Hydrological Research Research Articles

Prediction of runoff at ungauged areas employing interpolation techniques and deep learning algorithm

Most river basins across the world are ungauged, and just a handful are gauged. As a result, predicting runoff in an unmeasured watershed is a difficult problem for the researchers. This research takes into account the tropical monsoon region, which is primarily covered by mountains and has a changing climate. This research is also carried out by creating a model with a machine learning technique that comprises Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM). The hybrid model considerably improves runoff forecast accuracy, with the CNN-LSTM model reaching an overall accuracy of 99.29 % across many datasets. The study uses 25 years of meteorological data from gauged stations to calculate runoff predictions for four ungauged sites: Katigora, Subhang, Sonai, and Morang. The findings highlight the necessity of combining machine learning and classical approaches to improve flood forecasting skills, which are critical for successful water resource management in flood-prone areas. This novel technique not only fills a vital vacuum in hydrological research, but it also has practical implications for catastrophe risk mitigation initiatives worldwide.

Imperiled Great Basin terminal lakes: Synthesizing ecological and hydrological science gaps and research needs for waterbird conservation

Imperiled Great Basin terminal lakes: Synthesizing ecological and hydrological science gaps and research needs for waterbird conservation

Comparison and Integrated Application for Runoff Simulation Models in Small and Medium-Sized River Basins of Southeast China Coastal Area

Runoff simulation is of fundamental importance for hydrological research. This study evaluated the applicability of multiple hydrological models and their ensembles for simulating runoff in small and medium-sized river basins of southeastern coastal China, focusing on the Xixi tributary of Jinjiang River and the Songxi and Chongyang tributaries of Minjiang River in Fujian Province. Four lumped hydrological models were selected for analysis: GR4J, IHACRES, TVGM, and MISDc-2L. The Bayesian model averaging method was utilized to compare the performance of each individual model and the multi-model ensemble in runoff simulation. Results: (1) For the calibration and validation periods of four hydrological stations, the mean values of KGE, NS, and R2 for the models GR4J, IHACRES, TVGM, and MISDc-2L were all above 0.7, and the mean values of |RE| were below 8.3%, without significant simulation accuracy variations when basin size changes, demonstrating strong regional applicability for runoff simulation; (2) The multi-model ensemble simulations using Bayesian model averaging of GR4J, TVGM, and MISDc-2L exhibited higher accuracy than individual models; (3) The MISDc-2L model demonstrated strong applicability in daily runoff simulations for both small and medium-sized river basins in Fujian Province and the large-sized Dongting Lake basin, showing that it is worthy of further application in other river basins across China. The findings of this study provide a reference for the selection and application of hydrological models for runoff simulation in small and medium-sized river basins of southeastern coastal China.

CAMELS-DE: hydro-meteorological time series and attributes for 1582 catchments in Germany

Abstract. Comprehensive large-sample hydrological datasets, particularly the CAMELS datasets (Catchment Attributes and MEteorology for Large-sample Studies), have advanced hydrological research and education in recent years. These datasets integrate extensive hydro-meteorological observations with landscape features, such as geology and land use, across numerous catchments within a national framework. They provide harmonised large-sample data for various purposes, such as assessing the impacts of climate change or testing hydrological models on a large number of catchments. Furthermore, these datasets are essential for the rapid progress of data-driven models in hydrology in recent years. Despite Germany's extensive hydro-meteorological measurement infrastructure, it has lacked a consistent, nationwide hydrological dataset, largely due to its decentralised management across different federal states. This fragmentation has hindered cross-state studies and made the preparation of hydrological data labour-intensive. The introduction of CAMELS-DE represents a step forward in bridging this gap. CAMELS-DE includes 1582 streamflow gauges with hydro-meteorological time series data covering up to 70 years (median length of 46 years and a minimum length of 10 years), from January 1951 to December 2020. It includes consistent catchment boundaries with areas ranging from 5 to 15 000 km2 along with detailed catchment attributes covering soil, land cover, hydrogeologic properties, and data on human influences. Furthermore, it includes a regionally trained long short-term memory (LSTM) network and a locally trained HBV (Hydrologiska Byråns Vattenbalansavdelning) model that were used as quality control and that can be used to fill gaps in discharge data or act as baseline models for the development and testing of new hydrological models. Given the large number of catchments, including numerous relatively small ones (636 catchments < 100 km2), and the time series length of up to 70 years (166 catchments with 70 years of discharge data), CAMELS-DE is one of the most comprehensive national CAMELS datasets available and offers new opportunities for research, particularly in studying long-term trends and runoff formation in small catchments and in analysing catchments with strong human influences. This article describes CAMELS-DE version 1.0, which is available at https://doi.org/10.5281/zenodo.13837553 (Dolich et al., 2024).

Combination of dynamic TOPMODEL and machine learning techniques to improve runoff prediction

AbstractTOPMODEL has been widely employed in hydrology research, undergoing continuous modifications to broaden its practical applicability and enhance its simulation accuracy. To encompass spatial discretization, diffusion‐wave characteristics, depth‐dependent flow velocity, and flux estimation in the unsaturated zone, a generalized dynamic TOPMODEL is developed by introducing a greater number of physical parameters. The present study aims to evaluate the optimal combination of these parameters within the dynamic TOPMODEL framework using machine learning techniques to improve the accuracy of runoff predictions and bolster the model's reliability. An innovative training method is suggested to elevate the model's performance by integrating the Long Short‐Term Memory (LSTM) algorithm and a topological classification, which relies on the evolving spatial distribution of runoff conditions during floods. The research findings show that the proposed methodology achieves the lowest mean relative error (MRE) at 0.106, the highest Pearson correlation coefficient (PC) at 0.938, and the highest coefficient of determination (R2) at 0.906 among the three dynamic TOPMODEL types adopted in this study. The effective implementation of a case study in a river basin showcases the feasibility of the proposed method in conjunction with dynamic TOPMODEL and underscores the importance of employing the suggested training procedure.

History of hydrological research in Slovakia and its links to water management

ABSTRACT The article outlines the development of water resources management and research in hydrology in Slovakia in the last century, when the centralized water resources planning provided a robust framework for the development of hydrological research. We first present facts about the organization of research in hydrology from 1948 to 1990, which dealt with various problems of practical relevance. Experimental research and its international outreach in the collaborative programmes of the Danube countries or the United Nations Educational, Scientific and Cultural Organization (UNESCO) water programmes are described. Then, the main achievements related to experimental hydrological research in two small research catchments with long histories of operation, the Mošteník Brook and Bystrianka Creek catchments, are presented. Short biographies of researchers who significantly contributed to the overall development of water management or hydrological research in small catchments in Slovakia are provided.

Two-step hybrid model for monthly runoff prediction utilizing integrated machine learning algorithms and dual signal decompositions

Runoff is pivotal in water resource management and ecological conservation. Current research predominantly emphasizes enhancing the precision of machine learning-based runoff predictions, with limited focus on their physical interpretability. This study introduces an innovative two-step hybrid runoff prediction framework tailored for the headwater region of the Yellow River Basin (YRB) to improve prediction accuracy and elucidate the runoff modeling process. The framework integrates machine learning techniques with dual signal decomposition approaches, incorporating diverse hydrometeorological and geographic indicators. Long Short-Term Memory (LSTM) and eXtreme Gradient Boosting (XGBoost) algorithms were employed to predict monthly runoff generation in sub-basins delineated by the Soil and Water Assessment Tool (SWAT), which were subsequently integrated using a Recurrent Neural Network (RNN) for monthly runoff concentration prediction. Results indicate that the proposed models delivered superior prediction performance compared to the SWAT model (R2 = 0.86, NSE = 0.85), with the LSTM-based two-step hybrid model (R2 = 0.90, NSE = 0.90) outperforming the XGBoost-based model (R2 = 0.89, NSE = 0.88). The dual decomposition method, integrating seasonal-trend decomposition based on loess (STL) and successive variational mode decomposition (SVMD), demonstrated exceptional efficacy in addressing the complexities of hydrometeorological time series. Models decomposed by STL-SVMD exhibited the highest average R2 and NSE values, as well as the lowest RMSE and MAE values in sub-basin runoff calculations. The low standard deviations of performance metrics further underscored the stability of these models across all sub-basins. This study demonstrates the efficacy of the proposed two-step hybrid model for simulating physical runoff processes in the headwater region of the YRB, providing valuable insights for regional hydrological cycle research and hydro-ecological security.

Regional calibration of the Hargreaves model for estimation of reference evapotranspiration

Abstract Estimation of reference evapotranspiration values is crucial in climatological and hydrological research, agricultural engineering, and irrigation design. The Penman-Monteith method, endorsed by the Food and Agriculture Organization (FAO) of the United Nations and numerous research studies, is widely regarded as the gold standard. However, its extensive data requirements limit its applicability in regions with sparse meteorological networks or limited measurement capabilities. The Hargreaves method, which requires only basic air temperature inputs, offers an alternative solution. The aims of this study were to calibrate the Hargreaves model for Central European climate conditions, considering altitudinal dependence, and to evaluate the temporal stability of the model parameters. In the first part of the research, we regionalized the Hargreaves coefficients using a curve-fitting method to ensure the best accuracy across 60 climatological stations in Slovakia. The regionalization of the Hargreaves coefficient improved accuracy by 10.1%, reducing the weighted absolute percentage error (WAPE) to 17.9%. However, our results showed that the accuracy of the modified Hargreaves model decreased with the increasing altitude of a climatological station. Incorporating altitude into the Hargreaves equation significantly improved model accuracy in stations at higher altitudes, providing a consistent level of accuracy across all climatological stations, regardless of their location and altitude. The results also indicated that the optimal model coefficient values change over time, showing a decreasing trend of –0.5 for the B coefficient and –0.1 for the C coefficient between the periods 1981–2000 and 2001–2020. Although regionalizing the Hargreaves model coefficients for local conditions can achieve good model performance, the model’s accuracy is not stable over time. Thus, periodic validation of the model is necessary for short-term applications.

The Performance of GPM IMERG Product Validated on Hourly Observations over Land Areas of Northern Hemisphere from Global Meteorological Stations

The integrated multi-satellite retrievals for the global precipitation measurement (IMERG) data, which is the latest generation of multi-satellite fusion inversion precipitation product provided by the Global Precipitation Measurement (GPM) mission, has been widely applied in hydrological research and applications. However, the quality of IMERG data needs to be validated, as this technology is essentially an indirect way to obtain precipitation information. This study evaluated the performance of IMERG final run (version 6.0) products from 2001 to 2020, using three sets of gauge-derived precipitation data obtained from the Integrated Surface Database, China Meteorological Administration, and U.S. Climate Reference Network. The results showed a basic consistency in the spatial pattern of annual precipitation total between IMERG data and gauge observations. The highest and lowest correlations between IMERG data and gauge observations were obtained in North Asia (0.373, p < 0.05) and Europe (0.308, p < 0.05), respectively. IMERG data could capture the bimodal structure of diurnal precipitation in South Asia but overestimates a small variation in North Asia. The disparity was attributed to the frequency overestimation but intensity underestimation in satellite inversion, since small raindrops may evaporate before arriving at the ground but can be identified by remote sensors. IMERG data also showed similar patterns of interannual precipitation variability to gauge observation, while overestimating the proportion of annual precipitation hours by 2.5% in North America, and 2.0% in North Asia. These findings deepen our understanding of the capabilities of the IMERG product to estimate precipitation at the hourly scale, and can be further applied to improve satellite precipitation retrieval.

Daily runoff prediction based on lightweight Mamba with partial normalization

ABSTRACT Runoff prediction serves as the cornerstone for the effective management, allocation, and utilization of water resources, playing a key role in hydrological research. This study employs a newly reported deep learning model, Mamba, to forecast river daily runoff and compared the proposed model with various benchmark methods, including statistical models, machine learning methods, recurrent neural networks, and attention-based models. Application of these models is implemented on three hydrological stations situated along the middle and lower reaches of the Mississippi River. Daily runoff from 1983 to 2023 were used to build the model for 7-day prediction. Findings demonstrate the superiority of the Mamba model over its counterparts, showcasing its potential as a backbone model. In response to the necessity for a more lightweight approach, a refined variant of the Mamba model is proposed, called LightMamba. LightMamba incorporates partial normalization and MPM (Multi-Path-Mamba) to enhance its efficacy in discerning nonlinear trends and capturing long-term dependencies within the streamflow data. Notably, LightMamba achieves commendable performance with an average NSE of 0.904, 0.907, and 0.900 on the three stations. This study introduces an innovative backbone model for time series forecasting, which offers a novel approach to hybrid modeling for future daily runoff prediction.

Synthesis of a hydrological, water chemistry, and contaminants research program in the Peace-Athabasca Delta (Canada) to inform long-term monitoring of shallow lakes

In a multistressor world, evidence-based stewardship of aquatic ecosystems requires long-term monitoring data to understand the timing and magnitude of environmental change and potential causes. At the Peace-Athabasca Delta (PAD; northeastern Alberta, Canada), concern for aquatic ecosystem degradation has triggered renewed and urgent calls by Indigenous, national, and international governance bodies for implementation of a long-term lake monitoring program capable of tracking changes in hydrological conditions and contaminant deposition attributable to major energy projects located upstream, climate change, and other unnatural and natural processes. Challenges imposed by the delta's size, hydrological complexity, inaccessibility of lakes, and other factors, however, have long impeded implementation of a delta-wide lake monitoring program. To address this pressing need, here we review and synthesize results obtained during 7 years (2015–2021) of intensive, multifaceted research at 60 shallow lakes spanning the delta's broad hydroecological gradients to inform an integrated hydrology, water chemistry, and contaminants monitoring program. The research involved systematic, repeated measurements of water isotope composition, water depth variation, water chemistry and turbidity, and metal(loid) concentrations in lake surface sediment and periphytic biofilm. Results reveal marked spatial and temporal variation of hydrological processes and their affects on lake water balance and depth, strong association between hydrological processes and lake water chemistry, and that concentrations of nickel and vanadium (key oil sands indicators) remain within the range of natural variation. Correspondence of generalized additive model trendlines for isotope-derived lake evaporation-to-inflow ratios and water chemistry with climate indices (Pacific Decadal Oscillation, Oceanic Niño Index) demonstrates the sensitivity, and predictability, of lake ecosystem processes in the delta to large-scale climatic patterns. We provide recommendations for field sampling, sample analysis, data display, and integration of information for ongoing monitoring at the PAD. These approaches are readily transferable to other complex landscapes with abundant shallow waterbodies threatened by multiple stressors that may alter hydrological regimes and contaminant delivery.

Hydrological Simulation and Parameter Optimization Based on the Distributed Xin’anjiang Model and the Particle Swarm Optimization Algorithm: A Case Study of Xunhe Watershed in Shandong, China

Hydrological models serve as essential tools in hydrological research, allowing us to address practical hydrological issues. This study focuses on the Xunhe Watershed in Shandong Province, China, constructing a distributed Xin’anjiang hydrological model. Furthermore, traditional manual calibration and automatic calibration using the Particle Swarm Optimization (PSO) algorithm were employed to determine model parameters, followed by hydrological simulations, with the aim of investigating the applicability of the distributed Xin’anjiang model in this watershed. The research findings indicate that the distributed Xin’anjiang model accurately simulates the hydrological processes in the Xunhe Watershed. There is a high level of agreement between the observed data and the simulated results, including key indicators such as peak discharge, runoff volume, and peak time. After optimizing the model parameters using the PSO algorithm, the distributed Xin’anjiang model demonstrates improved simulation performance in the Xunhe Watershed. During the calibration period, the mean relative peak discharge error (RPE) is 4.1%, the mean relative runoff error (RRE) is 4.34%, and the average Nash–Sutcliffe efficiency (NSE) for simulating the flood events is 0.89. During the validation period, the mean RPE is 3.82%, the mean RRE is 6.1%, and the average NSE for the process is 0.83. This indicates that the distributed Xin’anjiang model has good applicability in this watershed, providing a reliable reference for flood control and disaster reduction in the Xunhe Watershed.

Assessment and Application of Multi-Source Precipitation Products in Cold Regions Based on the Improved SWAT Model

In hydrological modeling, the accuracy of precipitation data and the reflection of the model’s physical mechanisms are crucial for accurately describing hydrological processes. Identifying reliable data sources and exploring reasonable hydrological evolution mechanisms for hydrology and water resources research in high-altitude mountainous regions with sparse stations and limited data constitute a significant challenge and focus in the field of hydrology. This study focuses on the Yarkant River Basin in Xinjiang, which originates from glaciers and contains a substantial amount of meltwater runoff. A dynamic glacier melt module considering the synergistic effects of multiple meteorological factors was developed and integrated into the original Soil and Water Assessment Tool (SWAT) model. Four precipitation datasets (ERA5-land, MSWEP, CMA V2.0, and CHM-PRE) were selected to train the model, including remote sensing precipitation products and station-interpolated precipitation data. The applicability of the improved SWAT model and precipitation datasets in the source region of the Yarkant River was evaluated and analyzed using statistical indicators, hydrological characteristic values, and watershed runoff simulation effectiveness. The optimal dataset was further used to analyze glacier evolution characteristics in the basin. The results revealed the following: (1) The improved model fills the gap in glacier runoff simulation with respect to the original SWAT model, with the simulation results more closely aligning with the actual runoff variation patterns in the study area, better describing the meltwater runoff process. (2) CMA V2.0 precipitation data has the best applicability in the study area. This is specifically reflected in the rationality of the spatial and temporal distribution patterns of the inverted precipitation, the accuracy observed in capturing precipitation events and actual precipitation characteristics, the goodness of fit in driving hydrological models, and the observed precision in reflecting the composition of watershed runoff, all of which are superior to those pertaining to other precipitation products. (3) The glacier melt calculated using the improved SWAT model informed by CMA V2.0 shows that during the study period, the basin formed a pattern with a positive–negative glacier balance demarcation at 36.5° N, featuring melting at higher latitudes and accumulation at lower latitudes. The results of this study are of significant importance for hydrometeorological applications and hydrological and water resources research in this region.

Understanding and simulating of three-dimensional subsurface hydrological partitioning in an alpine mountainous area, China

Critical zone (CZ) plays a vital role in sustaining biodiversity and humanity. However, flux quantification within CZ, particularly in terms of subsurface hydrological partitioning, remains a significant challenge. This study focused on quantifying subsurface hydrological partitioning, specifically in an alpine mountainous area, and highlighted the important role of lateral flow during this process. Precipitation was usually classified as two parts into the soil: increased soil water content (SWC) and lateral flow out of the soil pit. It was found that 65%–88% precipitation contributed to lateral flow. The second common partitioning class showed an increase in SWC caused by both precipitation and lateral flow into the soil pit. In this case, lateral flow contributed to the SWC increase ranging from 43% to 74%, which was notably larger than the SWC increase caused by precipitation. On alpine meadows, lateral flow from the soil pit occurred when the shallow soil was wetter than the field capacity. This result highlighted the need for three-dimensional simulation between soil layers in Earth system models (ESMs). During evapotranspiration process, significant differences were observed in the classification of subsurface hydrological partitioning among different vegetation types. Due to tangled and aggregated fine roots in the surface soil on alpine meadows, the majority of subsurface responses involved lateral flow, which provided 98%–100% of evapotranspiration (ET). On grassland, there was a high probability (0.87), which ET was entirely provided by lateral flow. The main reason for underestimating transpiration through soil water dynamics in previous research was the neglect of lateral root water uptake. Furthermore, there was a probability of 0.12, which ET was entirely provided by SWC decrease on grassland. In this case, there was a high probability (0.98) that soil water responses only occurred at layer 2 (10–20 cm), because grass roots mainly distributed in this soil layer, and grasses often used their deep roots for water uptake during ET. To improve the estimation of soil water dynamics and ET, we established a random forest (RF) model to simulate lateral flow and then corrected the community land model (CLM). RF model demonstrated good performance and led to significant improvements in CLM simulation. These findings enhance our understanding of subsurface hydrological partitioning and emphasize the importance of considering lateral flow in ESMs and hydrological research.

Causes for overestimation of the moisture recycling in an alpine meadow ecosystem of the Shule River Basin, Tibetan Plateau, China

Terrestrial moisture recycling is an essential hydrological component and a significant source of the atmosphere’s humidity budget in arid and semi-arid inland regions. Investigations on moisture recycling using the stable hydrogen and oxygen isotopes is currently a focal point in hydrologic research. However, the direct quantification of recycling ratio based on isotopic composition of evapotranspiration vapor is lacking. So, this causes challenges to compare studies, based in the same region and time period but with different methodologies. In this study, we measured the isotopic compositions of evapotranspiration vapor (δ18OET/δDET) from June to September 2018 in an alpine meadow ecosystem at the Shule River Basin by combining the Keeling plot model and in-situ chamber measurement. Using this data, the moisture recycled rate (mrr) was assessed based on the two-component (Model A) and three-component isotopic mixing models (Model B), respectively. Based on Model A and δ18OET, the analysis revealed that the mean recycled rate for the entire observation period was 35 %, while it was 26 % during the westerly period (the months dominated by the north branch of prevailing westerlies) and 44 % during the monsoon period (when subtropical moisture from the Indian monsoon penetrated the region). The recycled rate based on Model A and δDET was slightly larger with a mean value of 41 % during the entire observation period. The recycled rate based on Model B was also significantly higher in comparison with Model A, while the causes for this difference could be the assumption of substituting xylem water for transpiration vapor and the plant water source δD offset. The contribution of recycled moisture was notable lower for heavy rainfall comparing with light rainfall. Our findings provided a new perspective for the investigations of alpine meadow ecosystem hydrological processes.

Machine Learning-Based Soil Moisture Retrieval Using CYGNSS and Interpolated SMAP Data

Abstract. Soil moisture is a crucial component of the global terrestrial ecosystem water vapor cycle, and higher spatial-temporal soil moisture product has significant impacts on research in agriculture, hydrology, and ecology. Spaceborne Global Navigation Satellite System Reflectometry (GNSS-R) is a new remote sensing technology which can be used to retrieve soil moisture. Cyclone Global Navigation Satellite System (CYGNSS), launched by NASA in 2016 to observe ocean surface hurricanes, has also collected a substantial amount of data of GNSS signals reflected over the land. This paper focuses on soil moisture retrieval using CYGNSS and SMAP dataset. In this study, three machine learning methods (GSVM, BP and RF) are used for spatial interpolation of SMAP soil moisture data. Result shows RF interpolation is the best with the RMSE of 0.044 cm3/cm3 and the CC of 0.959 compared with SMAP truly SM. Then soil moisture data after RF interpolation and conventional linear interpolation are respectively used as response variables to train the retrieval model by ELM, BP and RF machine learning methods respectively. Results show that, among the six regression models, the RF-RF regression model performs the best with the RMSE of 0.012 cm3/cm3 and the CC of 0.986. Using this model, we get a global soil moisture product and have higher temporal and spatial resolution.

Effects of land use change on runoff depth in the Songnen Plain, China.

Climate change and human activities both have a considerable impact on runoff depth, which are important parts of a changing ecosystem. Nevertheless, the main focus of hydrological response research has been on investigating the impact of climate change on the depth of runoff. In contrast, there has been limited emphasis on comprehending the precise mechanisms through which changes in land use, in relation to human activities, influence runoff depth. This paper employs the MIKE SHE/MIKE 11 model to simulate surface runoff in the study area from 1980 to 2020, assesses the effects of climate change and land use change on runoff depth using the runoff reduction method, and quantifies the influence of land use change on runoff depth through a spatio-temporal geographically weighted regression model. This study indicates that during the past 40 years, the average runoff depth in the Songnen Plain was 36.26mm, exhibiting a tendency of 'increasing-decreasing-increasing'. The impact of climate change on surface runoff depth is more substantial than that of land use change. During the impact period 1, the runoff depth diminished by 19.07mm, with climate change contributing to a decrease of 15.89mm (83.31% contribution). In the impact period 2, the runoff depth increased by 7.49mm relative to the baseline period, with climate change leading to an increase of 12.73mm (70.84% contribution). Changes in various land types within the watershed can be used to observe the influence of human activities on runoff depth. More precisely, a 10% rise in the rate of change of construction land, dry land, and unoccupied land results in an increase in runoff depth of 6.21mm, 2.45mm, and 1.14mm, respectively. Conversely, a 10% rise in the rate of alteration of marsh, paddy, and forest land leads to a reduction in the depth of runoff by 9.49mm, 6.46mm, and 3.07mm, respectively. This research can contribute to improving the efficiency of water and land resource utilization and optimizing land resource governance.

TSAE-UNet: A Novel Network for Multi-Scene and Multi-Temporal Water Body Detection Based on Spatiotemporal Feature Extraction

The application of remote sensing technology in water body detection has become increasingly widespread, offering significant value for environmental monitoring, hydrological research, and disaster early warning. However, the existing methods face challenges in multi-scene and multi-temporal water body detection, including the diverse variations in water body shapes and sizes that complicate detection; the complexity of land cover types, which easily leads to false positives and missed detections; the high cost of acquiring high-resolution images, limiting long-term applications; and the lack of effective handling of multi-temporal data, making it difficult to capture the dynamic changes in water bodies. To address these challenges, this study proposes a novel network for multi-scene and multi-temporal water body detection based on spatiotemporal feature extraction, named TSAE-UNet. TSAE-UNet integrates convolutional neural networks (CNN), depthwise separable convolutions, ConvLSTM, and attention mechanisms, significantly improving the accuracy and robustness of water body detection by capturing multi-scale features and establishing long-term dependencies. The Otsu method was employed to quickly process Sentinel-1A and Sentinel-2 images, generating a high-quality training dataset. In the first experiment, five rectangular areas of approximately 37.5 km2 each were selected to validate the water body detection performance of the TSAE-UNet model across different scenes. The second experiment focused on Jining City, Shandong Province, China, analyzing the monthly water body changes from 2020 to 2022 and the quarterly changes in 2022. The experimental results demonstrate that TSAE-UNet excels in multi-scene and long-term water body detection, achieving a precision of 0.989, a recall of 0.983, an F1 score of 0.986, and an IoU of 0.974, significantly outperforming FCN, PSPNet, DeepLabV3+, ADCNN, and MECNet.

High resolution concentration-discharge relationships in managed watersheds: A 30+ year analysis

Concentration-discharge (C-Q) relationships provide insight into solute transport and biogeochemical processes for watersheds. A 30+ year, high-resolution dataset from the North Appalachian Experimental Watershed (NAEW) offers an unparalleled opportunity to explore land use and land management impacts on C-Q relationships for small watersheds of varying land management histories (agricultural to forested). The NAEW was among the few hydrologic research sites where storm event runoff was sampled using proportional sampling. This method captures the concentration-discharge behavior associated with land use more effectively than instantaneous sampling, which favors rising limb or falling limb dynamics. In this study, we explore C-Q relationships by investigating baseflow and storm event flow across their total behavior. We also build a systems-understanding by comparing chemostatic behavior to soil geochemistry and land use history. Highly managed agricultural watersheds with no associated stream baseflow demonstrate near-chemostatic behavior for most solutes, while mixed use and forested watersheds with associated streams are more mutable depending on whether primary sources of water were groundwater or surface water. Using this unique high-resolution dataset, we show that concentration-discharge relationships are influenced by soil and baseflow geochemistry, pore fluid concentration, and land type/land use legacy effects.

Ensemble Machine-Learning-Based Framework for Estimating Surface Soil Moisture Using Sentinel-1/2 Data: A Case Study of an Arid Oasis in China

Soil moisture (SM) is a critical parameter in Earth’s water cycle, significantly impacting hydrological, agricultural, and meteorological research fields. The challenge of estimating surface soil moisture from synthetic aperture radar (SAR) data is compounded by the influence of vegetation coverage. This study focuses on the Weigan River and Kuche River Delta Oasis in Xinjiang, employing high-resolution Sentinel-1 and Sentinel-2 images in conjunction with a modified Water Cloud Model (WCM) and the grayscale co-occurrence matrix (GLCM) for feature parameter extraction. A soil moisture inversion method based on stacked ensemble learning is proposed, which integrates random forest, CatBoost, and LightGBM. The findings underscore the feasibility of using multi-source remote sensing data for oasis moisture inversion in arid regions. However, soil moisture content estimates tend to be overestimated above 10% and underestimated below 5%. The CatBoost model achieved the highest accuracy (R2 = 0.827, RMSE = 0.014 g/g) using the top 16 feature parameter groups. Additionally, the R2 values for Stacking1 and Stacking2 models saw increases of 0.008 and 0.016, respectively. Thus, integrating multi-source remote sensing data with Stacking models offers valuable support and reference for large-scale estimation of surface soil moisture content in arid oasis areas.