Pub Date : 2025-05-01Epub Date: 2025-03-05DOI: 10.1016/j.hydroa.2025.100201
Mousumi Ghosh , Aatish Anshuman , Mukesh Kumar
The field of geosciences is replete with problems where the target variable to be predicted is inherently class-imbalanced, meaning the events of interest are rare and infrequent. Examples include predicting landslides, ice jam breakups, preferential flow, and frozen ground. Such imbalance poses substantial challenges for modeling approaches. Using frozen ground prediction as a case study, this research examines how the frequency of event occurrence influences its prediction performance and proposes a data curation strategy to improve predictability. To this end, a data-driven approach utilizing a Long Short-Term Memory neural network is first implemented to predict soil temperature and determine frozen periods. Application of this approach at 25 gaging sites in Michigan reveals model underperformance, particularly at sites where the frozen data fraction (FDF) or the ratio of the frozen period to the total observation period, is low. The. study further demonstrates that under-sampling of more prevalent non-frozen period in training data improves detection of frozen periods. Greater improvements are experienced at sites with lower FDFs. However, performance peaks after a threshold FDF, plateauing or declining thereafter due to increased class imbalance and reduced training data length. The presented training data curation approach can be used for predictions of other class-imbalanced time series.
{"title":"Improving prediction of class-imbalanced time series through curation of training data: A case study of frozen ground prediction","authors":"Mousumi Ghosh , Aatish Anshuman , Mukesh Kumar","doi":"10.1016/j.hydroa.2025.100201","DOIUrl":"10.1016/j.hydroa.2025.100201","url":null,"abstract":"<div><div>The field of geosciences is replete with problems where the target variable to be predicted is inherently class-imbalanced, meaning the events of interest are rare and infrequent. Examples include predicting landslides, ice jam breakups, preferential flow, and frozen ground. Such imbalance poses substantial challenges for modeling approaches. Using frozen ground prediction as a case study, this research examines how the frequency of event occurrence influences its prediction performance and proposes a data curation strategy to improve predictability. To this end, a data-driven approach utilizing a Long Short-Term Memory neural network is first implemented to predict soil temperature and determine frozen periods. Application of this approach at 25 gaging sites in Michigan reveals model underperformance, particularly at sites where the frozen data fraction (FDF) or the ratio of the frozen period to the total observation period, is low. The. study further demonstrates that under-sampling of more prevalent non-frozen period in training data improves detection of frozen periods. Greater improvements are experienced at sites with lower FDFs. However, performance peaks after a threshold FDF, plateauing or declining thereafter due to increased class imbalance and reduced training data length. The presented training data curation approach can be used for predictions of other class-imbalanced time series.</div></div>","PeriodicalId":36948,"journal":{"name":"Journal of Hydrology X","volume":"27 ","pages":"Article 100201"},"PeriodicalIF":3.1,"publicationDate":"2025-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143550334","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-05-01Epub Date: 2025-06-01DOI: 10.1016/j.hydroa.2025.100204
Renato Amorim , Gabriele Villarini , Jeffrey Czajkowski , James Smith
During September 2024, Hurricane Helene devasted large areas of western North Carolina and eastern Tennessee, causing extensive loss of life and widespread damage due to heavy rainfall and extreme flooding. Despite the impacts of this storm, Helene’s heavy rainfall and resulting floods were not entirely unprecedented, as the region experienced several floods linked to tropical cyclones in the past, including multiple storms during the 2004 hurricane season. To make matters worse, this is an area with historically low market penetration by the National Flood Insurance Program, highlighting a strong asymmetry with respect to the coastal areas: while roughly 14% of the buildings in the eastern third of North Carolina were insured against floods, inland areas had less than a tenth of that coverage. Therefore, to improve resiliency and reduce the residual flood losses, it is critical to reconcile perceived versus actual flood risk and expand insurance coverage in hurricane-prone areas.
{"title":"Flooding from Hurricane Helene and associated impacts: A historical perspective","authors":"Renato Amorim , Gabriele Villarini , Jeffrey Czajkowski , James Smith","doi":"10.1016/j.hydroa.2025.100204","DOIUrl":"10.1016/j.hydroa.2025.100204","url":null,"abstract":"<div><div>During September 2024, Hurricane Helene devasted large areas of western North Carolina and eastern Tennessee, causing extensive loss of life and widespread damage due to heavy rainfall and extreme flooding. Despite the impacts of this storm, Helene’s heavy rainfall and resulting floods were not entirely unprecedented, as the region experienced several floods linked to tropical cyclones in the past, including multiple storms during the 2004 hurricane season. To make matters worse, this is an area with historically low market penetration by the National Flood Insurance Program, highlighting a strong asymmetry with respect to the coastal areas: while roughly 14% of the buildings in the eastern third of North Carolina were insured against floods, inland areas had less than a tenth of that coverage. Therefore, to improve resiliency and reduce the residual flood losses, it is critical to reconcile perceived versus actual flood risk and expand insurance coverage in hurricane-prone areas.</div></div>","PeriodicalId":36948,"journal":{"name":"Journal of Hydrology X","volume":"27 ","pages":"Article 100204"},"PeriodicalIF":3.1,"publicationDate":"2025-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144205547","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-05-01Epub Date: 2025-03-20DOI: 10.1016/j.hydroa.2025.100203
Ze Jiang, Ashish Sharma
Knowledge of impending drought can help significantly with water planning and management. This study introduces a novel forecasting framework for decadal drought projection which relies on climate model projections of Sea Surface Temperature Anomaly (SSTA) indices over the next decade and a spectral transformation methodology to maximise forecast skill. Decadal SSTA projections from the Decadal Climate Prediction Project (DCPP) undergo spectral transformation using Wavelet System Prediction (WASP). WASP modulates the frequency spectrum of predictor variables to better mimic the response spectrum of drought indices. The transformed SSTA indices are then used in a multiple linear regression (MLR) model to forecast drought indices across multiple time scales. This framework significantly improves drought forecasting skills, especially for lead times exceeding 24 months. While demonstrated for Australia, the MLR-WASP framework is transferable to other regions, offering a reliable tool for long-term water resource management by projecting drought risk over the coming decade. The implications of this research extend beyond hydroclimatology, impacting environmental science and engineering, sustainable planning, and adaptation efforts to climate change.
Plain language summary
Projecting drought risk over the next decade is essential for effective long-term water resources management. This study presents a new framework that reliably projects drought conditions up to 10 years ahead by optimizing decadal climate model data. It uses a spectral transformation technique to adjust predictors like Sea Surface Temperature Anomalies to better match drought patterns. These transformed predictors are then integrated into a regression model to forecast drought indices. When applied to Australia, this approach significantly outperformed existing methods, especially for 2-year forecasts. By combining advanced climate predictions with prediction-oriented data transformation, this framework enables reliable drought risk projections a decade out, offering invaluable insights for proactive planning in drought-prone regions worldwide.
{"title":"Decadal drought prediction via spectral transformation of projected Sea Surface Temperatures","authors":"Ze Jiang, Ashish Sharma","doi":"10.1016/j.hydroa.2025.100203","DOIUrl":"10.1016/j.hydroa.2025.100203","url":null,"abstract":"<div><div>Knowledge of impending drought can help significantly with water planning and management. This study introduces a novel forecasting framework for decadal drought projection which relies on climate model projections of Sea Surface Temperature Anomaly (SSTA) indices over the next decade and a spectral transformation methodology to maximise forecast skill. Decadal SSTA projections from the Decadal Climate Prediction Project (DCPP) undergo spectral transformation using Wavelet System Prediction (WASP). WASP modulates the frequency spectrum of predictor variables to better mimic the response spectrum of drought indices. The transformed SSTA indices are then used in a multiple linear regression (MLR) model to forecast drought indices across multiple time scales. This framework significantly improves drought forecasting skills, especially for lead times exceeding 24 months. While demonstrated for Australia, the MLR-WASP framework is transferable to other regions, offering a reliable tool for long-term water resource management by projecting drought risk over the coming decade. The implications of this research extend beyond hydroclimatology, impacting environmental science and engineering, sustainable planning, and adaptation efforts to climate change.</div></div><div><h3>Plain language summary</h3><div>Projecting drought risk over the next decade is essential for effective long-term water resources management. This study presents a new framework that reliably projects drought conditions up to 10 years ahead by optimizing decadal climate model data. It uses a spectral transformation technique to adjust predictors like Sea Surface Temperature Anomalies to better match drought patterns. These transformed predictors are then integrated into a regression model to forecast drought indices. When applied to Australia, this approach significantly outperformed existing methods, especially for 2-year forecasts. By combining advanced climate predictions with prediction-oriented data transformation, this framework enables reliable drought risk projections a decade out, offering invaluable insights for proactive planning in drought-prone regions worldwide.</div></div>","PeriodicalId":36948,"journal":{"name":"Journal of Hydrology X","volume":"27 ","pages":"Article 100203"},"PeriodicalIF":3.1,"publicationDate":"2025-05-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143703904","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01Epub Date: 2024-12-09DOI: 10.1016/j.hydroa.2024.100196
Ruhhee Tabbussum , Rajarshi Das Bhowmik , Pradeep Mujumdar
The natural variability in the occurrence of concurrent extremes of droughts and heatwaves is frequently attributed to climate change and anthropogenic causes, disregarding its association with large-scale global teleconnections. This study explores this association by demonstrating how concurrent droughts and heatwaves (CDHW) in India are temporally and spatially connected to multiple global teleconnections (referred to as climate variability modes). Composite and wavelet coherence analyses are implemented for the univariate evaluation of droughts and heatwaves—measured using the standardized precipitation index (SPI) and the standardized heat index (SHI), respectively—in relation to the climate variability modes. Furthermore, an attribution table framework is employed to examine the extremal dependence of concurrent heatwaves and droughts in India on the climate variability modes during 1951–2018. The results exhibit a higher probability of CDHW events when they are preceded by a large-scale global teleconnection. Overall, the insights drawn from this study suggest the possibility of relying on the climate variability modes to issue season-ahead forecasts of CDHW.
{"title":"Association of climate variability modes with concurrent droughts and heatwaves in India","authors":"Ruhhee Tabbussum , Rajarshi Das Bhowmik , Pradeep Mujumdar","doi":"10.1016/j.hydroa.2024.100196","DOIUrl":"10.1016/j.hydroa.2024.100196","url":null,"abstract":"<div><div>The natural variability in the occurrence of concurrent extremes of droughts and heatwaves is frequently attributed to climate change and anthropogenic causes, disregarding its association with large-scale global teleconnections. This study explores this association by demonstrating how concurrent droughts and heatwaves (CDHW) in India are temporally and spatially connected to multiple global teleconnections (referred to as climate variability modes). Composite and wavelet coherence analyses are implemented for the univariate evaluation of droughts and heatwaves—measured using the standardized precipitation index (SPI) and the standardized heat index (SHI), respectively—in relation to the climate variability modes. Furthermore, an attribution table framework is employed to examine the extremal dependence of concurrent heatwaves and droughts in India on the climate variability modes during 1951–2018. The results exhibit a higher probability of CDHW events when they are preceded by a large-scale global teleconnection. Overall, the insights drawn from this study suggest the possibility of relying on the climate variability modes to issue season-ahead forecasts of CDHW.</div></div>","PeriodicalId":36948,"journal":{"name":"Journal of Hydrology X","volume":"26 ","pages":"Article 100196"},"PeriodicalIF":3.1,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143103043","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01Epub Date: 2024-12-09DOI: 10.1016/j.hydroa.2024.100197
Akash Singh Raghuvanshi , Ricardo M. Trigo , Ankit Agarwal
This study enhances existing understanding of extreme precipitation spells induced by cloudburst-like (EPS-CBL) events in India, emphasizing climatology and geographical distribution often overlooked by traditional observations. EPS-CBL is defined as continuous rainfall exceeding 200 mm/day and intermittent extreme rates above 30 mm/hour or the 99.9th percentile threshold, differing from definitions proposed by the IMD and other studies. Our findings reveal significant biases in various precipitation products compared to IMD data. CMORPH consistently outperforms other datasets by capturing more extreme events and showing significant rising trends in regions influenced by orographic effects, such as the Himalayan foothills and the Western Ghats. Although IMERG aligns well with IMD overall, it exhibits variability in extreme events, while IMDAA tends to underestimate these extremes, especially in complex terrains. Analysis of EPS-CBL trends from 2000 to 2022 highlights regional differences across datasets. Both CMORPH and IMERG show an increase in EPS-CBL events in the hilly region, while IMDAA indicates a decline. Understanding EPS-CBL climatology provides valuable insights for modeling studies exploring the underlying mechanisms of these events.
{"title":"Climatology of extreme precipitation spells induced by cloudburst-like events during the Indian Summer Monsoon","authors":"Akash Singh Raghuvanshi , Ricardo M. Trigo , Ankit Agarwal","doi":"10.1016/j.hydroa.2024.100197","DOIUrl":"10.1016/j.hydroa.2024.100197","url":null,"abstract":"<div><div>This study enhances existing understanding of extreme precipitation spells induced by cloudburst-like (EPS-CBL) events in India, emphasizing climatology and geographical distribution often overlooked by traditional observations. EPS-CBL is defined as continuous rainfall exceeding 200 mm/day and intermittent extreme rates above 30 mm/hour or the 99.9th percentile threshold, differing from definitions proposed by the IMD and other studies. Our findings reveal significant biases in various precipitation products compared to IMD data. CMORPH consistently outperforms other datasets by capturing more extreme events and showing significant rising trends in regions influenced by orographic effects, such as the Himalayan foothills and the Western Ghats. Although IMERG aligns well with IMD overall, it exhibits variability in extreme events, while IMDAA tends to underestimate these extremes, especially in complex terrains. Analysis of EPS-CBL trends from 2000 to 2022 highlights regional differences across datasets. Both CMORPH and IMERG show an increase in EPS-CBL events in the hilly region, while IMDAA indicates a decline. Understanding EPS-CBL climatology provides valuable insights for modeling studies exploring the underlying mechanisms of these events.</div></div>","PeriodicalId":36948,"journal":{"name":"Journal of Hydrology X","volume":"26 ","pages":"Article 100197"},"PeriodicalIF":3.1,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143103044","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01Epub Date: 2024-12-12DOI: 10.1016/j.hydroa.2024.100199
Keegan E. Johnson, Paul C. Reneau, Matthew J. Komiskey
Using camera imagery to measure water level (camera-stage) is a well-researched area of study. Previous camera-stage studies have shown promising results when implementing this technology with tight constraints on test conditions. However, there is a need for a more comprehensive evaluation of the extensibility of camera-stage to practical applications. Therefore, the aim of this study was to test a camera-stage method under a wide variety of test conditions to better understand the successes and challenges of using this technology in real-world scenarios. In this study, this approach was tested during Water Year 2020 at three existing U.S. Geological Study (USGS) stream gaging stations in south central Wisconsin that had existing USGS water-level instrumentation. The specific reference objects tested were white pipes and a concrete wall. Since successful application of camera-stage relies on use of suitable images, all captured images in this study were visually inspected to determine suitability for application of camera-stage. Camera-stage measurements were then computed only on images deemed suitable and the results were compared with ground-truth stage values to determine the accuracy. For the purposes of this study, camera-stage values within ±0.10 ft of the actual stage were considered acceptable. One major challenge highlighted was the potential difficulty in obtaining suitable imagery, with the proportion of suitable images varying greatly between the four trials from 38 % to 92 %. The results from applying camera-stage to suitable images were encouraging though, with 79 % to 99 % of evaluated camera-stage values qualifying as acceptable among the four test trials.
{"title":"Practical application of time-lapse camera imagery to develop water-level data for three hydrologic monitoring sites in Wisconsin during water year 2020","authors":"Keegan E. Johnson, Paul C. Reneau, Matthew J. Komiskey","doi":"10.1016/j.hydroa.2024.100199","DOIUrl":"10.1016/j.hydroa.2024.100199","url":null,"abstract":"<div><div>Using camera imagery to measure water level (camera-stage) is a well-researched area of study. Previous camera-stage studies have shown promising results when implementing this technology with tight constraints on test conditions. However, there is a need for a more comprehensive evaluation of the extensibility of camera-stage to practical applications. Therefore, the aim of this study was to test a camera-stage method under a wide variety of test conditions to better understand the successes and challenges of using this technology in real-world scenarios. In this study, this approach was tested during Water Year 2020 at three existing U.S. Geological Study (USGS) stream gaging stations in south central Wisconsin that had existing USGS water-level instrumentation. The specific reference objects tested were white pipes and a concrete wall. Since successful application of camera-stage relies on use of suitable images, all captured images in this study were visually inspected to determine suitability for application of camera-stage. Camera-stage measurements were then computed only on images deemed suitable and the results were compared with ground-truth stage values to determine the accuracy. For the purposes of this study, camera-stage values within ±0.10 ft of the actual stage were considered acceptable. One major challenge highlighted was the potential difficulty in obtaining suitable imagery, with the proportion of suitable images varying greatly between the four trials from 38 % to 92 %. The results from applying camera-stage to suitable images were encouraging though, with 79 % to 99 % of evaluated camera-stage values qualifying as acceptable among the four test trials.</div></div>","PeriodicalId":36948,"journal":{"name":"Journal of Hydrology X","volume":"26 ","pages":"Article 100199"},"PeriodicalIF":3.1,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143103046","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01Epub Date: 2024-12-10DOI: 10.1016/j.hydroa.2024.100195
Vincent Lyne
The separation of streamflow into fastflow and slowflow components has been historically ambiguous, with existing separation methods like the Lyne-Hollick (LH) algorithm facing challenges due to subjective parameter choices. Here, we address this issue by developing the AutoVL algorithm which objectively and automatically partitions streamflow for no parameter input. AutoVL uses iterative statistical models, including a Signal Reconstructor for fastflow and an autoregressive moving-average (ARMA) model for slowflow, to estimate key hydrologic parameters. The algorithm couples the two models to iteratively estimate these parameters and to accurately separate streamflow. When applied to the Harvey River, Dingo Road station data, AutoVL identified significant seasonal and long-term variations in hydrologic parameters, reflecting the possible influence of climate change altering the temporal dynamics of catchment responses. The algorithm highlighted strongly coupled changes in infiltration and decay rates from altered streamflow patterns, offering a clearer understanding of streamflow responses to climate change. This performance suggests that AutoVL provides a more reliable, objective, efficient, and standard method for streamflow separation compared to previous approaches, enabling more accurate and confident hydrological modeling. By providing objective, dynamic insights into catchment behavior, AutoVL offers a promising tool for climate change studies and streamflow analysis.
{"title":"AutoVL: Automated streamflow separation for changing catchments and climate impact analysis","authors":"Vincent Lyne","doi":"10.1016/j.hydroa.2024.100195","DOIUrl":"10.1016/j.hydroa.2024.100195","url":null,"abstract":"<div><div>The separation of streamflow into fastflow and slowflow components has been historically ambiguous, with existing separation methods like the Lyne-Hollick (LH) algorithm facing challenges due to subjective parameter choices. Here, we address this issue by developing the AutoVL algorithm which objectively and automatically partitions streamflow for no parameter input. AutoVL uses iterative statistical models, including a Signal Reconstructor for fastflow and an autoregressive moving-average (ARMA) model for slowflow, to estimate key hydrologic parameters. The algorithm couples the two models to iteratively estimate these parameters and to accurately separate streamflow. When applied to the Harvey River, Dingo Road station data, AutoVL identified significant seasonal and long-term variations in hydrologic parameters, reflecting the possible influence of climate change altering the temporal dynamics of catchment responses. The algorithm highlighted strongly coupled changes in infiltration and decay rates from altered streamflow patterns, offering a clearer understanding of streamflow responses to climate change. This performance suggests that AutoVL provides a more reliable, objective, efficient, and standard method for streamflow separation compared to previous approaches, enabling more accurate and confident hydrological modeling. By providing objective, dynamic insights into catchment behavior, AutoVL offers a promising tool for climate change studies and streamflow analysis.</div></div>","PeriodicalId":36948,"journal":{"name":"Journal of Hydrology X","volume":"26 ","pages":"Article 100195"},"PeriodicalIF":3.1,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143103045","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Severe flood events are deemed more frequent in the near future with a changing climate. Headwater catchments, especially when karstified, exhibit a pronounced susceptibility to swift and substantial responses to precipitation events, leading to flooding. In this study, a karstified headwater catchment in SW Germany is investigated, focusing on gaining insights into the key processes controlling its discharge behavior. Intensive fieldwork was conducted and a variety of field data were collected and analyzed to determine the general system behavior during low flow and flood events. Field insights reveal a groundwater borne streamflow generation with a subsurface catchment largely differing from the surface catchment. Episodic and sporadic springs were identified as crucial contributors to stream flow generation.
The study was undertaken to evaluate the viability of simulating streamflow for flood warning using a lumped modeling approach at a sub-daily temporal scale, since lumped models are widely used for karst spring discharge modeling. Based on field data observations, a comparative analysis of different model structures was undertaken, aiming at assessing the required degree of model complexity for representing catchment runoff generation as well as the relevant system features and properties. In order to find an adequate model structure, a total of 21 models with varying degree of complexity were set up and run. Both, subsurface and surface catchment limits were considered. Results show that the hydrograph of the whole catchment can be represented by a rather simple lumped model in the present case under two prerequisites: (1) input needs to represent the groundwater catchment emphasizing the groundwater borne nature of flow and (2) the models need to allow for direct runoff, as the sporadic springs observed in the field contribute significant discharge to streamflow during flood events. It is revealed that it seems valid to start modeling with a relatively simple storage model as long as key processes in the catchment are represented. The general feasibility of such a simple modeling approach in this complex catchment encourages its feasibility in other headwater catchments.
{"title":"Effects of model complexity on karst catchment runoff modeling for flood warning systems","authors":"Paul Knöll , Ferry Schiperski , Antonia Roesrath , Traugott Scheytt","doi":"10.1016/j.hydroa.2024.100194","DOIUrl":"10.1016/j.hydroa.2024.100194","url":null,"abstract":"<div><div>Severe flood events are deemed more frequent in the near future with a changing climate. Headwater catchments, especially when karstified, exhibit a pronounced susceptibility to swift and substantial responses to precipitation events, leading to flooding. In this study, a karstified headwater catchment in SW Germany is investigated, focusing on gaining insights into the key processes controlling its discharge behavior. Intensive fieldwork was conducted and a variety of field data were collected and analyzed to determine the general system behavior during low flow and flood events. Field insights reveal a groundwater borne streamflow generation with a subsurface catchment largely differing from the surface catchment. Episodic and sporadic springs were identified as crucial contributors to stream flow generation.</div><div>The study was undertaken to evaluate the viability of simulating streamflow for flood warning using a lumped modeling approach at a sub-daily temporal scale, since lumped models are widely used for karst spring discharge modeling. Based on field data observations, a comparative analysis of different model structures was undertaken, aiming at assessing the required degree of model complexity for representing catchment runoff generation as well as the relevant system features and properties. In order to find an adequate model structure, a total of 21 models with varying degree of complexity were set up and run. Both, subsurface and surface catchment limits were considered. Results show that the hydrograph of the whole catchment can be represented by a rather simple lumped model in the present case under two prerequisites: (1) input needs to represent the groundwater catchment emphasizing the groundwater borne nature of flow and (2) the models need to allow for direct runoff, as the sporadic springs observed in the field contribute significant discharge to streamflow during flood events. It is revealed that it seems valid to start modeling with a relatively simple storage model as long as key processes in the catchment are represented. The general feasibility of such a simple modeling approach in this complex catchment encourages its feasibility in other headwater catchments.</div></div>","PeriodicalId":36948,"journal":{"name":"Journal of Hydrology X","volume":"26 ","pages":"Article 100194"},"PeriodicalIF":3.1,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142707302","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-01Epub Date: 2024-12-10DOI: 10.1016/j.hydroa.2024.100198
Abhinav Gupta , Sean A. McKenna
This study examines streamflow simulations using deep learning (DL) to understand the information extraction capability of global DL models trained on multiple watersheds. The study separately examined the entire streamflow time series and recession flow predictions. It introduces a global–local (GL) modeling strategy, where the global model outputs are fed as input to a locally trained model, with the hypothesis that the local model can leverage watershed-specific information that the global model may miss. The GL models demonstrate enhanced accuracy in recession flow prediction for 20-30% of the watersheds compared to the global and local models. However, considering the entire hydrograph, the GL models often perform worse than the global model. Further, the DL models were trained on two different objective functions. The performance of the global model in a watershed depended strongly upon the objective function used. These results suggest that the performance of global models is affected by watershed uniqueness, suggesting that even a global DL model should be tailored to individual watersheds for optimal performance.
{"title":"Hydrograph and recession flows simulations using deep learning: Watershed uniqueness and objective functions","authors":"Abhinav Gupta , Sean A. McKenna","doi":"10.1016/j.hydroa.2024.100198","DOIUrl":"10.1016/j.hydroa.2024.100198","url":null,"abstract":"<div><div>This study examines streamflow simulations using deep learning (DL) to understand the information extraction capability of global DL models trained on multiple watersheds. The study separately examined the entire streamflow time series and recession flow predictions. It introduces a global–local (GL) modeling strategy, where the global model outputs are fed as input to a locally trained model, with the hypothesis that the local model can leverage watershed-specific information that the global model may miss. The GL models demonstrate enhanced accuracy in recession flow prediction for 20-30% of the watersheds compared to the global and local models. However, considering the entire hydrograph, the GL models often perform worse than the global model. Further, the DL models were trained on two different objective functions. The performance of the global model in a watershed depended strongly upon the objective function used. These results suggest that the performance of global models is affected by watershed uniqueness, suggesting that even a global DL model should be tailored to individual watersheds for optimal performance.</div></div>","PeriodicalId":36948,"journal":{"name":"Journal of Hydrology X","volume":"26 ","pages":"Article 100198"},"PeriodicalIF":3.1,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143103047","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
<div><div>Our understanding of the impact of climate change on water availability and natural hazards in high-mountain regions is limited due to the spatial and temporal scarcity of ground observations of precipitation and snow. Freely available, satellite-based information about the snowpack is currently mainly limited to indirect measurements of snow-covered area or very coarse-scale snow water equivalent (SWE), but only for flat areas in lowlands without vegetation cover. Novel space-based laser altimeters, such as ICESat-2, have the potential to provide high-resolution snow depth data in worldwide mountain regions where no ground observations exist. However, these space-based laser altimeters come with spatial gaps between ground tracks, obtained without repetition at a give location. To overcome these drawbacks, here, we present a combined probabilistic data assimilation and deep learning approach to reconstruct spatio-temporal SWE from observations of snow depth along ground tracks, imitating ICESat-2 tracks in view of a potential future global application.</div><div>Our approach is based on assimilating SWE and snow cover information in a degree-day model with an iterative ensemble smoother (IES) which allows temporally reconstructing SWE along hypothetical ground tracks separated by 3 km. As input, the degree-day model uses daily precipitation and downscaled air temperature from the ERA5 reanalysis. A feedforward neural network (FNN) is then used for spatial propagation of the daily mean and standard deviation of the updated SWE ensemble members obtained from the IES. The combined IES-FNN approach provides uncertainty-aware spatio-temporally continuous estimates of SWE.</div><div>We tested our approach in the alpine Dischma valley (Switzerland) using high-resolution snow depth maps obtained from photogrammetric techniques mounted on airplanes and unmanned aerial system observations. Our results show that the IES-FNN model provides reliable estimates at a resolution of approximately 100 m. Even assimilating only one SWE observation during the year (combined with satellite-based melt-out date estimates) produces satisfying results when evaluating the IES-FNN SWE reconstructions on independent dates and smaller (<span><math><mrow><mo><</mo></mrow></math></span>4 km<sup>2</sup>) areas: mean absolute error of 86 mm (78 mm) at Schürlialp (Latschüelfurgga) for average SWE of 180 mm (254 mm), and average spatial linear correlation with the reference SWE of 0.51 (0.48). However, the assimilated SWE observation must not be too early in the accumulation season or too late in the melt season when the snowpack is starting or ending to accumulate or melt, respectively. Smaller distances between ground tracks (1500 m and 500 m) show improved performance of the IES-FNN approach in space, with no significant improvement in terms of temporal reconstruction.</div><div>Applying the IES-FNN approach to e.g., real ICESat-2 data, remains challenging due to t
由于缺乏对降水和积雪的时空地面观测，我们对气候变化对高山地区水资源供应和自然灾害的影响的了解十分有限。目前，基于卫星的免费积雪信息主要限于间接测量积雪覆盖面积或非常粗略的雪水当量（SWE），但仅限于低地无植被覆盖的平坦区域。新型天基激光测高仪（如 ICESat-2）有可能在没有地面观测数据的全球山区提供高分辨率雪深数据。然而，这些天基激光测高仪的地面轨迹之间存在空间差距，在特定地点获得的数据不重复。为了克服这些缺点，我们在此提出了一种概率数据同化和深度学习相结合的方法，以模仿 ICESat-2 的轨迹，根据沿地面轨迹的雪深观测数据重建时空 SWE，从而在未来实现潜在的全球应用。我们的方法基于将 SWE 和雪盖信息同化到一个度日模型中，并使用迭代集合平滑器（IES），从而可以沿相距 3 公里的假定地面轨迹重建 SWE。度日模型使用ERA5再分析的日降水量和降尺度气温作为输入。然后使用前馈神经网络（FNN）对从 IES 中获得的 SWE 更新集合成员的日平均值和标准偏差进行空间传播。我们使用安装在飞机上的摄影测量技术和无人机系统观测所获得的高分辨率雪深图，在瑞士高山迪施玛山谷测试了我们的方法。结果表明，IES-FNN 模型可在约 100 米的分辨率范围内提供可靠的估计值。在评估独立日期和较小（4 平方公里）区域的 IES-FNN SWE 重建时，即使只同化一年中的一次 SWE 观测（结合基于卫星的融化日期估计），也能得出令人满意的结果：平均 SWE 为 180 毫米（254 毫米）时，Schürlialp（Latschüelfurgga）的平均绝对误差为 86 毫米（78 毫米），与参考 SWE 的平均空间线性相关为 0.51（0.48）。不过，同化的 SWE 观测值不能在积雪开始或结束积雪或融化的季节过早或过晚进行。地面轨迹之间的距离越小（1500 米和 500 米），IES-FNN 方法的空间性能就越好，但在时间重建方面没有明显改善。将 IES-FNN 方法应用于 ICESat-2 等真实数据仍具有挑战性，因为这些数据的不确定性更高。不过，我们提出的方法仍有可能非常有助于解决高山地区降水和降雪地面观测资料匮乏的问题。
{"title":"A combined data assimilation and deep learning approach for continuous spatio-temporal SWE reconstruction from sparse ground tracks","authors":"Matteo Guidicelli , Kristoffer Aalstad , Désirée Treichler , Nadine Salzmann","doi":"10.1016/j.hydroa.2024.100190","DOIUrl":"10.1016/j.hydroa.2024.100190","url":null,"abstract":"<div><div>Our understanding of the impact of climate change on water availability and natural hazards in high-mountain regions is limited due to the spatial and temporal scarcity of ground observations of precipitation and snow. Freely available, satellite-based information about the snowpack is currently mainly limited to indirect measurements of snow-covered area or very coarse-scale snow water equivalent (SWE), but only for flat areas in lowlands without vegetation cover. Novel space-based laser altimeters, such as ICESat-2, have the potential to provide high-resolution snow depth data in worldwide mountain regions where no ground observations exist. However, these space-based laser altimeters come with spatial gaps between ground tracks, obtained without repetition at a give location. To overcome these drawbacks, here, we present a combined probabilistic data assimilation and deep learning approach to reconstruct spatio-temporal SWE from observations of snow depth along ground tracks, imitating ICESat-2 tracks in view of a potential future global application.</div><div>Our approach is based on assimilating SWE and snow cover information in a degree-day model with an iterative ensemble smoother (IES) which allows temporally reconstructing SWE along hypothetical ground tracks separated by 3 km. As input, the degree-day model uses daily precipitation and downscaled air temperature from the ERA5 reanalysis. A feedforward neural network (FNN) is then used for spatial propagation of the daily mean and standard deviation of the updated SWE ensemble members obtained from the IES. The combined IES-FNN approach provides uncertainty-aware spatio-temporally continuous estimates of SWE.</div><div>We tested our approach in the alpine Dischma valley (Switzerland) using high-resolution snow depth maps obtained from photogrammetric techniques mounted on airplanes and unmanned aerial system observations. Our results show that the IES-FNN model provides reliable estimates at a resolution of approximately 100 m. Even assimilating only one SWE observation during the year (combined with satellite-based melt-out date estimates) produces satisfying results when evaluating the IES-FNN SWE reconstructions on independent dates and smaller (<span><math><mrow><mo><</mo></mrow></math></span>4 km<sup>2</sup>) areas: mean absolute error of 86 mm (78 mm) at Schürlialp (Latschüelfurgga) for average SWE of 180 mm (254 mm), and average spatial linear correlation with the reference SWE of 0.51 (0.48). However, the assimilated SWE observation must not be too early in the accumulation season or too late in the melt season when the snowpack is starting or ending to accumulate or melt, respectively. Smaller distances between ground tracks (1500 m and 500 m) show improved performance of the IES-FNN approach in space, with no significant improvement in terms of temporal reconstruction.</div><div>Applying the IES-FNN approach to e.g., real ICESat-2 data, remains challenging due to t","PeriodicalId":36948,"journal":{"name":"Journal of Hydrology X","volume":"25 ","pages":"Article 100190"},"PeriodicalIF":3.1,"publicationDate":"2024-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142441862","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号