Water Resources Research最新文献_第2页

Modeling Incentive‐Driven Cooperation in Agricultural Water Use Under Unequal Withdrawal Condition 非等量取水条件下农业用水激励驱动合作模型

IF 5.4 1区地球科学

Water Resources Research Pub Date : 2026-10-01 DOI: 10.1029/2025wr042236

Jiheun Kim,Jeongwoo Hwang,Seung Beom Seo

{"title":"Modeling Incentive‐Driven Cooperation in Agricultural Water Use Under Unequal Withdrawal Condition","authors":"Jiheun Kim,Jeongwoo Hwang,Seung Beom Seo","doi":"10.1029/2025wr042236","DOIUrl":"https://doi.org/10.1029/2025wr042236","url":null,"abstract":"Abstract Water allocation has long been framed as a general optimization problem in which users compete for a finite resource, leading much of the existing literature to adopt a top‐down perspective on how water should be distributed among users. However, in practice, water withdrawals may also be influenced by physical access and operational arrangements, such as direct abstraction from river channels through intake facilities for agricultural and industrial use. In such systems, disparities may arise between institutional allocation principles and effective access to water during shortage conditions, potentially creating asymmetric impacts among upstream, midstream, and downstream stakeholders. This study quantifies these impact asymmetries across river locations and evaluates stage‐based intake restriction strategies designed to mitigate inequitable outcomes during droughts. The proposed framework is applied to the Yeongsan River Basin in southwestern Korea through computational experiments to demonstrate how drought‐stage‐based intake regulations can improve both net profit and equity among stakeholders. A game‐theoretic approach is further employed to assess the potential for compensation across Pareto‐optimal solutions by examining whether the additional gains generated under cooperation are sufficient to offset stakeholder losses relative to the non‐cooperative baseline. The analysis is extended to projected scenarios of intensified climatic and socio‐economic droughts to assess the robustness of the proposed policies under future stress conditions. Through this approach, the study introduces a generalized framework for identifying water allocation policies that reconcile efficiency, equity, and sustainability under institutional and hydrologic constraints.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"38 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149090821","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Diagnosing Latent Storage Representations in LSTM Reservoir Release Policies 诊断LSTM库释放策略中的潜在存储表示

IF 5.4 1区地球科学

Water Resources Research Pub Date : 2026-10-01 DOI: 10.1029/2025wr043135

Matthew Chen,Jonathan D. Herman

{"title":"Diagnosing Latent Storage Representations in LSTM Reservoir Release Policies","authors":"Matthew Chen,Jonathan D. Herman","doi":"10.1029/2025wr043135","DOIUrl":"https://doi.org/10.1029/2025wr043135","url":null,"abstract":"Abstract The influence of reservoirs on the water cycle introduces significant uncertainty for hydrologic prediction. The representation of reservoirs in hydrologic models ideally must be accurate, interpretable, and transferable across sites. Recent studies have highlighted the potential for data‐driven methods, including long short‐term memory (LSTM) networks, to accurately capture reservoir releases. However, the performance of LSTM models of reservoir releases has not yet been diagnosed on a large‐sample data set to understand whether their accuracy is physically justified. This study evaluates the ability of LSTMs to represent reservoir release policies across the continental U.S., leveraging the recently developed ResOpsUS data set. In particular, we focus on whether LSTMs can infer a latent storage representation without being provided observed storage as input. We find that in many cases the LSTM succeeds in encoding physically interpretable storage dynamics in its internal states, despite measurement errors in the data. However, the model accuracy is only weakly related to the strength of the learned storage representation; instead, it is log‐linearly related to the degree of regulation. In addition, LSTMs struggle to generalize storage dynamics and learned policies in space with pooled training. This study contributes to the growing literature on interpreting deep learning models of human‐hydrologic systems.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"11 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149210461","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Offshore Groundwater Salinization and Flushing in the Canterbury Bight During Glacial‐Interglacial Cycles 冰川-间冰期旋回期间坎特伯雷湾近海地下水盐渍化和冲刷

IF 5.4 1区地球科学

Water Resources Research Pub Date : 2026-10-01 DOI: 10.1029/2025wr043274

Chong Sheng,Christian Hensen,Aaron Micallef,Thomas Müller,Ariel T. Thomas,Mark Schmidt

{"title":"Offshore Groundwater Salinization and Flushing in the Canterbury Bight During Glacial‐Interglacial Cycles","authors":"Chong Sheng,Christian Hensen,Aaron Micallef,Thomas Müller,Ariel T. Thomas,Mark Schmidt","doi":"10.1029/2025wr043274","DOIUrl":"https://doi.org/10.1029/2025wr043274","url":null,"abstract":"Abstract Offshore freshened groundwater (OFG) is a significant yet underexplored component of global water resources, occurring widely along many continental margins. However, its dynamics remain poorly constrained due to limited offshore observations and the absence of robust evaluation models for unambiguous emplacement interpretation. To address this knowledge gap, we integrate existing porewater data and geological constraints from the Canterbury Bight with paleo‐hydrogeochemical modeling to investigate the origin, evolution, and biogeochemical processes related to OFG driven by sea‐level fluctuations. Our simulation results indicate that modern recharge from onshore aquifers contributes only a limited fraction of the present OFG. The majority of OFG has mean groundwater ages of less than ∼300 kyr, suggesting that it mainly originates from local meteoric recharge during sea‐level lowstands since the late Pleistocene. OFG volume and paleo‐submarine groundwater discharge along the continental shelf have varied periodically in response to sea‐level fluctuations throughout glacial‐interglacial cycles. The Canterbury Bight OFG system is currently in a non‐steady state, undergoing gradual salinization by downward intrusion from overlying seawater at an estimated rate of 0.06 cm yr −1 . The long‐term evolution of the OFG system has increased the complexity of biogeochemical processes within shelf sediments through freshwater–seawater mixing. By integrating paleo‐hydrogeological reconstruction, geological archives, and reactive transport modeling, this study provides a comprehensive framework for understanding OFG formation and evolution, advancing beyond previous models focused primarily on salinity.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"40 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149150592","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Spatiotemporal Variability of Water Transit Time Distribution Across an Alpine Stream Network 高山水系水运时间分布的时空变化特征

IF 5.4 1区地球科学

Water Resources Research Pub Date : 2026-10-01 DOI: 10.1029/2026wr044152

G. Grandi,P. Benettin,N. Durighetto,M. Peschiutta,M. Masiol,B. Stenni,G. Botter,E. Bertuzzo

{"title":"Spatiotemporal Variability of Water Transit Time Distribution Across an Alpine Stream Network","authors":"G. Grandi,P. Benettin,N. Durighetto,M. Peschiutta,M. Masiol,B. Stenni,G. Botter,E. Bertuzzo","doi":"10.1029/2026wr044152","DOIUrl":"https://doi.org/10.1029/2026wr044152","url":null,"abstract":"Abstract Water transit time (TT), the time elapsed from its entrance into a catchment and its exit through a stream section, is a useful indicator of catchment transport dynamics. Its distribution (TTD) is typically estimated at a single cross section where tracer data are available. However, spatial variability of TTD is key to understanding how heterogeneous source areas contribute to solute and water quality dynamics. Here, we investigate how TTD varies along a stream network. To this end, we analyzed stable isotopes in rainfall (three stations) and streamwater (eight sampling sites) in the Valfredda network draining an alpine catchment. To quantitatively analyze the data, we developed a spatially explicit model of TTD based on StorAge Selection (SAS) functions. The catchment is discretized into subcatchment units, each characterized by distinct SAS functions. Subcatchment TTDs are routed through the network to estimate TTDs at every cross section. Parameter estimation relies on a hierarchical Bayesian framework, which ensures consistency across subcatchments while allowing heterogeneity. Results indicate that streamflow preferentially samples older water from storage, with only a limited number of flow events triggering young water release. Despite pronounced heterogeneity in geomorphology and land cover, the inferred SAS functions exhibit only mild spatial variability. Differences in TTDs are driven primarily by variations in precipitation phase (liquid vs. solid) and the timing of snowmelt‐driven fluxes, rather than by intrinsic differences in mixing mechanisms. Both sources of variability are explicitly represented within the proposed framework, highlighting its general applicability to reveal transport dynamics at the catchment scale.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"57 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149150595","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Probable Maximum Precipitation Estimation Under Climate Change: Addressing Sampling and Bias Correction Uncertainties Using Large Ensemble Climate Simulations 气候变化下的可能最大降水估算：利用大集合气候模拟处理采样和偏差校正的不确定性

IF 5.4 1区地球科学

Water Resources Research Pub Date : 2026-10-01 DOI: 10.1029/2026wr043507

Md. Robiul Islam,Mohammad Reza Najafi,Alex J. Cannon

{"title":"Probable Maximum Precipitation Estimation Under Climate Change: Addressing Sampling and Bias Correction Uncertainties Using Large Ensemble Climate Simulations","authors":"Md. Robiul Islam,Mohammad Reza Najafi,Alex J. Cannon","doi":"10.1029/2026wr043507","DOIUrl":"https://doi.org/10.1029/2026wr043507","url":null,"abstract":"Abstract This study presents an integrated framework for estimating Probable Maximum Precipitation (PMP)‐equivalent return levels under climate change, addressing two critical challenges: sampling uncertainty for rare events and the validation of bias correction under different global warming scenarios. Leveraging large regional climate model ensembles, the framework combines spatial transposition, based on L‐moment homogeneity, with ensemble pooling to expand the effective sample size to nearly 10,000 independent annual maxima per location. Results show that spatial transposition significantly reduces uncertainty in rare‐event return levels, narrowing 95% confidence intervals by over 60% for the 10,000‐year event. Improvements were most notable in the upper tail, where empirical quantiles closely matched Generalized Extreme Value fits due to increased parameter stability from the expanded sample size. Subsequently, three bias correction techniques, CDF‐transform (CDFt), Quantile Delta Mapping (QDM), and Quantile Mapping (QM), were applied to CanRCM4‐LE and evaluated against CRCM5‐LE using a pseudo‐observation framework that allowed assessment under different warming scenarios. CDFt consistently provided the best agreement with distributional properties and empirical quantiles, followed by QDM. Using the best‐performing bias correction method applied to CRCM5‐LE with ERA5 reanalysis serving as the observational reference, PMP‐equivalent return levels at grid cells corresponding to 2,730 dam locations across Northeastern North America, show spatially coherent increases, with median 100,000‐year return levels rising by ∼57% under +1.5°C and ∼77% under +3.0°C warming. Broader spatial patterns mirror this intensification, underscoring the thermodynamic drivers of future extremes. The framework enables robust, reproducible PMP estimation and supports adaptive design strategies for critical infrastructure under climate uncertainty.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"24 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149090636","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Unveiling the Potential and Limitations of Large Observational Data Sets for the Calibration of Scaling Laws of Water Consumption 揭示大型观测数据集对用水量标度规律校正的潜力和局限性

IF 5.4 1区地球科学

Water Resources Research Pub Date : 2026-10-01 DOI: 10.1029/2025wr043200

A. Cozzolino,B. Sansone,R. Padulano,G. Del Giudice

{"title":"Unveiling the Potential and Limitations of Large Observational Data Sets for the Calibration of Scaling Laws of Water Consumption","authors":"A. Cozzolino,B. Sansone,R. Padulano,G. Del Giudice","doi":"10.1029/2025wr043200","DOIUrl":"https://doi.org/10.1029/2025wr043200","url":null,"abstract":"Abstract Scaling relationships for residential water consumption are crucial for transferring knowledge across spatial units and for developing predictive models that generalize beyond the observed data, key to the efficient and resilient management of urban water systems. However, calibration of their parameters is often limited by the size of the available observational data sets, which is usually small due to technological constraints, and by the violation of homogeneity assumptions underlying theoretical formulas. This research leverages a significantly large data set of observations collected at the morning peak hour in a real‐world community to provide reliable estimates of scaling laws, assessing the combined impacts of the Finite Population Effect (FPE) and cross‐correlation among users, and exploring the suitability of “artificial” data sets as surrogates of observations. Results show that scaling laws of variance and higher‐order moments are extremely sensitive to the dependence among users and to the distribution of sample moments across the population, with impacts difficult to disentangle from finite‐size sampling effects. The FPE mainly affects the estimate of uncertainty and the scaling behavior of correlation coefficients, with cascading effects on higher‐order moments, whereas average correlation among users directly impacts the scaling behavior of aggregated variance. Common empirical scaling formulas are tested, highlighting important limitations in their interpretation across broad aggregation ranges. Finally, limitations are highlighted regarding the use of surrogate data sets to infer the scaling behavior of a real‐world community.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"7 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149209438","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Assessing Hydrological Model Representation of Irrigation and Subsurface Flow in a Heavily Anthropogenic‐Influenced Catchment 在严重人为影响的集水区评估灌溉和地下流量的水文模型表示

IF 5.4 1区地球科学

Water Resources Research Pub Date : 2026-10-01 DOI: 10.1029/2026wr043528

Raul Mendoza,Willem van Verseveld,Frederiek Sperna Weiland,Anjali Bhagwat,Chris Seijger,Albrecht Weerts

{"title":"Assessing Hydrological Model Representation of Irrigation and Subsurface Flow in a Heavily Anthropogenic‐Influenced Catchment","authors":"Raul Mendoza,Willem van Verseveld,Frederiek Sperna Weiland,Anjali Bhagwat,Chris Seijger,Albrecht Weerts","doi":"10.1029/2026wr043528","DOIUrl":"https://doi.org/10.1029/2026wr043528","url":null,"abstract":"Abstract Evaluating water availability in catchments with widespread anthropogenic influences demands a joint analysis of natural hydrological processes and anthropogenic influences. Achieving this through hydrological modeling requires proper representation of relevant processes. Beyond the application of a single model set‐up, we assess the impact of different model representations of irrigation and subsurface flow dynamics by implementing three irrigation calendars based on remote sensing data sets and two subsurface flow model concepts in the distributed hydrological model wflow_sbm applied to the Hindon River Basin, part of the larger Ganges River Basin in India, featuring intensive irrigation, groundwater abstractions, and inter‐basin water transfer. A comprehensive inter‐comparison of the model output reveals the spatio‐temporal variability in simulated water balance partitioning. Irrigation calendar has a strong dynamic influence across the monsoon and non‐monsoon seasons, producing differences of 9–43 mm/year in river discharge and 32–147 mm/year in evapotranspiration. The impact of the model concept of subsurface flow on the magnitude of the simulated fluxes (differences of 81–90 mm/year in river discharge and 30–37 mm/year in evapotranspiration) is mostly consistent across the seasons but varies spatially, having the strongest influence in areas limited in groundwater storage and near the river. Model evaluation based on observed river discharge (KGE = 0.62–0.7) and groundwater level changes (MAE = 1.6–2.0 m) indicates varying performances. The study highlights the necessity of comprehensive assessment of water balance components and reduction of uncertainties in model estimates when utilizing remote sensing data sets to constrain irrigation timing and increasing the complexity of groundwater process model representation.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"484 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149092279","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

On the Adversarial Robustness of Hydrological Models 水文模型的对抗鲁棒性

IF 5.4 1区地球科学

Water Resources Research Pub Date : 2026-10-01 DOI: 10.1029/2026wr043726

Yang Yang,Joseph Janssen,Hoshin Gupta,Ting Fong May Chui

{"title":"On the Adversarial Robustness of Hydrological Models","authors":"Yang Yang,Joseph Janssen,Hoshin Gupta,Ting Fong May Chui","doi":"10.1029/2026wr043726","DOIUrl":"https://doi.org/10.1029/2026wr043726","url":null,"abstract":"Abstract The evaluation of hydrological models is essential for both model selection and reliability assessment. However, comparing predictions with observations alone is insufficient to fully characterize model behavior. This limitation is particularly relevant to deep learning (DL) models because of their complex internal structures. In risk‐averse operational settings, water managers require models whose reliability and failure modes have been carefully evaluated because unstable predictions can put critical infrastructure at risk. Motivated by the need to assess model reliability before operational deployment, we introduce and explore adversarial robustness analysis for hydrological modeling and evaluate whether small, targeted perturbations to meteorological forcings induce substantial changes in simulated discharge. We compare a physical‐conceptual model (HBV) with a deep learning‐based model (LSTM) across 1,347 German catchments using perturbations of varying magnitudes generated by the fast gradient sign method (FGSM). We find that FGSM perturbations generally reduce Kling–Gupta efficiency and increase mean squared error. However, widespread catastrophic failures under the specific FGSM perturbations are not observed, and LSTM generally exhibits less performance degradation than HBV. In the examined catchments, changes in the predicted hydrographs and time‐averaged internal model states often vary approximately linearly with perturbation magnitude, providing a compact summary of how errors grow under such perturbations. Similar patterns are observed along random perturbation directions in selected catchments, suggesting that, in those catchments, small input changes produce approximately proportional changes in model output. Overall, these findings characterize local predictive stability under controlled stress tests and motivate future studies of model robustness in real‐world operational settings.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"52 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149150589","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Estimating River Suspended Sediment Concentrations From Sentinel‐2: Expanding the Training Set and Applicability to Multi‐Dammed Systems 从哨兵- 2估算河流悬沙浓度：扩展训练集及其对多水坝系统的适用性

IF 5.4 1区地球科学

Water Resources Research Pub Date : 2026-10-01 DOI: 10.1029/2025wr041855

Ilan Valencius,Evan Dethier,Christopher Sheehan,Noah Snyder

{"title":"Estimating River Suspended Sediment Concentrations From Sentinel‐2: Expanding the Training Set and Applicability to Multi‐Dammed Systems","authors":"Ilan Valencius,Evan Dethier,Christopher Sheehan,Noah Snyder","doi":"10.1029/2025wr041855","DOIUrl":"https://doi.org/10.1029/2025wr041855","url":null,"abstract":"Abstract In situ measurements of riverine suspended sediment concentrations (SSC) are often sparse, limiting assessment of important patterns in space and changes in time. SSC measurements based on satellite data are a potential solution to this limitation, but training algorithms to estimate SSC from satellite data also requires substantial in situ data. This study presents new efforts to use Sentinel‐2 satellite data to infer SSC in large rivers (≥60 m width). Sentinel‐2 reflectance data has the benefit of a high relative frequency, with a nominal return period of 5 days, improving the likelihood that in situ SSC data was collected coincident with satellite overpass. We generated additional SSC training data using discharge—sediment flux rating curves and estimates derived from Landsat data using Dethier et al. (2020, https://doi.org/10.1029/2019jf005033 ) algorithms. The resulting algorithms were calibrated using 19,345 SSC observations, 23% of which were from this unconventional training data set. Despite the varied training base, these Sentinel‐2 algorithms can accurately predict suspended‐sediment concentrations (68% relative error). A case study on the heavily dammed Chattahoochee‐Apalachicola watershed accurately reproduced sparse in situ results and extended spatial and temporal coverage, highlighting the potential for improved understanding of river processes using Sentinel‐2‐based algorithms.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"53 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149210446","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

HydroDiffusion: Diffusion‐Based Probabilistic Streamflow Forecasting With a State Space Backbone 水扩散：基于扩散的概率流预测与状态空间主干

IF 5.4 1区地球科学

Water Resources Research Pub Date : 2026-10-01 DOI: 10.1029/2025wr043158

Yihan Wang,Annan Yu,Lujun Zhang,Charuleka Varadharajan,N. Benjamin Erichson

{"title":"HydroDiffusion: Diffusion‐Based Probabilistic Streamflow Forecasting With a State Space Backbone","authors":"Yihan Wang,Annan Yu,Lujun Zhang,Charuleka Varadharajan,N. Benjamin Erichson","doi":"10.1029/2025wr043158","DOIUrl":"https://doi.org/10.1029/2025wr043158","url":null,"abstract":"Abstract Recent advances have introduced diffusion models for probabilistic streamflow forecasting, demonstrating strong early flood‐warning skill. However, current implementations rely on Long Short‐Term Memory (LSTM) backbones and discrete‐time diffusion formulations. This leaves unresolved whether more advanced sequence‐modeling architectures and continuous‐time diffusion formulations can improve probabilistic streamflow forecasting. In this study, we developed HydroDiffusion, a diffusion‐based probabilistic forecasting framework integrating a decoder‐only state space model (SSM) backbone with a continuous‐time score‐based diffusion formulation. HydroDiffusion is evaluated across 531 watersheds in the contiguous United States (CONUS) in the CAMELS data set. We benchmark HydroDiffusion against two diffusion baselines with LSTM backbones, as well as the recently proposed Diffusion‐based Runoff Model (DRUM). Results show that HydroDiffusion achieves the strongest nowcast accuracy when driven by observed meteorological forcings. Moreover, HydroDiffusion delivers higher deterministic and probabilistic forecast skill than its LSTM‐based counterpart in operational forecasting. Attribution analysis further reveals that the SSM backbone excels in humid, densely vegetated, storage‐dominated watersheds with strong hydrologic memory, while its advantage becomes limited in arid regions and hydrologically complex systems such as karst and prairie pothole basins. These results establish HydroDiffusion as an effective generative modeling framework for medium‐range streamflow forecasting, providing both a new modeling benchmark and a foundation for future research on probabilistic hydrologic prediction at continental scales.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"4 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149090640","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0