied by a high-flow period and low-flow periods are more likely to occur in cold periods associated with the Asian Summer Monsoon and solar activity. [ABSTRACT FROM AUTHOR]

A multi-reservoir simulation-optimization model GRAPS, Generalized Multi-Reservoir Analyses using Probabilistic Streamflow Forecasts, is developed in which reservoirs and users across the basin are represented using a node-link representation. Unlike existing reservoir modeling software, GRAPS can handle probabilistic streamflow forecasts represented as ensembles for performing multi-reservoir prognostic water allocation and evaluate the reliability of forecast-based allocation with observed streamflow. GRAPS is applied to four linked reservoirs in the Jaguaribe Metropolitan Hydro-System (JMH) in Ceará, North East Brazil. Results from the historical simulation and the zero-inflow policy over the JMH system demonstrate the model's capability to support monthly water allocation and reproduce the observed monthly releases and storages. Additional analyses using streamflow forecast ensembles illustrate GRAP's abilities in developing storage-reliability curves under inflow-forecast uncertainty. Our analyses show that GRAPS is versatile and can be applied for 1) short-term operating policy studies, 2) long-term basin-wide planning evaluations, and 3) climate-information based application studies. • GRAPS is a generalized multi-reservoir simulation-optimization model for water allocation and reservoir management. • GRAPS is intended to assist engineers and stakeholders as a decision-supporting tool to support seasonal water allocation. • GRAPS can handle streamflow forecasts as ensembles to quantify the reliability of meeting target storage and demand. [ABSTRACT FROM AUTHOR]

The Ganges-Brahmaputra-Meghna river system is the third largest in the world with an annual discharge of approximately 40,000m3/s. The Brahmaputra alone contributes nearly half of the total discharge. Climatic controls on streamflow in the lower Brahmaputra basin are complex and vary seasonally, with glacial and snow melt from the Himalaya and Tibetan Plateau dominating in the fall, winter and spring, and Indian monsoon precipitation acting as the major contributor to summer flow. Here we present a suite of four seasonal and one annual reconstruction of past discharge at the Bahadurabad gauging station in north-eastern Bangladesh based on a careful evaluation of streamflow clustering across different months and streamflow and tree-ring predictor relationships. The spring transition season (May-June), monsoon season (July-September), post-monsoon season (October-December), and annual flow (January-December) reconstructions extend back six centuries from ~1400 to 2014 C.E. The dry winter season reconstruction (January-April) is millennial length and spans ~300 to 2014 C.E. These resulting reconstructions have little predictor overlap allowing us compare low frequency variability and recent trends if any across different seasons and the entire year. The short instrumental data (1956-2014 C.E.) shows increasing dry season flow and decreasing wet season flow, particularly since the 1990s. While these changes are likely associated with increased glacial melt and decreasing monsoon precipitation respectively, they lie within reconstructed paleo-discharge variability estimates. The deltaic flood plains of Bangladesh are extremely vulnerable to catastrophic seasonal floods caused by heavy summer monsoon precipitation. Major flooding events occurred in 1951, 1966, 1987, 1988, and 1998. Our reconstruction indicates unusually high summer discharge in all of these years. The single year highest discharge value over the length of the six-century long reconstruction occurred in 1998 suggesting that this event was highly unusual in the long-term context. Using our reconstructions, we find on average five single-year high discharge events related to flooding occur every century, but these events are more likely to occur during periods of decadal high flow. This suggests a potential to use our multi-centennial reconstruction for flood-hazard risk estimation. [ABSTRACT FROM AUTHOR]