Kling-Gupta Efficiency Research Articles

When ancient numerical demons meet physics-informed machine learning: adjoint-based gradients for implicit differentiable modeling

Abstract. Recent advances in differentiable modeling, a genre of physics-informed machine learning that trains neural networks (NNs) together with process-based equations, have shown promise in enhancing hydrological models' accuracy, interpretability, and knowledge-discovery potential. Current differentiable models are efficient for NN-based parameter regionalization, but the simple explicit numerical schemes paired with sequential calculations (operator splitting) can incur numerical errors whose impacts on models' representation power and learned parameters are not clear. Implicit schemes, however, cannot rely on automatic differentiation to calculate gradients due to potential issues of gradient vanishing and memory demand. Here we propose a “discretize-then-optimize” adjoint method to enable differentiable implicit numerical schemes for the first time for large-scale hydrological modeling. The adjoint model demonstrates comprehensively improved performance, with Kling–Gupta efficiency coefficients, peak-flow and low-flow metrics, and evapotranspiration that moderately surpass the already-competitive explicit model. Therefore, the previous sequential-calculation approach had a detrimental impact on the model's ability to represent hydrological dynamics. Furthermore, with a structural update that describes capillary rise, the adjoint model can better describe baseflow in arid regions and also produce low flows that outperform even pure machine learning methods such as long short-term memory networks. The adjoint model rectified some parameter distortions but did not alter spatial parameter distributions, demonstrating the robustness of regionalized parameterization. Despite higher computational expenses and modest improvements, the adjoint model's success removes the barrier for complex implicit schemes to enrich differentiable modeling in hydrology.

Enhancing streamflow simulation accuracy in ungauged catchments via parameter calibration with triple collocation-based merged evapotranspiration and streamflow features

Estimating model parameters for accurate hydrological simulations poses a challenge in ungauged catchments. Evapotranspiration (ET) data derived from remote sensing (RS) holds promise for parameter calibration in ungauged catchments. Nonetheless, uncertainties inherent in raw RS-based ET data can pose challenges in the calibration process. This study evaluates two calibration methods mitigating the impact of this uncertainty. The first, a single-variable calibration, utilizes merged ET data based on the triple collocation technique, while the second incorporates streamflow features alongside ET data for multi-variable calibration. Applying the variable infiltration capacity model to 49 Chinese catchments revealed the superiority of the merged ET-based calibration scheme over the three raw RS ET-based calibrations, with improvements noted in the Nash-Sutcliffe efficiency (NSE) and Kling-Gupta efficiency (KGE) coefficients for simulated streamflow. On average, the NSE and KGE for the simulated streamflow demonstrate improvements ranging from 0.023 to 0.080 and 0.026 to 0.061, respectively. The multi-variable calibration schemes exhibit a significant enhancement in the reliability of streamflow simulations compared to the ET-based single-variable calibration schemes. On average, the NSE and KGE for the simulated streamflow exhibit improvements of 0.054 to 0.133 and 0.062 to 0.122, respectively. The performance of these calibrations was generally higher in humid catchments but showed more pronounced improvements in arid catchments when compared to raw RS ET-based calibrations. In conclusion, the proposed calibrations using streamflow features and/or merged ET data offer viable solutions for parameter calibration in ungauged catchments.

Assessment of CMIP6 models performance in simulation precipitation and temperature over Iran and surrounding regions.

This study investigates the performance of CMIP6 models in reproducing historical temperature and precipitation data for Iran and neighboring countries (Afghanistan, Pakistan, Turkmenistan, Azerbaijan, Armenia, Turkey, and Iraq) from 1980 to 2014. Reanalysis data from the ECMWF database (ERA5) for temperature and precipitation were utilized as a reference for the period 1980-2014. Additionally, ten Atmosphere-Ocean General Circulation Models (AOGCMs) from CMIP6 were employed to simulate temperature and precipitation data for the study region based on the IPCC Sixth Assessment Report databases. The Kling-Gupta Efficiency (KGE) index was used to evaluate the accuracy of CMIP6 models in replicating daily temperature and precipitation. The results indicate that different CMIP6 models exhibit varying degrees of accuracy in simulating historical temperatures and precipitation, depending on the month and the country. For instance, the IPSL-CM6A-LR model demonstrated the best annual performance in estimating temperature in Azerbaijan (KGE = 0.5), while the HadGEM3-GC31-LL model showed the lowest annual performance in Pakistan (KGE = -1.4). Interestingly, the models were found to be more accurate in simulating temperatures during warm months compared to cold ones. Furthermore, the accuracy of different models in estimating annual precipitation varied significantly, ranging from -0.64 (MRI-EMS2-0 model in Afghanistan) to 0.05 (CMCC-ESM2 model in Armenia). Similar to temperature, the study found that models were generally more accurate in simulating precipitation during cold months compared to warm ones.

Performance and projections of the NEX‐GDDP‐CMIP6 in simulating precipitation in the Brazilian Amazon and Cerrado biomes

AbstractThe objective of this work is to provide projections of mean annual and monthly precipitation for the Brazilian Amazon and Cerrado biomes, in the near‐term (2021–2040), medium‐term (2041–2060) and long‐term (2081–2100). The intermediate and most pessimistic Intergovernmental Panel on Climate Change (IPCC) greenhouse gas emissions scenarios were considered. Thus, 34 high‐resolution global climate models (GCMs) from the NASA Earth Exchange Global Daily Downscaled Projections (NEX‐GDDP) Phase 6 of the Coupled Model Intercomparison Project (CMIP6) were evaluated. The base period evaluated was from 1981 to 2010. The NEX‐GDDP simulations are bias‐corrected and spatially disaggregated. The Climate Hazards Group InfraRed Precipitation with Station v2.0 was chosen as the source of observed data due to low availability in situ data. The Kling‐Gupta efficiency (KGE) and the global performance indicator were implemented in Google Earth Engine to evaluate the GCMs. The results show that the GCMs perform satisfactorily, except for KACE‐1‐0‐G and IITM‐ESM. The median KGE is 0.86 for the biomes. Thus, the Ensemble Model of 32 GCMs (EM‐32) indicates a reduction in precipitation in the biomes, except the northern Cerrado. In the most pessimistic scenario, changes in annual precipitation range from 3% to −33% until the end of the century. The north‐central Amazon and the northwestern Cerrado are the most affected regions. In general, the monthly precipitations between September and November show the most intense decreasing rates. It is estimated that 91% and 23% of areas in the Amazon and Cerrado biomes, respectively, show robust signs of reduction in mean annual precipitation. Thus, EM‐32 shows more intense and robust climate projections, in comparison to the total annual precipitation of the subset of 33 raw CMIP6 models from Working Group I of the IPCC Sixth Assessment Report. Therefore, the EM‐32 precipitation projections can be applied to future hydrological and hydrosedimentological investigations.

Utilization of the Google Earth Engine for the evaluation of daily soil temperature derived from Global Land Data Assimilation System in two different depths over a semiarid region

AbstractThe Google Earth Engine (GEE) was used to investigate the performance of the Global Land Data Assimilation System (GLDAS) soil temperature (ST) data against observed ST from 13 synoptic stations over a semiarid region in Iran. Three‐hourly ST data were collected and analyzed in two depths (0–10 cm; 40–100 cm) and 5 years. In each depth, GLDAS‐Noah ST data were evaluated for daily minimum, maximum, and average ST (i.e., Tmin, Tmax, and Tavg). Based on the correlation coefficient, Kling–Gupta Efficiency, and Nash–Sutcliffe Efficiency the overall performance of the GLDAS‐Noah is 0.96, 0.66, and 0.79 for Tmin; 0.97, 0.84, and 0.89 for Tavg; and 0.95, 0.89, and 0.89 for Tmax, respectively in the first layer. Likewise, 0.97, 0.85, and 0.86 for Tmin; 0.97, 0.77, and 0.80 for Tavg; and 0.97, 0.69, and 0.69 for Tmax are obtained in the second layer. However, there is a significant negative bias which tends to underestimate ST in the two investigated layers, given by an average bias over all the stations analyzed of −24%, −12%, and −5% for Tmin, Tavg, and Tmax in the first layer, and average bias of −8%, −13%, and −17% for Tmin, Tavg, and Tmax in the second layer. This study reveals that GLDAS‐Noah‐derived ST can be used in arid regions where little or no observation data is available. Moreover, GEE performed as an advanced geospatial processing tool in regional scale analysis of ST in different layers.

Multiple conceptual hydrological models for simulating streamflow in data-sparse river basins: an application of the Vietnamese Cau river basin

ABSTRACT Streamflow plays a critical role in water resources management, requiring precise estimation, especially in data-sparse river basins. In this study, multiple hydrological models (HYMOD, GR4J, and NAM) are presented and employed to simulate streamflow in the data-sparse Vietnamese Cau river basin at the Thac Rieng station, providing an illustrative example. Four indices named Nash–Sutcliffe efficiency (NSE), correlation coefficient (r), Kling–Gupta efficiency (KGE), and root mean square error (RMSE) are implemented to quantitatively assess model performance. Firstly, using the available hydrometeorological data from 1 January 1960 to 31 December 1981, parameters in each model are determined using the Shuffled Complex Evolution University of Arizona optimization method. The results show that three models reproduced acceptably observed streamflow, with NSE ranging from 0.65 to 0.75, r and KGE varying between 0.72 and 0.87, and RMSE is less than 6.5% of the observed streamflow magnitude. Secondly, the NAM model was found as the most suitable for simulating observed streamflow in the studied river basin. Thirdly, three models were applied to simulate streamflow in the period from 1 January 1982 to 31 December 2022, revealing similar magnitudes of four statistical indicators of observed and estimated streamflow. The capability of the three models in simulating streamflow in data-sparse river basins is finally discussed.

Development of temperature and mass transfer-based empirical models for estimating reference evapotranspiration in Nigeria

ABSTRACT The empirical models commonly employed as alternatives for estimating evapotranspiration provide constraints and yield inaccurate results when applied to Nigeria. This study aims to develop novel empirical models to enhance evapotranspiration (ET0) estimation accuracy in Nigeria. The coefficients of non-linear equations were optimised using the particle swarm optimisation (PSO) algorithm for the development of the two new ET0 models for Nigeria, Awhari1 (temperature-based) and Awhari2 (mass transfer-based). ERA5 reanalysis data with a 0.1° × 0.1° resolution was used. The models were rigorously assessed against the FAO-56 Penman–Monteith method, resulting in Kling–Gupta efficiency (KGE) and percentage bias (Pbias) values of 0.75, 6.49, and 0.92, 5.67, respectively. The spatial distribution analysis of performance metrics showed both equations exhibited superior accuracy in estimating ET0 across diverse climatic zones in Nigeria. The incorporation of PSO in model development, coupled with spatial analysis, highlights the study's multidimensional approach. The spatial performance of the models indicates that they can be valuable tools for water resource management, irrigation planning, and sustainable agriculture practices in Nigeria.

A long-term multivariate time series prediction model for dissolved oxygen

Accurate and efficient long-term prediction of marine dissolved oxygen (DO) is crucial for the sustainable development of aquaculture. However, the multidimensional time dependency and lag effects of marine chemical variables present significant challenges when handling multiple inputs in univariate long-term prediction tasks. To address these issues, we designed a multivariate time-series long-term prediction model (LMFormer) based on the Transformer architecture. The proposed multivariate time-series decomposition strategy effectively leverages the feature information of prediction variables at different scales, thereby reducing the loss of critical information. Additionally, a dynamic variable selection strategy based on a gating mechanism was designed to optimize the collinearity problem in the multivariate data feature extraction process. Finally, an efficient two-stage attention architecture is proposed to effectively capture the long-range dependencies between dynamic features. This study conducted high-precision 7-day advance DO long-term predictions in two case studies, the environmentally stable Shandong Peninsula in China and the San Juan Islands in the United States, which are affected by extreme conditions such as ocean currents. The experimental results demonstrate the superior prediction performance and generalizability of the designed model. In the Shandong Peninsula case, the mean absolute error (MAE), root mean square error (RMSE), coefficient of determination (R2), and Kling–Gupta efficiency (KGE) reached 0.0159, 0.126, 0.9743, and 0.9625, respectively. In the San Juan Islands case, the MAE was reduced by an average of 42.34% compared to that of the baseline model, the RMSE was reduced by an average of 24.57%, the R2 increased by 22.54%, and the KGE improved by an average of 12.04%. Overall, the proposed prediction model effectively achieves long-term prediction of multivariate marine chemical data, providing valuable references for sustainable management and decision-making in aquaculture.

Meta-LSTM in hydrology: Advancing runoff predictions through model-agnostic meta-learning

In the field of hydrology, deep learning has become a prevalent tool for runoff simulation. However, the limitations stem from their primary focus on normality, which fails to accurately capture rare events. This study introduces a novel Long Short-Term Memory (LSTM) model, termed Meta-LSTM, which is based on the Model-Agnostic Meta-Learning (MAML) framework. The Meta-LSTM model is capable of dynamically fine-tuning its parameters to ensure effective adaptation to various runoff scenarios. From this Meta-LSTM model, the Meta-enhanced paradigm is subsequently proposed, ensuring effective adaptation to diverse models and datasets. We conducted experiments with various models on the CAMELS and CAMELS-AUS datasets. Compared to conventional deep learning models for runoff simulation, Meta-enhanced models exhibit substantial improvements. Specifically, the Meta-LSTM model reduces the PBIAS from −13.86 % to −5.18 %, increasing the Kling-Gupta Efficiency (KGE) by 7 % on the CAMELS dataset. For the Meta-GR4J-CNN model, the PBIAS decreases by 21 %, and the Nash-Sutcliffe Efficiency (NSE) increases by 2 % on the CAMELS-AUS dataset. This demonstrates the ability of the Meta-enhanced model to more accurately represent observed data without significant additional time costs. Our approach overcomes the problem of uniform training in simulations of complex scenarios and promises to significantly improve the accuracy of future hydrologic studies.

Development and validation of an oil palm model for a wide range of planting densities and soil textures in Malaysian growing conditions

A semi-mechanistic oil palm growth and yield model called Sawit.jl was developed to account for a wide range of planting densities and soil textures under Malaysia's climate conditions. The model comprises components related to meteorology, photosynthesis, energy balance, soil water content, and crop growth. The model simulates instantaneous meteorological properties using daily weather data, calculates simultaneous evaporation from crop and soil with the Shuttleworth-Wallace model, determines soil water content through Darcy's law, and adapts a biochemical C3 model for photosynthesis. The model is also parameterized using updated measurements from the newer tenera oil palm, including temperature-dependent Rubisco kinetics, specific leaf area, and the partitioning of nutrients and dry matter between various tree parts. Sawit.jl was validated using historical field measurement data from seven Malaysian oil palm sites, encompassing palm ages spanning 1–23 years. These seven sites differed in soil type (Inceptisols and Ultisols), planting density (82–299 palms ha−1), soil texture (27–59 % clay and 7–67 % sand), and rainfall (1800–2800 mm yr−1). The model showed overall good accuracy in simulating oil palm parameters (except for trunk weight) across diverse conditions, with model agreement metrics ranging from 6 to 27 % for model absolute errors, −22 to +17 % for model bias, and 0.38 to 0.98 for the Kling-Gupta Efficiency index. The model also predicted the response of oil palm yield to abrupt rainfall changes, such as those during El Niño and La Niña events, while accounting for how soil texture, rainfall, and other meteorological factors influence water deficits and crop photosynthesis. However, model accuracy varied by site, planting density, and oil palm parameter. Model accuracy can be increased by more accurately representing the oil palm microclimate, incorporating fruiting activity, and refining the dry matter partitioning mechanism for the trunk.

A novel strategy for flood flow Prediction: Integrating Spatio-Temporal information through a Two-Dimensional hidden layer structure

In recent years, neural network models have been extensively applied in flood prediction due to their superior performance. However, most studies aimed at enhancing models have simply stacked basic models, overlooking the need for designs specific to the physical attributes of hydrological data. While stacking basic models can enhance flood prediction to a degree, they often introduce significant unnecessary complexity redundancy. Moreover, improvement strategies are challenging to generalize across different models. To address this issue, a widely applicable Two-Dimensional Hidden Layer (Td) architecture is proposed. Unlike single hidden layer architectures, this design specifically tackles hydrological data’s spatiotemporal characteristics. Discretizing spatiotemporal information in flood and rainfall data yields an abstract, two-dimensional representation of the hydrological data to enhance the model’s ability to utilize hydrological information. Using a Td structure instead of the single hidden layer in Recurrent Neural Network (RNN), Long Short-Term Memory (LSTM), Bidirectional Long Short-Term Memory (BiLSTM), and Convolutional Neural Network (CNN), the TdRNN, TdLSTM, TdBiLSTM, and TdCNN models are developed to assess the novel structure’s impact on foundational models. Experiments were conducted using empirical flood process data from the LouDe Station in Shandong, China, and the NingXiang Station in Hunan, China. The results indicate that, compared to benchmark models, network models incorporating a Td structure exhibit superior adaptability across various lead times. For instance, at the LouDe Station, with a one-hour lead time, the TdCNN model showed improvements in the Nash-Sutcliffe Efficiency Coefficient (NSE), Kling-Gupta Efficiency Coefficient (KGE), Mean Absolute Error (MAE), and Root Mean Square Error (RMSE) performance metrics by 26.24%, 33.48%, 62.51%, and 56.88%, respectively, compared to the CNN model. The new models also demonstrated significant performance enhancements under other conditions. The experimental results demonstrate that the Td structure is an effective and versatile model improvement strategy. It enhances the prediction accuracy by strengthening the model’s ability to utilize information. This study may contribute to the development of a flood prediction system and provide data to support subsequent reservoir scheduling and disaster mitigation planning.

On the value of hydrochemical data for the interpretation of flow and transport processes in the Baget karst system, France

Continuous hourly time series of hydrochemical data can provide insights into the subsurface dynamics and main hydrological processes of karst systems. This study investigates how high-resolution hydrochemical data can be used for the verification of robust conceptual event-based karst models. To match the high temporal variability of hydrochemical data, the LuKARS 2.0 model was developed on an hourly scale. The model concept considers the interaction between the matrix and conduit components to allow a flexible conceptualization of binary karst systems characterized by a perennial spring and intermittent overflow as well as possible surface water bypassing the spring. The model was tested on the Baget karst system, France, featuring a recharge area defined by the coexistence of karst and nonkarst areas. The Morris screening method was used to investigate parameter sensitivity, and to calibrate the model according to the Kling-Gupta Efficiency (KGE). Model verification was performed by considering additional hydrochemical constraints with the aim of representing the internal dynamics of the systems, i.e., water contributions from the various compartments of the conceptual model. The hydrochemical constraints were defined based on high-temporal resolution time series of SO42− and HCO3−. The results of this study show that the simulation with the highest KGE among 9,000 model realizations well represents the dynamics of the spring discharge but not the variability of the internal fluxes. The implementation of hydrochemical constraints facilitates the identification of realizations reproducing the observed relative increase in the flow contribution from the nonkarst area.

Integrated metaheuristic algorithms with extreme learning machine models for river streamflow prediction

Accurate river streamflow prediction is pivotal for effective resource planning and flood risk management. Traditional river streamflow forecasting models encounter challenges such as nonlinearity, stochastic behavior, and convergence reliability. To overcome these, we introduce novel hybrid models that combine extreme learning machines (ELM) with cutting-edge mathematical inspired metaheuristic optimization algorithms, including Pareto-like sequential sampling (PSS), weighted mean of vectors (INFO), and the Runge–Kutta optimizer (RUN). Our comparative assessment includes 20 hybrid models across eight metaheuristic categories, using streamflow data from the Aswan High Dam on the Nile River. Our findings highlight the superior performance of mathematically based models, which demonstrate enhanced predictive accuracy, robust convergence, and sustained stability. Specifically, the PSS-ELM model achieves superior performance with a root mean square error of 2.0667, a Pearson’s correlation index (R) of 0.9374, and a Nash–Sutcliffe efficiency (NSE) of 0.8642. Additionally, INFO-ELM and RUN-ELM models exhibit robust convergence with mean absolute percentage errors of 15.21% and 15.28% respectively, a mean absolute errors of 1.2145 and 1.2105, and high Kling-Gupta efficiencies values of 0.9113 and 0.9124, respectively. These findings suggest that the adoption of our proposed models significantly enhances water management strategies and reduces any risks.

Statistical Performance of Gridded Rainfall Datasets Over Ungauged Jalaur River Basin, Philippines

The study presented aims to find the most appropriate climate dataset for the data-scarce Jalaur River Basin (JRB), Iloilo, Philippines, by evaluating the statistical performance of five rainfall datasets (APHRODITE, CPC NOAA, ERA5, SA-OBS, and PGF-V3) with resolutions of 0.25° and 0.5° having a time domain of 1981 to 2005. Bilinear interpolation implemented through Climate Data Operator (CDO) was used to extract and process grid climate datasets with Linear scaling as bias correction to minimize product simulation uncertainties. The datasets were compared to the lone meteorological station nearest to JRB investigated at monthly and annual timescales using six statistical metrics, namely, Pearson’s correlation coefficient (r), coefficient of determination (R2), modified index of agreement (d1), Kling-Gupta efficiency, Nash-Sutcliffe efficiency (NSE), and RMSE-observations standard deviation ratio (RSR). The results indicate a strong positive correlation with the observed data for both rainfall and temperature (r > 0.8; R2, d1 > 0.80). Although graphical observation shows an underestimation of rainfall, goodness-of-fit values indicate very good model performance (NSE, KGE > 0.75; RSR < 0.50). In terms of temperature, variable responses are observed with significant overestimation for maximum temperature and underestimation for minimum temperature. SA-OBS proved to be the best-performing dataset, followed by ERA5 and PGF-V3. These key findings supply useful information in deciding the most appropriate gridded climate dataset for hydrometeorological investigation in the JRB and could enhance the regional representation of global datasets.

A multi-criteria decision analysis approach for ranking the performance of CMIP6 models in reproducing precipitation patterns over Abaya-Chamo sub-basin, Ethiopia

The most suitable multi-model ensemble set of general circulation models is used to reduce the uncertainty associated with GCM selection and improve the accuracy of the model simulations. This study evaluated the performance of 20 global climate models participating in the Coupled Model Intercomparison Project Phase 6 (CMIP6) in reproducing precipitation patterns over the Abaya-Chamo Sub-basin, Ethiopia. For the validation and selection of the models' capabilities, datasets from the Climate Hazards Infrared Precipitation with Stations (CHIRPS) were used after comparing them with ground observational datasets. The objective was to identify the most suitable multi-model ensemble (MME) of a subset of CMIP6 GCMs to capture the rainfall for the 1981–2014 period over the region. Climate Data Operators (CDOs) were used in climate data processing and extraction, and the Mann-Kendall test and Theil-Sen slope estimator methods were utilized to analyze the trends of the CMIP6 simulations. Four statistical metrics (Nash-Sutcliffe coefficient, percent bias, normalized root mean square error, and Kling-Gupta efficiency) were used to further assess the performance of the models. A multi-criteria decision analysis approach, namely, the technique for order preferences by similarity to an ideal solution (TOPSIS) method, was used to obtain the overall ranks of CMIP6 models and to select the best-performing CMIP6 model in the region. The results indicated that CHIRPS and most of the CMIP6 simulations generally reproduced bimodal precipitation patterns over the region. The CESM2-WACCM, NorESM2-MM, NorESM2-LM, and NorESM2-LM models performed better than the other models in reproducing seasonal patterns for the winter, spring, summer, and autumn seasons, respectively. On the other hand, FGOALS-f3-L revealed the trends of the reference datasets for all seasons. In terms of the NSE, PB, NRMSE, and KGE metrics, EC-Earth3-C, EC-Earth3, EC-Earth3-C, and EC-Earth-C, respectively, were considered good at representing the observed features of precipitation over the region. EC-Earth3-C,EC-Earth3, EC-Earth3-Veg-LR, ACCESS-CM2, MPI-ESM1-2-HR, and CNRM-CM6-1-HR exhibited the best performances in the Abaya-Chamo Sub-basin.

Performance evaluation and ranking of CMIP6 global climate models over upper blue nile (abbay) basin of Ethiopia

The use of Global Climate Models (GCMs) data is the most practical way to conduct studies on climate science. However, performance evaluation and the selection of appropriate GCMs are vital. In this research, the effectiveness of eight selected CMIP6 GCMs in simulating the annual and seasonal rainfall observed over the Ethiopian Upper Blue Nile Basin from 1988 to 2014 was assessed. Five performance metrics (PMs) were used in the study: the correlation coefficient, root mean square error, bias percentage, Kling-Gupta efficiency and Nash-Sutcliffe efficiency. The scores of the various PMs were then combined into one, and the CMIP6 GCMs were ranked using Compromised Programming (CP). The findings from the CP were verified using a spatial, Taylor Diagram (TD), and areal average annual and seasonal evaluations. Even though the PMs produced some contradicting results, the study exhibited that CP was capable to evaluate the CMIP6 GCMs consistently. A regional evaluation of the CMIP6 GCMs relative to the observed data revealed that the best-ranked CMIP6 GCMs by using CP were capable to more accurately replicate the observed annual and seasonal rainfall. The lowest-ranking CMIP6 GCMs were found to have either spatially overvalued or undervalued the amount of rainfall over the basin. The best three CMIP6 GCMs for annual rainfall, according to the results of the CP method, are BCC-CSM2-MR, MIROC6, and NorESM2-MM; for the Kiremt season, the best CMIP6 GCMs are BCC-CSM2-MR, GISS-E2-2-G, and EC-Earth3. INM-CM5-0, MIROC6, and MRI-ESM2-0 ranked highest for Bega season, and EC-Earth3, BCC-CSM2-MR, and MRI-ESM2-0 for Belg season. It is recommended using the above-ranked CMIP6 GCMs to predict the characteristics of rainfall in the UBNB. Furthermore, results suggest that the CMIP6 GCMs be evaluated with a range of PMs across the whole temporal scales and that techniques such as CP be used to identify the best-performing CMIP6 GCMs.

Evaluation of Gridded Rainfall Products in Three West African Basins

In recent years, accessing rainfall data from ground observation networks maintained by national meteorological services in West Africa has become increasingly challenging. This is primarily due to high acquisition costs and the often sparse distribution of rainfall gauges across the region, which limits their use in hydrological studies and related research. At the same time, the rising availability of precipitation products derived from satellite/earth observations, reanalysis datasets, and in situ measurements presents exciting prospects for hydrological applications. Nonetheless, these datasets constitute indirect measurements, necessitating rigorous validation against ground-based rainfall data. This study comprehensively assesses twenty-three gridded rainfall products, including sixteen from satellites, six from reanalysis data, and one from in situ measurements, across the Senegal, Gambia, and Casamance River basins. Performance evaluation is conducted across distinct climatic zones, both pre- and post-resampling against observed rainfall data gathered from forty-nine rainfall stations over a six-year period (2003–2008). Evaluation criteria include the Kling–Gupta Efficiency (KGE) and Percentage of Bias (PBIAS) metrics, assessed at daily, monthly, and seasonal time steps. The results reveal distinct performance levels among the evaluated rainfall products. RFE, ARC2, and CPC notably yield the highest KGE scores at the daily time step, while GPCP, CHIRP, CHIRPS, RFE, MSWEP, ARC2, CPC, TAMSAT, and CMORPHCRT demonstrate superior performance at the monthly time step. During the rainy season, these products generally exhibit robustness. However, rainfall estimates derived from reanalysis datasets (ERA5, EWEMBI, MERRA2, PGF, WFDEICRU, and WFDEIGPCC) perform poorly in the studied basins. Based on the PBIAS metric, most products tend to underestimate precipitation values, while only PERSIANN and PERSIANNCCS lead to significant overestimations. Spatially, optimal performance of the products is observed in the Casamance basin and the Sudanian and Sahelian climatic zones within the Gambia and Senegal basins. Conversely, in the Guinean zone of the Gambia and Senegal Rivers, the rainfall products displayed the poorest performance.

Inter-Comparison of Multiple Gridded Precipitation Datasets over Different Climates at Global Scale

Comprehensive evaluations of global precipitation datasets are imperative for gaining insights into their performance and potential applications. However, the existing evaluations of global precipitation datasets are often constrained by limitations regarding the datasets, specific regions, and hydrological models used for hydrologic predictions. The accuracy and hydrological utility of eight precipitation datasets (including two gauged-based, five reanalysis and one merged precipitation datasets) were evaluated on a daily timescale from 1982 to 2015 in this study by using 2404 rain gauges, 2508 catchments, and four lumped hydrological models under varying climatic conditions worldwide. Specifically, the characteristics of different datasets were first analyzed. The accuracy of precipitation datasets at the site and regional scale was then evaluated with daily observations from 2404 gauges and two high-resolution gridded gauge-interpolated regional datasets. The effectiveness of precipitation datasets in runoff simulation was then assessed by using 2058 catchments around the world in combination with four conceptual hydrological models. The results show that: (1) all precipitation datasets demonstrate proficiency in capturing the interannual variability of the annual mean precipitation, but with magnitudes deviating by up to 200 mm/year among the datasets; (2) the precipitation datasets directly incorporating daily gauge observations outperform the uncorrected precipitation datasets. The Climate Precipitation Center dataset (CPC), Global Precipitation Climatology Center dataset (GPCC) and multi-source weighted-ensemble precipitation V2 (MSWEP V2) can be considered the best option for most climate regions regarding the accuracy of precipitation datasets; (3) the performance of hydrological models driven by different datasets is climate dependent and is notably worse in arid regions (with median Kling–Gupta efficiency (KGE) ranging from 0.39 to 0.65) than in other regions. The MSWEP V2 posted a stable performance with the highest KGE and Nash–Sutcliffe Efficiency (NSE) values in most climate regions using various hydrological models.

Monitoring spatiotemporal changes in urban flood vulnerability of Peninsular Malaysia from satellite nighttime light data

Urban flood vulnerability monitoring requires a large amount of socioeconomic and environmental data collected at regular time intervals. However, collecting such a large volume of data poses a significant constraint in assessing changes in flood vulnerability. This study proposed a novel method to monitor spatiotemporal changes in urban flood vulnerability from satellite nighttime light (NTL) data. Peninsular Malaysia was chosen as the research region as floods are the most devastating and recurrent phenomena in the region. The study developed a flood vulnerability index (FVI) based on socioeconomic and environmental data from a single year. This FVI was then linked to NTL data using an Adaptive neuro-fuzzy inference system (ANFIS) machine learning algorithm. The model was calibrated and validated with administrative unit scale data and subsequently used to predict FVI at a spatial resolution of 10 km for 2000–2018 using NTL data. Finally, changes in estimated FVI at different grid points were evaluated using the Mann-Kendall trend method to determine changes in flood vulnerability over time and space. Results showed a nonlinear relationship between NTL and flood vulnerability factors such as population density, Gini coefficient, and percentage of foreign nationals. The ANFIS technique performed well in estimating FVI from NTL data with a normalized root-mean-square error of 0.68 and Kling-Gupta Efficiency of 0.73. The FVI revealed a high vulnerability in the urbanized western coastal region (FVI ∼ 0.5 to 0.54), which matches well with major contributing regions to flood losses in Peninsular Malaysia. Trend assessment showed a significant increase in flood vulnerability in the study area from 2000 to 2018. The spatial distribution of the trend indicated an increase in FVI in the urbanized coastal plains, particularly in rapidly developing western and southern urban regions. The results indicate the potential of the technique in urban flood vulnerability assessment using freely available satellite NTL data.

Understanding the role of the spatial-temporal variability of catchment water storage capacity and its runoff response using deep learning networks

The land surface of a watershed acts as a large reservoir, with its catchment water storage capacity (CWSC) influencing rainfall-runoff relationship. Estimating CWSC at global grid scale is challenging due to calibration complexity, limited spatial continuity, and data scarcity. To address this, a deep learning-based approach incorporates spatial reconstruction and temporal transfer for capturing spatio-temporal variations in watershed characteristics. The study focuses on the Global Runoff Data Centre dataset and presents a grid-based hydrological model. Findings demonstrate accurate identification of CWSC distribution, with the model achieving an R 2 of 0.92 and the runoff Kling–Gupta efficiency of 0.71 during validation. According to the CMIP6 projections, the global CWSC is anticipated to undergo a significant increase at a rate of 1.7 mm per decade under the SSP5-8.5 emission scenario. Neglecting spatio-temporal CWSC variability globally overestimates climate change’s impact on runoff, potentially reducing the projected long-term increase by up to 41%.