Discharge Time Series Research Articles

A parsimonious model for springs discharge reconstruction and forecast for drought management: Lessons from a case study in Central Italy

Study regionKarst springs located in Central Apennine ridge (Central Italy), in the Tiber River basin. Study focusThe assessment of water availability is a key issue in a water supply system because of increasing drought and water scarcity events. Analysing and predicting the dynamic behaviour of groundwater resources is challenging to conceptualize and model, especially in poorly-monitored systems. A parsimonious model based on linear regression between the monthly spring discharge time series and Standardized Precipitation Index is proposed. The model is conceived for management purposes and suitable for users with a limited background in modelling techniques, who can take advantage from an initial knowledge of the aquifers hydrological regime. New hydrological insights for the regionThe model developed for long-term monitored springs is used to reconstruct the historical groundwater hydrographs and to make predictions for poorly-monitored springs with similar properties, exploiting the “similarity principle”. Results highlight the notable performance of this approach, which represents a useful tool for overcoming the limitations in spring discharge monitoring networks. Moreover, the tool is used to test forecast performance enabling water managers to develop a monthly early-warning system fostering a sustainable water resource exploitation and limiting the critical issues of the water supply system, especially during drought periods. Results are discussed from the perspective of the water utilities entrusted to manage their resources in the study region.

Forecasting the design maxima water discharges of floods on the Latorica river according to the data of the Mukachevo gauging station using plotting position formulas

This article presents the results of forecasting design maxima discharges on the Latorica River within Mukachevo town based on hydrological observation data at the “Mukachevo” gauging station using plotting position formulas. While solving the task, a novel non-parametric method of forecasting using observation data is applied. The method includes extrapolating the discrepancy (divergence, disagreement) between the estimates of the statistical annual probabilities of exceedance obtained by different plotting position formulas. The task is considered in the frame of the stationarity hypothesis of the maximum river flow employing a time series of maximal discharges of the Latorica River observed at the “Mukachevo” gauging station from 1947 to 1999.We involved the thirteen plotting position formulas. There was no specific criterion for choosing them to solve the task. All applied formulas were considered admissible options, and results obtained after using them – expert judgments reflecting decision-makers’ predisposition to more cautious or less expensive decision options in flood management strategies.The epistemic uncertainty of the different plotting positions was reduced by employing the Fishburn rule. According to this rule, the significance of various plotting positions was given by arranging their estimates in descending order of importance of their values under decision-making. Depending on the selected significance option assignment of the different plotting position formulas, such rank-weighted estimates of the design peak discharges (each of them for annual exceedance probability 1%, 0.5%, and 0.2%) were computed: (1) the rank-weighted upper bound estimate (sup-estimate) corresponding to the predisposition to more cautious decision options; (2) the rank-weighted lower bound estimate (inf-estimate) corresponding to the predisposition to less expensive decision options. As possible control theoretical alternatives for forecasting design maximal discharges considered were five parametric probability distributions: 1) the Kritskyi-Menkel three-parameter gamma distribution; 2) Pearson’s type III distribution; 3) the Extreme value type I distribution (Gumbell’s type I distribution); 4) the Logarithmic Pearson type III distribution; and 5) the Two-parameters logarithmic-normal distribution. The population statistical parameters for these parametric probability distributions were estimated from the sample statistics by the method of moments.

Assessing Future hydrological response of an urban watershed using machine learning based LULC forecasting models

Urbanization in terms of land-use/cover (LULC) change has a long-term significant impact on the hydrological cycle as the LULC is one of the most important influencing parameters to produce curve number (CN). The drastic change in LULC changes the CN. This change directly affects surface water including peak flows. This study aims to assess the change in surface runoff due to changes in LULC. Hydrological modeling is done for the consistent long-term behavioral study of surface runoff. The area focused on the study is the Kharun river, a tributary of the Mahanadi River. For the assessment of the impact of LULC change on the catchment discharge, a daily-step conceptual model soil and water assessment tool (SWAT) was applied. Landuse maps were prepared with the Landsat Thematic Mapper satellite images. The future land use was forecasted with the technique of spatial statistical modeling. The machine learning (ML) tool is used for quantifying the influences on the LULC change dynamics and producing the LULC map for 2024 and 2030. Remote sensing (RS) and geographic information system (GIS) analysis were coupled with hydrological SWAT modeling to investigate the connection between the LULC change and hydrologic regime. The SWAT Model’s calibration efficiency is verified by comparing the simulated and observed discharge time series at the Patharidih gauge and discharge station. The monthly and daily calibrations were quite satisfactory, with Nash-Sutcliffe an efficiency coefficient of 0.86 and 0.67. This modeling provides reliable information for sustainable management of available water resources of the catchment.

Suspended Load Estimation in Data Scarce Rivers

AbstractSediment rating curves (SRCs) are tools of satisfactory reliability in the attempt to describe the sediment regime in catchments with limited or poor-quality records. The study valorised the most suitable SRC development method for the estimation of the coarse suspended sediment load at the outlet of nine Mediterranean sub-watersheds. Four established grouping techniques were assessed, to minimize the uncertainty of the results, namely simple rating curve, different ratings for the dry and wet season of the year, hydrographic classification, and broken line interpolation, at three major Greek rivers (Aliakmon, Acheloos – upper route, Arachthos). The methods’ performance was benchmarked against sediment discharge field records, utilizing statistical measures and graphical analyses. The necessary observations were conducted by the Greek Public Power Corporation. The results were site/station dependent, and no methodology emerged as universally accepted. The analysis designated that the simple rating curve performs best at the cross-sections Moni Ilarion, Moni Prodromou, and Arta bridge, the different ratings for the dry and wet season of the year at Grevena bridge and Gogo bridge, the hydrographic classification at Velventos and Plaka bridge, and the broken line interpolation at Avlaki dam and Tsimovo bridge. In this regard, the study advocates the use of multiple SRC methods. Despite its limitations, the method merits a rather simple and cost-effective generation of a (continuous, detailed, sufficiently accurate) synthetic suspended sediment discharge timeseries, with high interpolating, extrapolating and reproducibility potential. The success of the application could benefit, among others, water quality restoration and dam management operations.

A Fully Connected Neural Network (FCNN) Model to Simulate Karst Spring Flowrates in the Umbria Region (Central Italy)

In the last decades, climate change has led to increasingly frequent drought events within the Mediterranean area, creating an urgent need of a more sustainable management of groundwater resources exploited for drinking and agricultural purposes. One of the most challenging issues is to provide reliable simulations and forecasts of karst spring discharges, whose reduced information, as well as the hydrological processes involving their feeding aquifers, is often a big issue for water service managers and researchers. In order to plan a sustainable water resource exploitation that could face future shortages, the groundwater availability should be assessed by continuously monitoring spring discharge during the hydrological year, using collected data to better understand the past behaviour and, possibly, forecast the future one in case of severe droughts. The aim of this paper is to understand the factors that govern different spring discharge patterns according to rainfall inputs and to present a model, based on artificial neural network (ANN) data training and cross-correlation analyses, to evaluate the discharge of some karst spring in the Umbria region (Central Italy). The model used is a fully connected neural network (FCNN) and has been used both for filling gaps in the spring discharge time series and for simulating the response of six springs to rainfall seasonal patterns from a 20-year continuous daily record, collected and provided by the Regional Environmental Protection Agency (ARPA) of the Umbria region.

Adjoint-based sensitivity analysis and assimilation of multi-source data for the inference of spatio-temporal parameters in a 2D urban flood hydraulic model

This contribution presents a novel approach for the calibration of distributed parameters in a 2D urban flood hydraulic model. It focuses on the challenging issue of inferring distributed friction parameters from multi-source heterogeneous spatio-temporal observations of their hydraulic signatures in the context of an urban flash flood in a complex street network. A variational data assimilation algorithm is used to infer high-dimensional multi-variate parameters (spatialized friction and inflow discharge time series) using multi-source observations. This method relies on a differentiable 2D shallow water hydraulic model which enables to generate high-resolution sensitivity maps of local gradients and Derivative-based Global Sensitivity Measures (DGSM), enabling to guide adequate definition of parameter spatialization for the data assimilation process. Assimilated data include real local limnigraphic measurements and high-water marks collected after a major flood event, as well as modeled flow velocity used in twin experiments setups. This study is the first to leverage high-water marks with a variational method for the calibration of distributed parameters in an urban flood model. The multi-source data is used to infer inflow hydrographs and distributed friction parameters in setups of varying complexity. In the main setup, the complex structure of the street network, along with the sensitivity maps and hydraulics expertise led to define a model configuration with 45 friction patches. A high-dimensional parameter vector composed of these friction values and an upstream inflow is inferred simultaneously by assimilating real limnigraphic data and high-water marks. This leads to an increase in model fit to observations and satisfying parameter estimates.

Forecasting of compound ocean-fluvial floods using machine learning

Flood modelling and forecasting can enhance our understanding of flood mechanisms and facilitate effective management of flood risk. Conventional flood hazard and risk assessments usually consider one driver at a time, whether it is ocean, fluvial or pluvial, without considering the compound nature of flood events. In this paper, we developed a novel approach for modelling and forecasting compound coastal-fluvial floods using a two-step framework. In step one, a hydrodynamic model is used to simulate floodwater propagation; while in step two, machine learning (ML) models are used to generate flood forecasts. The architecture of hydrodynamic-ML forecasting system incorporates a hydrodynamic model covering a specific domain, with individual ML models trained for each pixel. In total 7 ML models including: Support Vector Regression (SVR), Support Vector Machine (SVM), Radial Basis Function (RBF), Linear Regression (LR), Gaussian Process Regression (GPR), Decision Tree (DT), and Artificial Neural Network (ANN) were applied in this study. Forecasting compound floods is achieved using two sets of inputs: timeseries of river discharges in the upstream fluvial section and downstream ocean water levels in the coastal areas. The accuracy of the flood forecasting system is demonstrated for Cork City, Ireland; and modelling performance was evaluated using several statistical tools. Results show that the proposed models can provide reliable estimates of flood inundation and associated water depths. Overall, the RBF model exhibits the best performance. Despite the complexity of compound multi-driver floods, this study shows that the coupled hydrodynamic-ML approach can forecast coastal-fluvial flood with limited hydraulic and hydrological input data. This system overcomes the limitations of traditional hydrodynamic model-based systems where trade-offs between the always competing numerical model accuracy and computational time prohibit the model to be used for short-term flood forecasting. Once trained, the ML component of the coupled system can perform flood forecasting in near real-time, potentially integrating into a flood early warning system. Accurate flood forecasting has a wide range of positive societal impacts, including improved flood preparedness, increased confidence, better resource allocation, reduced flood damage, and potentially even flood prevention.

Exploring Drivers of Historic Mercury Trends in Beluga Whales Using an Ecosystem Modeling Approach.

While mercury occurs naturally in the environment, human activity has significantly disturbed its biogeochemical cycle. Inorganic mercury entering aquatic systems can be transformed into methylmercury, a strong neurotoxicant that builds up in organisms and affects ecosystem and public health. In the Arctic, top predators such as beluga whales, an ecologically and culturally significant species for many Inuit communities, can contain high concentrations of methylmercury. Historical mercury concentrations in beluga in the western Canadian Arctic's Beaufort Sea cannot be explained by mercury emission trends alone; in addition, they could potentially be driven by climate change impacts, such as rising temperatures and sea ice melt. These changes can affect mercury bioaccumulation through different pathways, including ecological and mercury transport processes. In this study, we explore key drivers of mercury bioaccumulation in the Beaufort Sea beluga population using Ecopath with Ecosim, an ecosystem modeling approach, and scenarios of environmental change informed by Western Science and Inuvialuit Knowledge. Comparing the effect of historical sea ice cover, sea surface temperature, and freshwater discharge time series, modeling suggests that the timing of historical increases and decreases in beluga methylmercury concentrations can be better explained by the resulting changes to ecosystem productivity rather than by those to mercury inputs and that all three environmental drivers could partially explain the decrease in mercury concentrations in beluga after the mid-1990s. This work highlights the value of multiple knowledge systems and exploratory modeling methods in understanding environmental change and contaminant cycling. Future work building on this research could inform climate change adaptation efforts and inform management decisions in the region.

BaHys—A Bayesian Modeling Framework for Long‐Term Concentration‐Discharge Hysteresis: A Case Study on Chloride

AbstractImproving river water quality requires a thorough understanding of the relationship between constituent concentration and water discharge during runoff events (i.e., C‐Q hysteresis), which may be strongly non‐linear. Analysis of C‐Q hysteresis on large temporal scales provides unprecedented insights into event dynamics and long‐term concentration trends in surface and groundwater. Despite the increasing availability of time series data on water quality, there are still limited quantitative modeling frameworks that enable this analysis. Here, we combine Bayesian modeling and an existing mass balance to model long‐term C‐Q hysteresis dynamics in multi‐decade constituent concentration and water discharge time series. We focus on the case study of chloride and demonstrate that our model can simultaneously characterize the size and rotation of C‐Q hysteresis, and diffuse and low‐flow inputs to constituent concentration using only time‐series data from the river Rhine. Over 28 years of monitoring, we find that chloride exhibits a dominant clockwise dilution behavior that does not vary considerably under different hydro‐climatic conditions, hinting to similar mobilization mechanisms over time. We also show decreasing chloride concentrations in surface and groundwater, due to the cessation of mining activities in the Rhine. Our approach uses uncertainty estimates to show the range within which model parameter values lie, aiding decision‐makers in a robust assessment of river water quality. We conclude that Bayesian modeling of C‐Q hysteresis provides a powerful framework for investigating long‐term contamination dynamics that can be extended to several constituents to find factors controlling their export, ultimately suggesting mitigation measures for river contamination.

Staged SWAT calibration with bias-corrected precipitation product for enhancing flow data continuity in tributaries of the Mekong River

Accurate and continuous flow data are crucial for effective water management in large river basins worldwide. However, these catchments often face challenges regarding data continuity in the mainstream and their tributaries. This study proposes a methodological framework for enhancing flow data continuity that uses a staged calibration scheme of Soil and Water Assessment Tool (SWAT) modeling with an appropriate satellite precipitation product (SPP) for each watershed. This framework is successfully applied in the Mun-Chi River Basin, the largest tributary of the Lower Mekong River, over a 20-year period at 34 stations. The staged calibration involves partitioning the catchment into smaller sub-catchments, guided by the flow monitoring stations and flow discontinuity structures such as dams. The sequential calibration from upstream to downstream partitions the overall model calibration challenge into smaller problems and contributes to a more efficient and accurate calibration and validation. Alternative SPPs were considered to overcome monitoring gaps. Their bias was corrected using quantile mapping, and their performance was evaluated with flow simulations using the SWAT model. The assessment indicates that the CMORPH-CRT product, with a spatial resolution of 0.25°, demonstrates good suitability for hydrological modeling of the Mun-Chi River Basin. The proposed methodological framework provides a continuous time series of flow discharge at multiple stations within the watershed, offering valuable insights for sustainable water resource management strategies in river systems under changing climate and land use conditions, and supporting future studies on environmental issues.

Groundwater Recharge Assessment in Central Benin: The Case of the Collines Region (West Africa)

The objective of this study was to assess groundwater recharge in the hard-rock central region of Benin so as to compare it with the water needs of the local population. To reach this objective, we applied the Water Table Fluctuation (WTF) method, which requires long-term monitoring of groundwater level fluctuations. Groundwater level time series were used in combination with other data (including time series of surface water discharge and rainfall) to estimate groundwater recharge but also to shed further light on the relationship between surface water and groundwater. The results demonstrated that the minimum inter-annual groundwater recharge amount is about 1.09 × 109 m3, which is enough to cover the basic water needs of the local population. It should be highlighted that in sub-regions where the density of the population is high, water shortage can still occur with the above estimated groundwater recharge amount. This study has also illustrated that when applying the WTF method, sites with a highly uncertain specific yield can be detected.

Pulsed plasma thruster exhaust reconstruction

The transient exhaust of an ablative pulsed plasma thruster is characterized experimentally for three capacitances and three discharge voltages, for discharge energies below 11 J. A novel analysis technique is introduced, which reconstructs the exhaust as a superposition of different Maxwellian-like ion groups. Each ion group is characterized by its own mean velocity, thermal spread, initial density, generation time and divergence rate. The time series of three probes working in the ion saturation regime are used to determine the model parameters by least-squares fitting. This approach allows a higher level of accuracy and insight than time-of-flight analysis based on direct feature correlation alone. A good fit of the main part of the discharge time series is achieved with just three ion groups, with mean velocities ranging in 50–70, 30–45 and 10–25 km s−1 respectively. Each ion group differs in the lateral divergence rate and axial thermal spread, and potentially corresponds with a different charge/mass ratio and/or creation time. Some trends with bank capacitance and discharge voltage are identified and discussed.

FOCA: a new quality-controlled database of floods and catchment descriptors in Italy

Abstract. Here we present FOCA (Italian FlOod and Catchment Atlas), the first systematic collection of data on Italian river catchments for which historical discharge time series are available. Hydrometric information, including the annual maximum peak discharge and average daily annual maximum discharge, is complemented by several geomorphological, climatological, extreme rainfall, land-cover and soil-related catchment attributes. All hydrological information derives from the most recently released datasets of discharge and rainfall measurements. To enhance the reproducibility and transferability of the analysis, this paper provides a description of all the raw data and the algorithms used to build the basin attribute dataset. We also describe the approaches adopted to solve problems encountered during the digital elevation model elaboration in areas characterized by a complex morphology. Details about the data quality-control procedure developed to detect and correct errors are also reported. One of the main novelties of FOCA with respect to other national-scale datasets is the inclusion of a rich set of geomorphological attributes and extreme rainfall features for a large set of basins covering a wide range of elevations and areas. Using this first nationwide data collection (available at https://doi.org/10.5281/zenodo.10446258, Claps et al., 2023), a wide range of environmental applications, with a particular focus on flood studies, can be undertaken within the Italian territory.

Large-Scale Hydrological Models and Transboundary River Basins

Large-scale hydrological modeling is an emerging approach in river hydrology, especially in regions with limited available data. This research focuses on evaluating the performance of two well-known large-scale hydrological models, namely E-HYPE and LISFLOOD, for the five transboundary rivers of Greece. For this purpose, discharge time series at the rivers’ outlets from both models are compared with observed datasets wherever possible. The comparison is conducted using well-established statistical measures, namely, coefficient of determination, Percent Bias, Nash–Sutcliffe Efficiency, Root-Mean-Square Error, and Kling–Gupta Efficiency. Subsequently, the hydrological models’ time series are bias corrected through scaling factor, linear regression, delta change, and quantile mapping methods, respectively. The outputs are then re-evaluated against observations using the same statistical measures. The results demonstrate that neither of the large-scale hydrological models consistently outperformed the other, as one model performed better in some of the basins while the other excelled in the remaining cases. The bias-correction process identifies linear regression and quantile mapping as the most suitable methods for the case study basins. Additionally, the research assesses the influence of upstream waters on the rivers’ water budget. The research highlights the significance of large-scale models in transboundary hydrology, presents a methodological approach for their applicability in any river basin on a global scale, and underscores the usefulness of the outputs in cooperative management of international waters.

Global hydrological reanalyses: The value of river discharge information for world‐wide downstream applications – The example of the Global Flood Awareness System GloFAS

AbstractGlobal hydrological reanalyses are modelled datasets providing information on river discharge evolution everywhere in the world. With multi‐decadal daily timeseries, they provide long‐term context to identify extreme hydrological events such as floods and droughts. By covering the majority of the world's land masses, they can fill the many gaps in river discharge in‐situ observational data, especially in the global South. These gaps impede knowledge of both hydrological status and future evolution and hamper the development of reliable early warning systems for hydrological‐related disaster reduction. River discharge is a natural integrator of the water cycle over land. Global hydrological reanalysis datasets offer an understanding of its spatio‐temporal variability and are therefore critical for addressing the water–energy–food–environment nexus. This paper describes how global hydrological reanalyses can fill the lack of ground measurements by using earth system or hydrological models to provide river discharge time series. Following an inventory of alternative sources of river discharge datasets, reviewing their advantages and limitations, the paper introduces the Copernicus Emergency Management Service (CEMS) Global Flood Awareness System (GloFAS) modelling chain and its reanalysis dataset as an example of a global hydrological reanalysis dataset. It then reviews examples of downstream applications for global hydrological reanalyses, including monitoring of land water resources and ocean dynamics, understanding large‐scale hydrological extreme fluctuations, early warning systems, earth system model diagnostics and the calibration and training of models, with examples from three Copernicus Services (Emergency Management, Marine and Climate Change).

Remote Sensing-Based Extension of GRDC Discharge Time Series - A Monthly Product with Uncertainty Estimates

The Global Runoff Data Center (GRDC) data set has faced a decline in the number of active gauges since the 1980s, leaving only 14% of gauges active as of 2020. We develop the Remote Sensing-based Extension for the GRDC (RSEG) data set that can ingest legacy gauge discharge and remote sensing observations. We employ a stochastic nonparametric mapping algorithm to extend the monthly discharge time series for inactive GRDC stations, benefiting from satellite imagery- and altimetry-derived river width and water height observations. After a rigorous quality assessment of our estimated discharge, involving statistical validation, tests and visual inspection, results in the extension of discharge records for 3377 out of 6015 GRDC stations. The quality of discharge estimates for the rivers with a large or medium mean discharge is quite satisfactory (average KGE value > 0.5) however for river reaches with a low mean discharge the average KGE value drops to 0.33.The RSEG data set regains monitoring capability for 83% of total river discharge measured by GRDC stations, equivalent to 7895 km3/month.

What controls the tail behaviour of flood series: rainfall or runoff generation?

Abstract. Many observed time series of precipitation and streamflow show heavy-tail behaviour. For heavy-tailed distributions, the occurrence of extreme events has a higher probability than for distributions with an exponentially receding tail. If we neglect heavy-tail behaviour we might underestimate the magnitude of rarely observed, high-impact events. Robust estimation of upper-tail behaviour is often hindered by the limited length of observational records. Using long time series and a better understanding of the relevant process controls can help with achieving more robust tail estimations. Here, a simulation-based approach is used to analyse the effect of precipitation and runoff generation characteristics on the upper tail of flood peak distributions. Long, synthetic precipitation time series with different tail behaviour are produced by a stochastic weather generator. These are used to force a conceptual rainfall–runoff model. In addition, catchment characteristics linked to a threshold process in the runoff generation are varied between model runs. We characterize the upper-tail behaviour of the simulated precipitation and discharge time series with the shape parameter of the generalized extreme value (GEV) distribution. Our analysis shows that runoff generation can strongly modulate the tail behaviour of flood peak distributions. In particular, threshold processes in the runoff generation lead to heavier tails. Beyond a certain return period, the influence of catchment processes decreases and the tail of the rainfall distribution asymptotically governs the tail of the flood peak distribution. Beyond which return period this is the case depends on the catchment storage in relation to the mean annual rainfall amount.

A New Index to Assess the Effect of Climate Change on Karst Spring Flow Rate

Karstic aquifers, because of their conduit system, are susceptible to climate change. Ten karst springs in the Zagros region were selected to investigate the impact of climate change under three CMIP6 scenarios: SSP1-1.9, SSP2-4.5, and SSP5-8.5. This study was conducted in three steps: downscaling climate projection, analyzing spring discharge time series, and introducing a new index to assess the impact of climate change on spring flow rate. Applying LARS-WG6, precipitation was downscaled at 14 stations in the study area. Moreover, time series and trend analysis showed that the selected springs have experienced a decrease in their flow rate. Assuming the covariance function between precipitation and spring discharge is constant, new indices (i.e., IQd, IdQd, and Icc) were introduced to highlight the effect of climate change according to the three scenarios. dQd is the variability of spring discharge from past to future, IdQd is spring discharge variability over the historical data, and Icc is the effect of precipitation and spring discharge change together. Icc has a range from −0.25 to 0.25 below and above, which is indicative that two extreme conditions including the spring dryness and overflow are in effect, respectively. The main results revealed that the degree of impact at each spring is a function of climate change scenarios and hydrogeological characteristics of the karstic systems. A more noticeable negative trend in spring flow rate is observed for the karst springs characterized by a dominant conduit flow regime and low matrix storage, located in the areas with low cumulative rainfall, and has a stronger relationship with precipitation. Based on the results, decisions on the management of karst water resources should be made considering where the springs bear free surface and pressurized flow conditions.

Operational low-flow forecasting using LSTMs

This study focuses on exploring the potential of using Long Short-Term Memory networks (LSTMs) for low-flow forecasting for the Rhine River at Lobith on a daily scale with lead times up to 46 days ahead. A novel LSTM-based model architecture is designed to leverage both historical observation and forecasted meteorological data to carry out multi-step discharge time series forecasting. The feature and target selection for this deep learning (DL) model involves evaluating the use of different spatial resolutions for meteorological forcing (basin-averaged or subbasin-averaged), the impact of incorporating past discharge observations, and the use of different target variables (discharge Q or time-differenced discharge dQ). Then, the model is trained using the ERA5 dataset as meteorological forcing, and employed for operational forecast with ECMWF seasonal forecast (SEAS5) data. The forecast results are compared to a benchmark process-based model, wflow_sbm. This study also explores the flexibility of the DL model by fine-tuning the pretrained model with limited SEAS5 dataset. Key findings from feature and target selection include: (1) opting for subbasin-averaged meteorological variables significantly improves model performance compared to a basin-averaged approach. (2) Utilizing dQ as the target variable greatly boosts short-term forecast accuracy compared to using Q, with a mean absolute error (MAE) of 25 m3 s−1 and mean absolute percentage error (MAPE) of 0.02 for the first lead time, ensuring reliability and accuracy at the onset of the forecast horizon. (3) While incorporating historical discharge improves the forecasting of Q, its impact on predicting dQ is less pronounced for short lead times. In the operational forecast with SEAS5, compared to the wflow_sbm model, the DL model exhibits skill in forecasting low flows as evidenced by Continuous Ranked Probability Skill Score (CRPSS) median values of all lead times above zero, and better accuracy in forecasting drought events within short lead times. The wflow_sbm model shows higher accuracy for longer lead times. In the exploration of fine-tuning approach, the fine-tuned model generates marginal short-term enhancements in forecasting low-flow events over a non-fine-tuned model. Overall, this study contributes to advancing the field of low-flow forecasting using deep learning approach.

STREAMFLOW GENERATION OF RIVER NIGER BASIN KOGI USING MODIFIED THOMAS FIERRING MODEL

There is a concerted effort towards protection of environment as a result of incessant change of climate and its resultant effects on human live and property. Studies have shown that flooding is ranked the most devastating natural disaster in the world; and it is a common natural environmental hazard in various parts of Nigeria. In recent times, flooding has been reported to constitute a major challenge to about 1.5 billion people or 19% of the world population; whereas 70 million people globally exposed to flooding every year while over 800 million people are living in flood-prone areas. The 2022 destructive flooding in River Niger Basin Lokoja Nigeria has attracted attention of experts with a view to curtailing the future reoccurrence. Based on this rationale, this study aimed to determine the trend in the hydrometric dataset and to provide extensive time series analysis of the streamflow in the study area using Modified Thomas Fierring Model (MTFM). In order to achieve the objectives of the study, the dataset for 20 years was sourced from the NIWA (National Inland Waterway Authority) whereby the daily and monthly data was extensively analyzed and the (Thomas Fierring) T-F model the data for the “n” years was divided into 12 sections while the data of each month was re-aggressed to the previous month so that there were 12 linear regressions. Also, MTFM was used to predict the monthly flows from River Niger from 2000 to 2019 so as to compute 36 parameters (12) means (), standard deviations (), and Lag-1 autocorrelation coefficients (r1)). After the estimation of the parameters of the MTFM, the series of historical monthly flows for 20 years were generated. These series were compared with the recorded monthly flows of River Niger for the hydrological period of 20 years (2000-2019), with the use of basic statistical characteristics like mean, standard deviation, maximum, minimum, coefficient of variation, skewness and Kurtosis as shown in Table 1 and Figure 2. The findings revealed that monthly discharge time series of River Niger Basin from 2000 to 2019 are consistent with slightly variation in the trends after modified modified Thomas Fierring Model application; and the study further underscored the relevance of MTFM for analyzing streamflow of River Niger Basin Lokoja. Among others, the study recommended up-to-date hydrometric data of River Niger Basin for optimum control and management of streamflow in the study area; and concluded that up-to-date information in this direction will help future researcher to consolidate on the existing body of knowledge in the field.