Hydrological Sciences Journal

We use stable isotopes, radiocarbon, and geochemistry to determine the contribution of glacial melt to the watersheds surrounding the Coropuna ice cap in the Peruvian Andes. The Andes are deglaciating at an alarming rate and characterizing the glacial water contribution to these hydrological systems is an important step in building resilient and sustainable water use strategies in the Andes. Download at my website www.d18Olson.com/glaciallossintheandes/

With the requirement of a macroscopic approach to understanding the relationship between water resources and hydrological phenomena, such as severe droughts under climate change, this study investigates South Korea’s long-term terrestrial water storage (TWS) using GRACE satellite data. A detailed analysis of water balance in TWS, based on the GLDAS model outputs in Google Earth Engine, reveals the following results: (1) TWS anomaly shows an average decrease of −33.5 mm year−1 from 2003 to 2016; (2) spatial shifts in TWS anomaly (−1.176 to −0.424 cm) unveil regional water storage dynamics, indicating negative temporal slope changes per grid cell (−0.393 to −0.225); (3) contributions of precipitation to TWS are not always straightforward, due to runoff inefficiencies affecting water storage and groundwater; (4) accessible water, integrating surface water and groundwater linked only to the shallow layer’s soil moisture, constrains deep groundwater accounting, emphasizing the need for local groundwater surveys in further research.

Characterization of debris flow is critical to both risk assessment and hazard mitigation. Recent technologies enable on-site environmental monitoring sensors for geological disaster monitoring. However, the spatio-temporal understanding of debris flow in remote mountainous areas is still limited due to difficulties in observation networks and its complex driving conditions. Here we apply a coupled surface-subsurface watershed model to examine the characteristics of the water movements near the debris flow sites in Southwest China. Our approach captured the temporal dynamics of infiltration, redistribution of soil moisture, groundwater storage, and lateral groundwater fluxes. The lateral groundwater flux and groundwater storage were informative indicators in identifying debris flow location. Such informativeness was only effective when hourly dynamics were analyzed. Our findings provide new insight into quantifying the debris flow susceptibility. This study suggests that the coupled surface-subsurface hydrologic modeling approach can be informative for preliminary monitoring, planning and management problems of debris flow.

This study aims, as a first attempt, at establishing a framework for multi-objective optimization diagnostic model calibration and uncertainty analysis based on information theory, including mutual information (MI) and variation of information (VI). Moreover, a global sensitivity analysis was performed using the information coefficient of correlation to determine the incremental contribution of each model parameter. The applicability of the proposed framework was tested in an arid region of Oman with complex hydrogeological conditions and hardrock-alluvial aquifer systems. The findings highlight the capability of the information theory approach to simultaneously calibrate the model and assess uncertainty, while providing valuable diagnostics for identifying areas of the model that require further refinement. Furthermore, the results indicate that the proposed framework successfully reproduced 96% of the observed data, demonstrating its ability to reduce parameter uncertainty and ensure an accurate match between the simulated and observed data.

This study investigates event-based rainfall characteristics in Sinai (Egypt) using hourly precipitation data from the Global Satellite Mapping of Precipitation. A hierarchical cluster analysis of a 19-year dataset (2003-2021) identified five different regions in Sinai. Distinct storms were identified using a minimum inter-event time of 5 hours. The analysis of storm characteristics revealed that rainfall events in Sinai last from 1.7 to 3.6 hours, with a mean storm volume of 6.4 mm. Rainfall intensity ranges from 1.7 to 4 mm/hr, and the average dry period duration is 34 days. The northern region has the highest frequency of storms (25 events/year). The Weibull distribution was found to fit the best for all rainfall characteristics except for intensity, which was best represented by the Generalized Extreme Value distribution. This study provides valuable insights about rainfall events in Sinai that can be applied to improve flood mitigation strategies and water resources management.

Productive agricultural supply chains require the support of functional ecosystems, but intense agricultural practices change local hydrological systems (e.g. river diversion). In this study, the impact of farm dams was assessed for the Verlorenvlei catchment, a sensitive ecosystem currently under a state of hydrological change in South Africa. We developed a new module for the Jena Adaptable Modelling System (JAMS)/J2000 rainfall–runoff model to assess the streamflow impact from the points of abstraction, losses during storage and irrigation. The model achieved a satisfactory streamflow calibration with efficiencies Nash Sutcliffe Efficiency (NSE, logNSE) of 0.52 and 0.51. The irrigated area reduced simulated streamflow by 12 to 19%. The results from the study agree with remote sensed evapotranspiration, measured lake surface water levels and streamflow, but uncertainty remains in the total simulated dam evaporation. While many catchments lack the data required for a detailed irrigation impact assessment, this approach considers total water use, dam storage to area relationships and general farming practices.

Karst aquifers consist of complex networks of conduits in which groundwater flows and recharge/discharge processes are generally more dynamic than in other types of aquifers. Due to their intrinsic heterogeneity and anisotropy, monitoring, quantifying, and analysing natural responses of karst springs is an efficient tool. Unlike Cenozoic and Mesozoic rocks, in Neoproterozoic karst systems, groundwater circulates and stores generally in dissolution features known as tertiary porosity, as the rock's primary porosity is recrystallized, considered negligible. This article studies the hydrodynamics of a karst portion of the São Miguel River basin, southwest of the state of Minas Gerais, Brazil. The region is predominantly composed of Neoproterozoic carbonate rocks, dating from about 570 to 540 million years ago. During a hydrological year (2019–2020), three karst springs (S1, S2, and S3) were daily monitored through their natural responses (variations of electrical conductivity, EC, temperature, T, and discharge, Q) to rainfall episodes. The data were interpreted based on the analysis of spring hydrographs, time series, recession curves (seasonal and intra‐annual), and statistics of EC, T, and Q variations. The results show the three springs generally exhibit quick flow, typically karstic, in the case of hydrosystems with a well‐structured and functional underground drainage network. The time series indicate the hydrosystem drained by S1 presents slower circulation and a lower degree of linearity, resulting from the higher sinuosity of the system, while the hydrosystems of S2 and S3 have similar behaviours, of quick water circulations immediately after a rainy episode. The degrees of karstification classify S1 and S2 as complex and extensive karst systems consisting of several subsystems, and S3 as a system in which the conduit network is more developed at the upper epiphreatic zone than near the outlet.

Quantification of the greenhouse effect is a routine procedure in the framework of hydrological calculations of evaporation. According to the standard practice, this is made considering the water vapour in the atmosphere, without any reference to the concentration of carbon dioxide (CO2), which, however, in the last century has escalated from 300 to about 420 ppm. As the formulae used for the greenhouse effect quantification were introduced 50-90 years ago, we examine whether these are still representative or not, based on eight sets of observations, distributed in time across a century. We conclude that the observed increase of the atmospheric CO2 concentration has not altered, in a discernible manner, the greenhouse effect, which remains dominated by the quantity of water vapour in the atmosphere, and that the original formulae used in hydrological practice remain valid. Hence, there is no need for adaptation due to increased CO2 concentration.

Dams are an integral component of water resources systems and management in the Upper Ganga Basin (UGB). However, the impact of dam operations on the characteristics of hydrologic extremes and their regional spatial pattern is less explored. Based on observational datasets, we have studied the changes in the flood and drought characteristics at local and regional scales by comparing them during pre-and post-dam periods. It is observed that the annual frequency of streamflow is missing from the spectral representation of daily streamflow at the Tehri catchment during the post-dam period. The dam operation has reduced the flood peaks but increased its duration and volume, whereas the dam has influenced the variability in drought duration, intensity, and deficit. The operation of dams has changed the timing and strength of seasonality of these extremes. However, the dam has increased the risk of regional occurrences of droughts and minimized the risk of floods.

This study represented a new method based on three types of Deep Learning-based Models (DLM) for estimation of water parameters. The DLM models were recurrent neural networks (RNN), long short-term memory (LSTM), and bidirectional long short-term memory (BiLSTM). The study areas were the Colorado River basin in the United States and the Mighan Wetland in IRAN. The electrical conductivity (EC), dissolved oxygen (DO), total dissolved solids (TDS), chloride ion (Cl), and river flow rate (debi) were simulated by DLM models. The Wilson score uncertainty analysis (WS) results for Colorado modeling showed that LSTMdebi, RNNDO, and RNNEC were the certain models in simulating due to having the lowest errors (Mean ei equal to 0.36, -1.50, and -0.59), respectively. Finally, the highest value of the R2 index, equal to 0.998 was assigned to the LSTM model in modeling the debi parameter, and 0.996 in EC modeling in the Mighan wetland.

Addressing post-fire impacts largely depends on burn “severity.” A singular severity classification that encompasses the holistic effects of fire on all ecosystem processes does not currently exist. Lumping vegetation burn severity and soil burn severity into one metric, or using them interchangeably, can induce large inaccuracies and uncertainties in the intended ecosystem response to forcing. Often, burn “severity” reflects fire impacts on vegetation, which can be measured through remote sensing. Vegetation burn severity is likely more apropos for ecological research, whereas soil burn severity is more relevant for hydrological analyses. This paper reviews different remotely sensed vegetation severity products currently (mis)used for hydrological modeling, provides examples of when vegetation burn severity may (not) match soil burn severity, and summarizes the potential synergistic future of remote sensing with in situ severity metrics. While the focus in this paper is on the western United States, the lessons and principles apply universally.

Quantitative precipitation estimates obtained by weather radars are prone to errors. Gauge-based observations are known to be complementary data for mitigating radar-based estimation. This study investigates, implements, and validates four gauge adjustment and four time-independent bias correction methods over all the operating radars of Turkey during the years 2014–2019. The objective is to investigate the performance of methods over large regions using long time series, where such implementations are rarely done. The results provide detailed information regarding the performance of these methods in different spatiotemporal scenarios. Gauge adjustment methods can mitigate the mean error and/or the dispersion of the error in the original radar data. On average, gauge adjustment methods reduce the mean error from −0.81 to −0.05 mm/h, the root mean squared error from 2.63 to 1.50 mm/h, and the correlation coefficient from 0.53 to 0.83. Time-independent methods can improve the mean error from −0.81 to −0.08 mm/h.

Shallow crystalline groundwater of semi-arid hinterland of Ceará exhibits brackish or saline water with mixed-chloride or sodium-chloride facies. Very few hydrochemical data are available in the area and the drivers behind this salinity are not clearly identified. In this study, an extensive field data collection work was performed to provide new information about the hydrogeological functioning and the salinization processes, through the implementation of piezometric, hydrogeochemical, isotopic (18O, 2H) and multi-tracer dating (14C, 3H, CFC, SF6) monitoring. Piezometric and isotopic data evidence fast flow circulation processes and a high contribution of evaporated surface water to aquifer recharge. Multi-tracer dating shows that groundwater is essentially composed of seasonal vertical infiltration flows that mix with older waters stored in the aquifer. Chemical analyses suggest that groundwater, originally low mineralized, has become progressively saltier due to leaching of salts that were evapoconcentrated either in surface waters or the unsaturated zone during drier periods.

This study presented a novel paradigm for forecasting 12-step-ahead monthly precipitation at 126 California gauge stations. First, the satellite-based precipitation time series from CHIRPS, Terraclimate, ERA5, and PERSIANN-CDR products were bias-corrected using historical precipitation data. Four methods were tested, and Quantile mapping (QM) was the best. After pre-processing data, 19 machine-learning models were developed. Random Forest, GBOOST, extreme gradient boosting, support vector machine, multi-layer perceptron, and K-nearest-neighbors were chosen as the best models based on COPRAS measurement. After hyperparameter adjustment, the Bayesian backpropagation regularization algorithm fused the results. The superior models' predictions were considered inputs, and the target's initial step was labeled. The next 11 steps at each station followed this approach, and the fusion models accurately predicted all steps. The 12th step's average NSE, MSE, R2, and R were 0.937, 52.136, 0.880, and 0.869, respectively, demonstrating the framework's effectiveness at high forecasting horizons to help policymakers manage water resources.

Increasing temperatures in Antarctica have resulted in the enlargement of proglacial regions on the Antarctic Peninsula, following glacier melt. This melt has increased river activity yet direct runoff measurements remain scarce in Antarctica, despite it acting as a proxy for glacial ablation. Here, we present discharge and water temperature data from 2013 for three streams on Vega Island and discuss their relationship with air temperature. The average discharge at the largest stream was 0.523 m³s⁻¹ with a maximum of 5.510 m³s⁻¹; one of the highest recorded in Antarctica. The rivers continued to flow even when temperatures dropped to -7°C, indicating that a large proportion of the total runoff originated subglacially. This is supported by the one-day time lag between air and water temperatures. Using river discharge as a proxy, we measured 124.5 ±14.4 mm w.e. of ablation. This indirect measurement proved to be an effective tool to complement classic glaciological observations.

The state of Paraná is one of the largest corn producers in Brazil, cultivating in two growing seasons. This study evaluated bivariate correlations between precipitation and the second-crop corn yield between the first 10 days of April and the first 10 days of October using spatial statistics. The spatial analysis of yield was evaluated using geostatistical models and validated by cross-validation. Spatial analysis was performed using the local indicator of spatial association and a cross-semivariogram, and the co-dispersion coefficient was used to validate the correlation between variables. Cross-validation identified the exponential model as the best fit, indicating a spatial dependence of 317.26 km. Spatial associations between municipalities were significant throughout the state of Paraná, Brazil. The cross-semivariogram allowed fitting the wave, spherical, Gaussian, and exponential models, with a spatial correlation of 37.04 to 570.38 km between the precipitations and yield. Moreover, the co-dispersion coefficient validated the correlation between the two variables for all 10-day periods.

This paper proposes rainfall-runoff models based on machine learning to estimate daily streamflows in Oued Sebaou Watershed, a Mediterranean Coastal Basin located in northern Algeria. Therefore, we applied Random Forest (RF), Artificial Neural Networks (ANN) - under different training algorithms -, and Local Weighted Linear Regression (LWLR) using as input combinations of current and past amounts of rainfalls and previous values of streamflow. We selected streamflow and rainfall records to calibrate and validate the stated approaches. The study considered Root Mean Square Error (RMSE) and Correlation Coefficient (R) to evaluate the accuracy of the models. Analyses of the results show that RF provided the best outcomes for both training (RMSE = 4.7458 and R = 0.9834) and validation (RMSE = 2.3617 and R = 0.9719). The ANN calibrated with the Levenberg-Marquardt algorithm presented the second-best result, outperforming its counterparts and LWLR.

The Indian Summer Monsoon (ISM) characterizes the hydro-meteorological variability across the north Indian region and contributes more than 75% of the annual rainfall during the monsoon (July – September) season. In the present study, we analyzed the long-term monsoon rainfall for the Ganga River Basin to investigate its spatio-temporal variability. A statistically increasing (10 to 17 mm/y; p<0.05) trend is observed in ISM rainfall for the mountainous region since 1980, accompanied by increased temperature. We further note that high, very high, and extreme rainfall events are also increasing, enhancing the flash flood risk in the mountainous region. In contrast, the ISM rainfall in the alluvial region is observed to be statistically decreasing (-5 to -20 mm/y; p<0.05) with the combined influence of reduced vegetation. These findings provide valuable insights into the variations in the regional hydrology of the Ganga River Basin (GRBA) caused by natural and anthropogenic factors.

This study investigates the evolution characteristics of potential evapotranspiration (PET) over the Three-River Headwaters Region (TRHR) based on observations at 14 stations during 1960-2014. First, the spatial-temporal changes were analyzed at annual and seasonal scales, and the results indicated that: 1) the central part had the lowest annual mean PET; 2) the mean PET in spring and summer was much higher than that in autumn and winter; and 3) PET firstly decreased, and then increased from 1980s. Second, the monthly PET over the TRHR for the near future was predicted using BPNN and LSTM, and the predictions showed increasing trends. The major contribution of this study is to build the models for prediction of the monthly PET based on limited observations. Overall, the outcomes can help to better understand the future hydrometeorological conditions in the TRHR, which would be valuable for better protection of the ecological environment in this region.

In this study, Advanced Scatterometer (ASCAT) soil moisture data has been employed to compute the Basin Water Index (BWI) over six river basins of India for ten years (2007-2016). The BWI time series is assessed for the development of its relationship with observed streamflow. Further, a popular ensemble learning technique viz. Random Forest is employed to compute the ten-daily streamflow using the BWI time series. Moreover, the results are compared with the classical rainfall–runoff model forced with satellite-based precipitation and evapotranspiration, BWI-rainfall–runoff model, and GloFAS. The performance of the model is evaluated in terms of multiple efficiency measures viz., Nash-Sutcliffe Efficiency (NSE), correlation coefficient (R), and root mean square error (RMSE). The results reveal the BWI-rainfall–runoff model as the most accurate model for prediction of discharge. Out of six catchments, the performance of the BWI-rainfall–runoff model is very good over four catchments and good to satisfactory over the remaining catchments.

The variation of hydrological processes has been extensively discussed worldwide, yet little is known about the relative impact of human activities, and the precipitation–water level relationship in urbanized, watery areas. Thus, the change in water level and the influential variables are analysed for 1960–2014 in the urbanized and watery Taihu Basin, China. The results indicate that the water level displays a significant increasing trend. Furthermore, low-oscillation and high-oscillation periods were found to have occurred in the 1960s–1970s and 2000s, respectively, by the quantile perturbation method. A strong relationship was shown between water level and precipitation in the 1960s–1980s, especially in the flood season. Since then, human activities, such as land-use change, river system degradation and hydrological structures, have played distinct roles and caused more than 82% of the annual and flood-seasonal water level variation. The results may provide a more comprehensive understanding of the hydrological processes and provide a good reference for flood control.

Hydrological models require different inputs for the simulation of processes, among which precipitation is essential. For hydrological simulation, four different precipitation products – Asian Precipitation Highly Resolved Observational Data Integration Towards Evaluation of Water Resources (APHRODITE); European Centre for Medium-Range Weather Forecasts (ECMWF) Re-Analysis (ERA-Interim); Tropical Rainfall Measuring Mission (TRMM) Multi-satellite Precipitation Analysis (TMPA) real time (RT); and Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN) – are compared against ground-based datasets. The variable infiltration capacity (VIC) model was calibrated for the Sefidrood River Basin (SRB), Iran. APHRODITE and ERA-Interim gave better rainfall estimates at daily time scale than other products, with Nash-Sutcliffe efficiency (NSE) values of 0.79 and 0.63, and correlation coefficient (CC) of 0.91 and 0.82, respectively. At the monthly time scale, the CC between all rainfall datasets and ground observations is greater than 0.9, except for TMPA-RT. Hydrological assessment indicates that PERSIANN is the best rainfall dataset for capturing the streamflow and peak flows for the studied area (CC: 0.91, NSE: 0.80).

A hybrid hydrologic model (Distributed-Clark), which is a lumped conceptual and distributed feature model, was developed based on the combined concept of Clark’s unit hydrograph and its spatial decomposition methods, incorporating refined spatially variable flow dynamics to implement hydrologic simulation for spatially distributed rainfall-runoff flow. In Distributed-Clark, the Soil Conservation Service (SCS) curve number method is utilized to estimate spatially distributed runoff depth and a set of separated unit hydrographs is used for runoff routing to obtain a direct runoff flow hydrograph. Case studies (four watersheds in the central part of the USA) using spatially distributed (Thiessen polygon-based) rainfall data of storm events were used to evaluate the model performance. Results demonstrate relatively good fit to observed streamflow with a Nash-Sutcliffe efficiency (ENS) 0.84 and coefficient of determination (R²) 0.86 as well as better fit in comparison with outputs of spatially averaged rainfall data simulations for two models including HEC-HMS.