Search

Stochastic hydrology produces ensembles of time series that represent plausible future streamflow to simulate and test the operation of water resource systems. A premise of stochastic hydrology is that ensembles should be statistically representative of what may occur in the future. In the past, the application of this premise has involved producing ensembles that are statistically equivalent to the observed or historical streamflow sequence. This requires a number of metrics or statistics that can be used to test statistical similarity. However, with climate change, the past may no longer be representative of the future. Ensembles to test future systems operations should recognize non‐stationarity and include time series representing expected changes. This poses challenges for their testing and validation. In this paper, we suggest an evidence‐based analysis in which streamflow ensembles, whether statistically similar to and representative of the past or a changing future, should be characterized and assessed using an extensive set of statistical metrics. We have assembled a broad set of metrics and applied them to annual streamflow in the Colorado River at Lees Ferry to illustrate the approach. We have also developed a tree‐based classification approach to categorize both ensembles and metrics. This approach provides a way to visualize and interpret differences between streamflow ensembles. The metrics presented, along with the classification, provide an analytical framework for characterizing and assessing the suitability of future streamflow ensembles, recognizing the presence of non‐stationarity. This contributes to better planning in large river basins, such as the Colorado, facing water supply shortages. Plain Language Summary: Long‐range water supply planning in many river basins requires an assessment of ensembles of plausible future streamflow time series used to simulate and test the operation of water resource systems. With climate change, and growing recognition that hydrologic processes are changing over time, the past may no longer be representative of the future. This poses challenges when using statistical metrics to test future streamflow ensembles. In this paper, we suggest an evidence‐based approach in which streamflow ensembles, whether statistically similar to and representative of the past or a changing future, should be characterized using an extensive set of statistical metrics. We have assembled a broad set of metrics and applied them to annual streamflow in the Colorado River at Lees Ferry to illustrate the approach. We have also developed an approach to categorize both ensembles and metrics. The metrics presented and the classification provide an analytical framework for characterizing and assessing the suitability of future streamflow ensembles for water resources system planning. The metrics and classification developed advance and contribute to better planning in large river basins facing water supply shortages. Key Points: Many ensembles representing plausible future streamflow are available for the Colorado River BasinMetrics are presented to provide an evidence‐based framework for evaluating these streamflow ensemblesA classification approach was developed to group similar ensembles and assess their suitability for planning in different future scenarios [ABSTRACT FROM AUTHOR]

This paper presents a method to analyze and improve the set of equations constituting a rainfall‐runoff model structure based on a combination of a data assimilation algorithm and polynomial updates to the state equations. The method, which we have called "Data Assimilation Informed model Structure Improvement" (DAISI) is generic, modular, and demonstrated with an application to the GR2M model and 201 catchments in South‐East Australia. Our results show that the updated model generated with DAISI generally performed better for all metrics considered included Kling‐Gupta Efficiency, NSE on log transform flow and flow duration curve bias. In addition, the elasticity of modeled runoff to rainfall is higher in the updated model, which suggests that the structural changes could have a significant impact on climate change simulations. Finally, the DAISI diagnostic identified a reduced number of update configurations in the GR2M structure with distinct regional patterns in three sub‐regions of the modeling domain (Western Victoria, central region, and Northern New South Wales). These configurations correspond to specific polynomials of the state variables that could be used to improve equations in a revised model. Several potential improvements of DAISI are proposed including the use of additional observed variables such as actual evapotranspiration to better constrain internal model fluxes. Plain Language Summary: This paper presents a data‐driven method to improve rainfall‐runoff models used to generate future water resources scenario in climate change studies. The method, which we have called "Data Assimilation Informed model Structure Improvement" (DAISI) is generic, modular, and demonstrated with an application to monthly streamflow simulations over a large data set of catchments in South‐East Australia. Our results show that DAISI improves model performance for a wide range of metrics and increases the sensitivity of the model to climate inputs, which is critical in climate change scenarios. Finally, the improvements identified by DAISI take a simple mathematical form with distinct regional patterns in three sub‐regions of the study domain (Western Victoria, central region, and Northern New South Wales). Several improvements of DAISI are discussed including the inclusion of additional observed variables such as evapotranspiration to better constrain model simulations. Key Points: Data Assimilation Informed model Structure Improvement method diagnoses hydrological model structures by combining data assimilation with a polynomial update of state equationsThe method was applied to the GR2M rainfall‐runoff model with significantly improved streamflow simulations in 201 Australian catchmentsThe method identified updates to state equations with marked regional characteristics that could guide future improvement of GR2M [ABSTRACT FROM AUTHOR]

Previous studies have shown that regional deep learning (DL) models can improve runoff prediction by leveraging large hydrological datasets. However, training a DL regional model using all data without screening may degrade local performance. This study focuses on constructing enhanced local models through the utilization of cross‐basin data. To this end, we propose an approach that employs a novel training strategy to optimize DL model training for specific basins. The approach measures the impact of any one basin's gradient on the loss of the basin of interest, providing insights into the relationships between different basins. The approach was validated using 531 basins from the CAMELS dataset. Results suggest that local performance degradation is a common occurrence in regional models, and imbalanced data are likely to result in a specific pattern dominating the entire regional model. In comparison to a regional model simply trained with all basins, the median Nash‐Sutcliffe efficiency (NSE) for our models is 0.031 higher. In particular, the increase in NSE can exceed 0.2 for some dry basins. Our findings indicate that this novel DL strategy can significantly improve model performance in specific basins using large hydrological datasets, while mitigating local performance loss. Plain Language Summary: In the realm of deep learning, incorporating more data into the training process typically results in a more potent model. Conventionally, large datasets have been employed to train regional models with the intention of predicting rainfall‐runoff processes across all basins. However, such regional models often encounter performance degradation when applied to local basins. This degradation can be attributed to the extraction of overly general features, leading to a loss of specificity. In this paper, our objective is to harness information from a large dataset to establish a more robust local model. We have introduced a method that autonomously learns the similarities in rainfall‐runoff behavior among basins. Subsequently, we utilize this learned similarity to selectively choose data that proves advantageous for training the local model. Our results demonstrate that this strategy can significantly enhance runoff prediction, particularly in arid basins. Key Points: Training a DL regional model using all data without screening may degrade local performanceProper selection of training data is crucial for enhancing DL model training for individual basins, especially arid basinsData from different basins might act as mutual noise during the training process [ABSTRACT FROM AUTHOR]

Currently, climate change is considered as an important factor affecting nutrient loads introduced through riverine systems into the Baltic Sea. Although the prospect of a large increase in pollution has long seemed very real, it still does not translate into planning of effective remedial actions. One of the factors limiting the development of such activities is the scale of simulations, focusing generally on catchment outlet profiles. To fill this gap and enable a step forward in understanding responses toward future predictions in a higher resolution scale (subcatchment), we assessed nutrient load contribution using calculation profiles localized along a main watercourse and its tributaries. To track spatial and seasonal changes of total nitrogen and phosphorus for the Wełna River (central Poland), we used climate change data and the SWAT model. Having at our disposal a catchment model with a good performance we could follow not only total load changes in particular subcatchments, but also track localization of the pollution sources and their direct impact on load estimations. Our results showed an increase of the loads, especially from the agricultural land use type, up to 34% for TN and 85% for TP, in the most extreme scenario. Moreover, forest areas have been noted as highly reactive to climate change, and through their localisation are able to distinctly alter nutrient outflow. Finally, the contribution of urban areas should be further investigated since the dynamics of nitrogen and phosphorus release from impervious surfaces is noticeably different here than from the other diffuse sources. Plain Language Summary: This paper describes how climate change will affect the amount of nutrients in a small river catchment in the Baltic Sea region. While it is known that climate change can increase nutrient loads, effective actions to prevent them are still lacking. The "big picture" based on a whole catchment is still poor in terms of finding nutrient vulnerable areas. In this research, we looked at a more detailed scale to see where the nutrients are coming from and how they are changing over time. We used computer modeling to show that the amount of nutrients coming from agriculture, forests, and city areas will increase due to climate change. Overall, the amount of nitrogen can raise by 34% and the amount of phosphorus by 85%. Our results can be the basis for making decisions regarding actions aimed at improving the condition of surface waters and counteracting climate change effects. Key Points: Spatio‐temporal trends of nutrient loads have been tracked in a Baltic Sea region catchment under climate change scenariosClimate change suggests a potential load increase from the whole catchment by 34% for total nitrogen and 85% for total phosphorusOutputs from individual nutrient sources could grow by even 187% for total nitrogen and 302% for total phosphorus [ABSTRACT FROM AUTHOR]

Stochastic methods have been typically used for the design and operations of hydraulic infrastructure. They allow decision makers to evaluate existing or new infrastructure under different possible scenarios, giving them the flexibility and tools needed in decision making. In this paper, we present a novel stochastic streamflow simulation approach able to replicate both temporal and spatial dependencies from the original data in a multi‐site basin context. The proposed model is a multi‐site extension of the modified Fractional Gaussian Noise (mFGN) model which is well‐known to be efficient to maintain periodic correlation for several time lags, but presents shortcomings in preserving the spatial correlation. Our method, called Weighted‐mFGN (WmFGN), incorporates spatial dependency into streamflows simulated with mFGN by relying on the Cholesky decomposition of the spatial correlation matrix of the historical streamflow records. As the order in which the decomposition steps are performed (temporal then spatial, or vice‐versa) affects the performance in terms of preserving the temporal and spatial correlation, our method searches for an optimal convex combination of the resulting correlation matrices. The result is a Pareto‐curve that indicates the optimal weights of the convex combination depending on the importance given by the user to spatial and temporal correlations. The model is applied to a number of river basins in Chile, where the results show that the WmFGN approach maintains the qualities of the single‐site mFGN, while significantly improving spatial correlation. Key Points: Weighted modified Fractional Gaussian Noise model improves spatial correlations without significantly sacrificing temporal correlationThe method searches for an optimal convex combination of the spatial and temporal correlation matrices according on the user's priorityOur results in several Chilean basins demonstrate that Weighted‐modified Fractional Gaussian Noise (WmFGN) represents a significant improvement over the mFGN in preserving correlations [ABSTRACT FROM AUTHOR]

Future water availability is threatened by changes in both climate and water demand. Water rights with differing priorities are an important foundation of demand‐side tools (e.g., buyback, water pricing, and water market) to improve water use efficiency and reduce water scarcity, especially in highly regulated river systems. This paper assesses the impact of climate change on water yields from carryover storage with dual‐priority (high/low) water rights allocation systems using a simple and rapid analytical method. The method characterizes reservoir inflows using readily available flow characteristics (annual mean and Cv). We evaluate this method against a water resource simulation model in the Goulburn River basin, Australia. In general, our analytical "dual‐priority" Gould‐Dincer model reproduces water allocation estimates from the simulation model. We further demonstrate this method across 12 Australian catchments to investigate the climate change impact on "dual‐priority" water rights yield at the catchment scale. The hydrological projections show decreasing mean annual runoff and increasing annual runoff variability, except for some catchments in northern Australia. Water yield for high‐priority water rights (HPWRs) and low‐priority water rights (LPWRs) decreases for most catchments except for some catchments in northern Australia. South Dandalup in the 2070s (RCP8.5) shows the largest percentage decrease in HPWR and LPWR yield (about −53.53% and −56.81%, respectively). Our results show that changes in mean annual inflow have a more significant influence on water yield of HPWR and LPWR than Cv. Overall, the simple method provides a rapid assessment of water yields with "dual‐priority" water rights which is applicable across multiple sites at regional or even global scale. Key Points: A simple analytical method is used to evaluate the impact of climate change on the yield of "dual‐priority" water rightsWater yield for both high‐ and low‐priority water rights generally decreases under future climate in AustraliaFuture change in mean annual inflow has a more significant influence on water yield than change in annual flow variability [ABSTRACT FROM AUTHOR]

Optimization of spatially consistent parameter fields is believed to increase the robustness of parameter estimation and its transferability to ungauged basins. The current paper extends previous multi‐objective and transferability studies by exploring the value of both multi‐basin and spatial pattern calibration of distributed hydrologic models as compared to single‐basin and single‐objective model calibrations, with respect to tradeoffs, performance and transferability. The mesoscale Hydrological Model (mHM) is used across six large central European basins. Model simulations are evaluated against streamflow observations at the basin outlets and remotely sensed evapotranspiration patterns. Several model validation experiments are performed through combinations of single‐ (temporal evaluation through discharge) and multi‐objective (temporal and spatial evaluation through discharge and spatial evapotranspiration patterns) calibrations with holdout experiments saving alternating basins for model evaluation. The study shows that there are very minimal tradeoffs between spatial and temporal performance objectives and that a joint calibration of multiple basins using multiple objective functions provides the most robust estimations of parameter fields that perform better when transferred to ungauged basins. The study indicates that particularly the multi‐basin calibration approach is key for robust parametrizations, and that the addition of an objective function tailored for matching spatial patterns of ET fields alters the spatial parameter fields while significantly improving the spatial pattern performance without any tradeoffs with discharge performance. In light of model equifinality, the minimal tradeoff between spatial and temporal performance shows that adding spatial pattern evaluation to the traditional temporal evaluation of hydrological models can assist in identifying optimal parameter sets. Plain Language Summary: Hydrological models typically require local observations of river flow to calibrate the models and test their predictive capability. This limits the possibility for predictions in ungauged basins. This study used holdout tests to investigate the robustness of hydrological predictions for ungauged basins. Particularly we investigate how adding more basins and observed spatial patterns of evapotranspiration in the calibration of these models impact this robustness and transferability of model parameters to basins not used for calibration. Results show that transferability and spatial consistency of parameters increase when adding more basins and spatial pattern observations. Key Points: Clear improvements in simulated spatial patterns can be achieved with very limited tradeoff with discharge performanceMulti‐basin and spatial pattern‐based calibration improve parameter realism and transferability to ungauged basins [ABSTRACT FROM AUTHOR]