Nash-Sutcliffe Efficiency Values Research Articles

Forecasting Lake Nokoué Water Levels Using Long Short-Term Memory Network

The forecasting of hydrological flows (rainfall depth or rainfall discharge) is becoming increasingly important in the management of hydrological risks such as floods. In this study, the Long Short-Term Memory (LSTM) network, a state-of-the-art algorithm dedicated to time series, is applied to predict the daily water level of Lake Nokoué in Benin. This paper aims to provide an effective and reliable method to enable the reproduction of the future daily water level of Lake Nokoué, which is influenced by a combination of two phenomena: rainfall and river flow (runoff from the Ouémé River, the Sô River, the Porto-Novo lagoon, and the Atlantic Ocean). Performance analysis based on the forecasting horizon indicates that LSTM can predict the water level of Lake Nokoué up to a forecast horizon of t + 10 days. Performance metrics such as Root Mean Square Error (RMSE), coefficient of correlation (R2), Nash–Sutcliffe Efficiency (NSE), and Mean Absolute Error (MAE) agree on a forecast horizon of up to t + 3 days. The values of these metrics remain stable for forecast horizons of t + 1 day, t + 2 days, and t + 3 days. The values of R2 and NSE are greater than 0.97 during the training and testing phases in the Lake Nokoué basin. Based on the evaluation indices used to assess the model’s performance for the appropriate forecast horizon of water level in the Lake Nokoué basin, the forecast horizon of t + 3 days is chosen for predicting future daily water levels.

A novel urban flood risk assessment framework based on refined numerical simulation technology

Due to urbanization and climate change, urban areas are becoming more susceptible to flooding, posing significant risks to human life and property. Numerical simulation techniques can provide valuable support in mitigating urban flood risks. The objective of this study is to establish a novel framework for flood risk assessment based on refined urban flood simulations. To achieve refined urban flood simulations, the Torricelli’s law is incorporated into a two-dimensional fluid dynamics model, and a numerical analysis model is developed to estimate the inundation within buildings and accurately simulate the entire inundation process. The model was used to reproduce a real flood event in Omihachiman City, Japan, with the Nash-Sutcliffe Efficiency (NSE) values of 0.6488 and 0.7182, demonstrating the credibility of the model. Based on the simulation results, the hydrological causes of the flood event were analyzed, and flood prevention measures were proposed. After implementing these measures, the maximum water depths at the study area were reduced to 0.14 m, 0.62 m, and 0.12 m, respectively, proving the effectiveness of the flood prevention measures. To address the issue of inundation depth not fully reflecting flood risk, the instability of human bodies in floodwaters is conducted to evaluate flood risk for adults and children in the target area. In response to increasingly frequent extreme rainfall events, the extreme precipitation events of three different return periods (10-year, 50-year, and 100-year) are used to assess the flood risk and the effectiveness of flood prevention measures. Based on the assessment results, the flood prevention measures not only reduced the flood risk levels but also decreased the flood risk area, further confirming their effectiveness.

Climate change impacts on the hydrological behaviour of watershed in northeastern Tunisia (Oued El Abid watershed)

Climate change significantly impacts watershed hydrology, altering water supplies, natural disasters, and eco-hydrologic processes. These effects vary geographically, with some regions facing severe droughts while others experience increased precipitation and flooding. Climate change influences atmospheric evaporation demand, precipitation patterns, vegetation composition, streamflow characteristics, and groundwater dynamics. The temperature increases and rainfall changes in some watersheds can lead to increased potential evapotranspiration and decreased streamflow and soil water. Flood frequency analyses indicate an overall increasing trend in the latter half of the 21st century. The Oued El Abid catchment area, as like all catchments in northern Tunisia, is likely to be subject to major climatic and hydrological variations as a result of climate change. Advanced hydrological modelling using the HBV-Light model was set up to simulate current conditions and future scenarios based on Afric-Cordex climate models. The model demonstrates robust performance, achieving Nash-Sutcliffe Efficiency (NSE) values of 0.71 in calibration (1972-1992), and 0.66 in validation from 1993-2001. This climate change has a direct effect on the hydrological projections, which reveal significant changes, with annual flow reductions ranging from 35% to 43%, with more severe impacts under the RCP8.5 scenario for 2069-2099. Spring and summer seasons are particularly vulnerable, showing flow decreases of up to 68% and over 40%, respectively, extreme monthly variability is projected, reaching 88% reduction in July under RCP8.5. This research provides important results for water resource managers, highlighting the necessity of long-term planning and sustainable water management practices in the face of climate change.

Evaluation of hydrological models for streamflow prediction: a case study of the Mille River, Lower Awash Basin, Ethiopia

ABSTRACT Accurate prediction of stream flow is essential for effective water resource management, particularly in regions with complex hydrological dynamics. The Mille River, which flows through this basin and has a watershed area of 4,862.3 km², plays a crucial role in the region's water resources. However, selecting the most suitable hydrological model for this task remains a challenge due to varying model performances. This study addresses the problem of identifying the most effective hydrological model for predicting stream flow in the Mille River by evaluating three models: Soil and Water Assessment Tool (SWAT), Hydrologiska Byråns Vattenbalansavdelning (HBV-IHMS), and Artificial Neural Network (ANN). The objectives are to compare these models’ abilities to simulate daily stream flow and to determine their accuracy. The significance of this evaluation lies to guide water management decisions by identifying which model provides the most reliable predictions. The methodology involves using performance metrics, Nash-Sutcliffe Efficiency (NSE) values, to assess the accuracy of each model. The results show that SWAT achieved the highest NSE values of 0.81 and 0.82, indicating the best performance in simulating stream flow. ANN followed with NSE values of 0.79 and 0.81, while HBV-IHMS had lower NSE values of 0.73 and 0.74. Although all models demonstrated potential, SWAT and ANN outperformed HBV-IHMS in capturing the complex hydrological processes of the Mille River. The findings from this study indicate that the choice of model significantly impacts the accuracy and reliability of streamflow predictions. The SWAT and ANN models show strong potential for use in operational streamflow forecasting in the Mille River and similar basins. In conclusion, further research is recommended to refine model calibration and validation processes. Incorporating additional data sources, such as remote sensing and climate projections, could further enhance model accuracy and provide a more comprehensive understanding of the region's hydrological dynamics.

Evaluating nitrogen dynamic and utilization under controlled-release fertilizer application for sunflowers in an arid region: Experimental and modeling approach

Traditional nitrogen fertilizers (TNF), such as urea, percolate easily in arid fields, posing low nitrogen use efficiency (NUE) and a high non-point pollution risk. Controlled-release fertilizers (CRF) exhibit significantly lower deep seepage, rendering it a favorable choice in arid fields due to its ability to enhance NUE through slow-release mechanisms. However, current models do not fully account for the soil nitrogen dynamics and crop interactions under controlled-release conditions, and lack quantification. This study improved the APSIM model by adjustment the urea hydrolysis rate to assess the impact of CRF and TNF applications on soil health, crop growth, and water quality. Calibration and validation were conducted through experiments in the Hetao Irrigation District of China from 2019 to 2020, with different nitrogen application rates (135, 225, and 315 kg/ha). The model accurately simulated soil NO3-N concentration (SNC), cumulative NO3-N leaching (CNL), nitrogen uptake (NU), and sunflower yield. During the validation process, R2 and Nash-Sutcliffe efficiency (NSE) values were both above 0.75. Results indicated that the average SNC, NU, and yield under CRF application were significantly higher than those under TNF application, with increases of 38.62%, 44.92%, and 18.38%, respectively. Notably, the proportion of soil nitrogen available (PSNA), a novel metric proposed in this study, was 159.50% higher in the 0–40 cm soil layer with CRF compared to TNF. Additionally, CNL and NO3-N leaching loss rate (NLLR) decreased by 25.76% and 25.77%, respectively. Scenario simulations indicated that the optimal fertilization strategy for this region is to use 180–193.5 kg/ha of CRF with a release period of 80–85.5 d to balance agricultural productivity and ecological protection. This study confirms the significant advantages of CRF in enhancing yield, improving nitrogen management, and promoting environmental sustainability, providing a scientific basis for CRF management strategies and supporting the shift towards more efficient and environmentally friendly agricultural practices.

Ensemble of artificial intelligence and physically based models for rainfall–runoff modeling in the upper Blue Nile Basin

ABSTRACT This study investigated the performance of the adaptive neuro-fuzzy inference system (ANFIS), feed forward neural network (FFNN), Soil and Water Analysis Tool (SWAT), Hydrologic Engineering Center's Hydraulic Modeling System (HEC-HMS), Hydrologiska Byråns Vattenbalansavdelning (HBV), and support vector regression (SVR) models for rainfall–runoff modeling using gauged and satellite rainfall, and their fusions in the Gilgel Abay watershed, Ethiopia. Afterward, simple average ensemble (SAE), weighted average ensemble (WAE), and neural network ensemble (NNE) techniques were applied to combine the outputs of individual models under three scenarios. The performance of the models was evaluated using Nash–Sutcliffe efficiency (NSE) and root mean square error (RMSE). The results demonstrated that the ANFIS model outperformed all the other single models with validation stage NSE values of 0.864 and 0.875, and RMSE values of 23.58 and 21.84 m3/s for gauge and fusion rainfall data, respectively. Among the physical-based models, SWAT gave better modeling performance with the validation stage NSE values of 0.81 and 0.821 for gauge and fusion rainfall data, respectively. Moreover, an ensemble of artificial intelligence and physical-based models greatly improved the overall modeling performance. The NNE improved the performance of single models up to 15.7 and 21.2 5% for fusion and satellite-based rainfall modeling, respectively.

Combining signal decomposition and deep learning model to predict noisy runoff coefficient

Predicting runoff coefficient (Rc), as an indicator of the catchment’s response to the rainfall-runoff process, remains a persistent challenge using different modelling techniques, especially in catchments with strong human manipulation. This study investigates the efficiency of the Long Short-Term Memory (LSTM) method in predicting Rc for the Rur catchment, in Germany. The period from 1961 to 2021 is considered, which is subject to human intervention and significant urbanization especially in the northern part of the catchment. An LSTM structure is defined by employing inputs at a monthly resolution including temperature, precipitation, soil water storage, and total evaporation with a look-back window of 1 to 6 months to model noisy Rc data of the study area. Two approaches using either undecomposed or decomposed Rc were employed in conjunction with the LSTM method, to mitigate the impact of noise associated with Rc. The results show that in the case of undecomposed Rc, the best performance of the LSTM structure was obtained with a 4-month look-back window, yielding Nash-Sutcliffe efficiency (NSE) of 0.55, 0.46, and 0.15 for training, validation, and test sets, respectively. These results highlight inadequate accuracy in accounting for the presence of noise in Rc. Therefore, in the second novel approach, we used maximal overlap discrete wavelet transform (MODWT) to decompose the Rc up to level 3 to reduce the complexity and distribute the noise effects across each level. The new approach showed high accuracy in modelling noisy data of Rc with NSE values of 0.97, 0.95, and 0.90 for training, validation, and test sets, respectively. The obtained results underscore the pivotal role of decomposition techniques in conjunction with LSTM to account for the presence of noise, especially in catchments with strong human manipulation.

Uncertainty of canopy interception modeling in high-altitude Picea crassifolia forests of Semi-arid regions

The study of physically-based rainfall interception is crucial for comprehending the water balance within forest ecosystems and the contribution of vegetation to the hydrological cycle, particularly in arid/semi-arid ecosystems. Despite its importance, there is a lack of comprehensive sensitivity analysis and parameter optimization, resulting in uncertain or suboptimal predictive accuracy. To mitigate these shortcomings, this research involved the establishment and assessment of three quintessential forest canopy interception models namely, the power Návar model, the reformulated Gash model, and the Liu model, within semi-arid forest environments at two different elevations. A global sensitivity analysis conducted on the three physical models indicated that the canopy saturation point and the mean rainfall intensity required for canopy saturation were the parameters to which the reformulated Gash and Liu models were most sensitive when applied to high-altitude settings. Conversely, for the Návar model, the most sensitive parameters were the interception coefficient of the linear equation, and the parameters of the power equation k and c. The quantification indices of model sensitivity exert a certain influence on the ranking of parameter sensitivities. However, for models with a limited number of parameters, the impact of these results is constrained. Conversely, the identification and utilization of characteristics specific to the parameter tuning process can significantly enhance the efficiency of model calibration. The three models employed by the research institute have all demonstrated commendable performance in modeling the canopy interception process of subalpine P. crassifolia in arid, high-altitude regions, achieving a "good" rating with Nash-Sutcliffe Efficiency values exceeding 0.7. In practical applications, we recommend giving priority to the use of the Liu model. The findings of this study provide a reference for model selection, sensitivity analysis, parameter calibration, and model evaluation in the context of extensive canopy interception modeling in arid areas with significant altitudinal variation. This constitutes an important theoretical support for the refined modeling of hydrological processes in high-altitude forests within arid zones.

Improving carbon flux estimation in tea plantation ecosystems: A machine learning ensemble approach

Tea plant (Camellia sinensis) is a major global crop consumed as a drink after water. Quantifying carbon flux, specifically the net ecosystem exchange (NEE), in tea plantations is essential for determining carbon sequestration and ecosystem carbon balance. The Eddy covariance (EC) system is widely used for continuous monitoring of carbon flux but high costs associated with installation and maintenance limit its widespread adoption. In addition, EC flux data is often discarded due to malfunction of instruments caused by adverse weather conditions. Therefore, additional approaches for estimating NEE are necessary to overcome these challenges and ensure accurate NEE measurement. For this purpose, three standalone tree-based machine learning (ML) models were used for NEE estimation using EC flux data collected from tea ecosystem located in subtropical region (Danyang county of Zhenjiang city) of China. To address the accuracy limitations inherent in standalone ML models, the ensemble mechanism based on voting regressor method was proposed. In addition, k-fold cross-validation based on early stopping process was also used to enhance the performance of standalone ML models. Based on visual plots (scatter diagram, heatMap, Taylor diagram) and performing indices (root-mean-square error (RMSE), determination coefficient (R2), mean absolute error (MAE), mean absolute percentage error (MAPE), Nash–Sutcliffe efficiency (NSE), correlation coefficient (r), Kling Gupta Efficiency (KGE) and index of agreement (d)), the findings indicated that non-linear ensemble-generalized regression neural network (NLE-GRNN) significantly improved standalone ML model's results. In current study, the highest NSE, r and d in case of standalone ML model (DT) achieved 0.49, 0.73 and 0.75 respectively while our proposed NLE-GRNN model improved 48 % in NSE value (NSE = 0.97), 25 % in r value (r = 0.98) and 24 % in d value (d = 0.99). Likewise, NLE-GRNN significantly reduce errors (MAE, MAPE and RMSE) and provides NEE estimate closet to the observed value. The impact of climatic variables on NEE using shapley additive explanations (SHAP) analysis revealed that Rg (solar radiation) and Tair (air temperature) were the prime factors controlling NEE variation in the tea ecosystem. Considering the high accuracy and stability of the studied ML models, it is recommended to apply developed ensemble ML model (NLE-GRNN) for significant improvement of NEE estimate in the tea biomes or other ecosystems.

Flood risk assessment using machine learning, hydrodynamic modelling, and the analytic hierarchy process

ABSTRACT The objective of this study was to develop a theoretical framework based on machine learning, the hydrodynamic model, and the analytic hierarchy process (AHP) to assess the risk of flooding downstream of the Ba River in the Phu Yen. The framework was made up of three main factors: flood risk, flood exposure, and flood vulnerability. Hazard was calculated from flood depth, flood velocity, and flood susceptibility, of which depth and velocity were calculated using the hydrodynamic model, and flood susceptibility was built using machine learning, namely, support vector machines, decision trees, AdaBoost, and CatBoost. Flood exposure was constructed by combining population density, distance to the river, and land use/land cover. Flood vulnerability was constructed by combining poverty level and road density. The indices of each factor were integrated using the AHP. The results showed that the hydraulic model was successful in simulating flood events in 1993 and 2020, with Nash–Sutcliffe efficiency values of 0.95 and 0.79, respectively. All machine learning models performed well, with area under curve (AUC) values of more than 0.90; among them, AdaBoost was most accurate, with an AUC value of 0.99.

Quantile mapping technique for enhancing satellite-derived precipitation data in hydrological modelling: a case study of the Lam River Basin, Vietnam

ABSTRACT Accurate precipitation is crucial for hydrological modelling in sparse gauge regions like the Lam River Basin (LRB) in Vietnam. Gridded precipitation data from satellite and numerical models offer significant advantages in such areas. However, satellite precipitation estimates (SPEs) are subject to uncertainties, especially in high variable of topography and precipitation. This study focuses on enhancing the accuracy of Integrated Multi-satellitE Retrievals for Global Precipitation Measurement (IMERG), Climate Prediction Center morphing technique (CMORPH) using the Quantile Mapping (QM) technique, aligning the cumulative distribution functions of the observed precipitation data with those of the SPEs, and assessing the impact on hydrological predictions. The study highlights that the post-correction IMERG precipitation using QM performs better than other data sets, enhancing the hydrological model's performance for the LRB at different temporal scales. Nash–Sutcliffe efficiency values increased from 0.60 to 0.77, surpassing the original IMERG's 0.52 to 0.74, and correlation coefficients improved from 0.79 to 0.89 (compared with the previous 0.75–0.86) for hydrological modelling. Additionally, Percent Bias (PBIAS) decreased from approximately −1.66 to −2.21% (contrasting with the initial −20.22 and 4.6%) with corrected SPEs. These findings have implications for water resource management and disaster risk reduction initiatives in Vietnam and other countries.

Hybridization of deep learning, nonlinear system identification and ensemble tree intelligence algorithms for pan evaporation estimation

A reliable pan evaporation (Epan) estimation over a daily scale is vital for sustainable water and agriculture management, especially for designing water use allocations, irrigation system management, and demand and utilization assessments. However, designing a reliable computational methodology is a challenging task for Epan estimation due to its stochastic and random characteristics. In this study, the potential of five machine learning (ML) models namely Hammerstein-Weiner (HW), Random Forest (RF), Boosted Regression Tree (BRT), Long Short-Term Memory (LSTM), and Stepwise Linear Regression (SWLR) models were evaluated for Epan modeling in Lake Hawassa catchment, Ethiopia. Afterward, hybrid SWLR-LSTM, SWLR-RF, SWLR-BRT, and SWLR-HW models were developed to benefit from the strength of the non-linear and linear characteristics of the models. The estimation results were evaluated using determination coefficient (R2), Nash-Sutcliffe efficiency (NSE), mean absolute error (MAE) and Root Mean Square Error (RMSE), Index of Agreement (IA) and Kling-Gupta efficiency (KGE). The performance analysis result of individual models showed that the RF model led to more accurate estimation with NSE = 0.938, IA = 0.956, R2 = 0.94, KGE = 0.852, RMSE = 0.483 mm/day and Pbias = -1.193 mm/day during the validation phase. Among the hybrid models, SWLR-RF provides the best predictive performance improving SWLR, HW, LSTM, BRT and RF by 26.563 %, 21.348 %, 15.577 %, 9.091 % and 3.514 %, respectively based on the validation phase NSE value. Generally, the results of the current study demonstrated the capability of deep learning, ensemble tree and hybrid linear-nonlinear models for Epan estimation in the data-scarce catchment.

Streamflow Prediction Using Complex Networks.

The reliable prediction of streamflow is crucial for various water resources, environmental, and ecosystem applications. The current study employs a complex networks-based approach for the prediction of streamflow. The approach consists of three major steps: (1) the formation of a network using streamflow time series; (2) the calculation of the clustering coefficient (CC) as a network measure; and (3) the use of a clustering coefficient-based nearest neighbor search procedure for streamflow prediction. For network construction, each timestep is considered as a node and the existence of link between any node pair is identified based on the difference (distance) between the streamflow values of the nodes. Different distance threshold values are used to identify the critical distance threshold to form the network. The complex networks-based approach is implemented for the prediction of daily streamflow at 142 stations in the contiguous United States. The prediction accuracy is quantified using three statistical measures: correlation coefficient (R), normalized root mean square error (NRMSE), and Nash-Sutcliffe efficiency (NSE). The influence of the number of neighbors on the prediction accuracy is also investigated. The results, obtained with the critical distance threshold, reveal that the clustering coefficients for the 142 stations range from 0.799 to 0.999. Overall, the prediction approach yields reasonably good results for all 142 stations, with R values ranging from 0.05 to 0.99, NRMSE values ranging from 0.1 to 12.3, and the NSE values ranging from -0.89 to 0.99. An attempt is also made to examine the relationship between prediction accuracy and the catchment characteristics/streamflow statistical properties (drainage area, mean flow, coefficient of variation of flow). The results suggest that the prediction accuracy does not have much of a relationship with the drainage area and the mean streamflow values, but with the coefficient of variation of flow. The outcomes from this study are certainly promising regarding the application of complex networks-based concepts for the prediction of streamflow (and other hydrologic) time series.

Exploring PCSWMM for Large Mixed Land Use Watershed by Establishing Monitoring Sites to Evaluate Stream Water Quality

Extensive hydrologic and water quality modeling within a watershed benefits from long-term flow and nutrient data sets for appropriate model calibration and validation. However, due to a lack of local water quality data, simpler water quality modeling techniques are generally adopted. In this study, the monitoring sites were established at two different locations to collect hydraulic data for the hydraulic calibration and validation of the model. In addition, water quality samples were collected at eight monitoring sites and analyzed in the lab for various parameters for calibration. This includes total suspended solids (TSS), soluble phosphorus, five-day biochemical oxygen demand (BOD5), and dissolved oxygen (DO). The Personal Computer Storm Water Management Model (PCSWMM) 7.6 software was used to simulate all the pollutant loads using event mean concentrations (EMCs). The performance of the model for streamflow calibration at the two USGS gauging stations was satisfactory, with Nash–Sutcliffe Efficiency (NSE) values ranging from 0.51 to 0.54 and coefficients of determination (R2) ranging from 0.71 to 0.72. The model was also validated with the help of historical flow data with NSE values ranging from 0.5 to 0.79, and R2 values ranging from 0.6 to 0.95. The hydraulic calibration also showed acceptable results with reasonable NSE and R2 values. The water quality data recorded at the monitoring stations were then compared with the simulated water quality modeling results. The model reasonably simulated the water quality, which was evaluated through visual inspection using a scatter plot. Our analysis showed that the upstream tributaries, particularly from agricultural areas, were contributing more pollutants than the downstream tributaries. Overall, this study demonstrates that the PCSWMM, which was typically used for modeling urban watersheds, could also be used for modeling larger mixed land use watersheds with reasonable accuracy.

Interpretable machine learning on large samples for supporting runoff estimation in ungauged basins

The distribution of flowmeter data and basin characteristic information exhibits substantial disparities, with most flow observations being recorded at a limited number of well-monitored locations. The perennial challenge of achieving reliable and robust hydrological modeling in ungauged catchments through regionalization has persisted. The increasing availability of large-scale hydrological datasets, coupled with recent advancements in machine learning techniques, offers new opportunities to explore patterns of association between basin attributes and hydrological parameters to enhance streamflow predictions. A novel parameter cross-regional transfer approach based on interpretable machine learning (XGBoost) is proposed to accurately predict runoff processes in ungauged regions by leveraging well-trained models across numerous basins within climate zones. We validate the effectiveness of this framework across 5,764 basins in a large sample dataset (Caravan), employing Nash-Sutcliffe Efficiency (NSE), RMSE and bias to assess performance. And a comparison is made with deep transfer learning based on LSTM and Transformer. Results indicate that the proposed method achieves NSE values exceeding 0.2 for 75 % of the ungauged basins, demonstrating superior performance and more stable accuracy compared to pure deep learning models, owing to its incorporation of physical constraints. Furthermore, the response of parameters to basin attributes within different climatic zones in the large-sample context is elucidated through SHAP values, enriching the understanding of hydrological features through data-driven inverse inference. These findings underscore the capability of interpretable machine learning to leverage hydro-physical regularities extracted from abundant basin features, thereby enhancing the accuracy of runoff predictions in ungauged regions.

Spatial mapping and driving factor Identification for salt-affected soils at continental scale using Machine learning methods

Soil salinization is one of the most serious land-degrading threats globally, particularly in Asia, where there are 21 countries, e.g., Kazakhstan, China, Iran, and Indonesia, lack of accurate spatial information of salt-affected soils currently. Mapping the distribution of salt-affected soils and identifying their main driving factors is critical for sustainable development. Current studies for mapping the spatial distribution of salt-affected soils focused on recognizing the soil salinity by using the indicator of electrical conductivity of saturated soil paste extract (ECe), while cannot distinguish the soil sodicity. Meanwhile, the absence of uncertainty assessment is another problem for large-scale studies. To address these gaps, we developed an effective spatial prediction model that considered both soil salinity and sodicity using machine learning (ML) models and soil salinization data. This study used ECe, pH, and exchangeable sodium percent (ESP) as indicators, with sample sizes of 22,803, 30,821, and 12,961, respectively, to produce salt-affected soil maps with uncertainty assessments in Asia, and conducted the analysis of the drivers of soil salinization. The ML models involved cubist, random forest, quantile regression forest, quantile regression neural network, and a simple multiple linear regression model for comparison. The results showed that among these models, quantile regression forest exhibited the best performance, with mean absolute percentage error values of 2.54 %∼52.06 %, coefficient of determination values of 0.40 ∼ 0.74, and Nash-Sutcliffe efficiency values of 0.36 ∼ 0.71. The predicted maps, with 1 km resolution, demonstrated minimal uncertainty across over 60.4 % of the study area. Approximately 466.36 million hectares of topsoil (0–30 cm) and 377.68 million hectares of subsoil (30–100 cm) in Asia are salt-affected, mainly concentrated in arid, semi-arid, and coastal regions. Climatic factors (the annual precipitation and maximum temperature), topography (digital elevation model and slope), and salinity index V (SI5) were identified as the main drivers of soil salinization. Overall, this study provides important guidelines for soil salinization management and achieving ecological sustainability.

Calibration of CAMS PM2.5 data over Hungary: a machine learning approach

Air pollution is a major environmental problem, and reliable monitoring of particulate matter (PM) concentrations is critical for assessing its impact on human health and the environment. The Copernicus Atmosphere Monitoring Service (CAMS) offers vital data on PM2.5 concentrations by applying a worldwide modelling system. This study compares in situ PM2.5 measurements and raw CAMS data at 0.1° × 0.1° resolutions for 2019 and 2020 in Hungary. It proposes a calibration method to improve the accuracy of CAMS PM2.5 data at the scale of air monitoring stations. In the study, the accuracy of the raw CAMS PM2.5 data is assessed based on the chosen air quality stations. Then, to improve the precision, we employed machine learning algorithms (LightGBM, Random Forest (RF), and Multiple Linear Regression (MLR)) for calibration. Initial assessment of the raw CAMS PM2.5 data showed positive hourly Spearman correlation coefficient values (SR between 0.64 and 0.87 for the 14 air quality stations used), indicating a positive relationship between the datasets but a systemic underestimation. Our findings highlight LightGBM as the most effective method, consistently demonstrating elevated correlation SR and coefficient of determination R2 values reaching up to 0.95 and 0.93, respectively, and very good RSR (Root mean square error ratio) and NSE (Nash-Sutcliffe Efficiency) values (lower than 0.5 and higher than 0.75 for RSR and NSE, respectively). In contrast, RF yields mixed results, and MLR exhibits variable performance. By correcting underestimation and lowering modelling biases, the calibrated PM2.5 data better matches ground-based observations, which can be promising for using the obtained model for accurate estimation at individual air monitoring stations.

Assessing the accuracy and reliability of satellite-derived precipitation products in the Kosi River basin (India).

The present research endeavors to examine the effectiveness of four gridded precipitation datasets, namely Integrated Multi-satellite Retrievals for GPM (IMERG), Tropical Precipitation Measuring Mission (TRMM), Modern-Era Retrospective Analysis for Research and Applications Version 2 (MERRA-2), and Precipitation Estimation from Remotely Sensed Information using Artificial Neural Networks (PERSIANN), with the observed rainfall data of eight rain gauge stations of India Meteorological Department (IMD) from 2001 to 2019 in Kosi River basin, India. Various statistical metrics, contingency tests, trend analysis, and rainfall anomaly index were utilized at daily, monthly, seasonal, and annual time scales. The categorical metrics namely probability of detection (POD) and false alarm ratio (FAR) indicate that MERRA-2 and IMERG datasets have the highest level of concurrence with the observed daily data. Statistical analysis of gridded datasets with observed dataset of IMD showed that the performance of the IMERG dataset is better than MERRA-2, PERSIANN, and TRMM datasets with "very good" coefficient of determination (R2) and Nash-Sutcliffe Efficiency values for monthly data. Trend analysis of gridded seasonal data of IMERG showed similar trends of observed seasonal data whereas other dataset differs. IMERG also performed well in identifying wet and dry years based on annual data. Discrepancies of the satellite sensor in capturing the precipitation have also been discussed. Thus, the IMERG dataset can be effectively used for hydro-meteorological and climatological investigations in cases of lack of observed datasets.

Enhancing hydrological data completeness: A performance evaluation of various machine learning techniques using probabilistic fusion imputer with neural networks for streamflow data reconstruction

The present-day accessibility of streamflow data, particularly in the developing countries, is often marked by a multitude of data shortfalls or distortions. This study investigates the estimate of missing streamflow data using machine learning approaches, including K-nearestneighbour (KNN), Predictive Mean Matching (PMM), Random Forest (RF) and a novel technique of Probabilistic Fusion Imputer with Neural Networks (PROFINN). This study tackles the issue of data insufficiency in such time series considering the inclusion of numerous hydrological parameters by means of RF feature selector, to determine the most significant ones among them. The study explores the efficacy of selected models on various data gaps under different hydrological scenarios, including diverse flow characteristics (mean, high and low flows) and gap lengths (long and short gaps, continuous and discontinuous gaps) and presents a pattern of ranking system that assesses the level of suitability of each technique for various data gaps. This study underscore PROFINN’s remarkable performance across all scenarios, yielding an average Root Mean Square Error (RMSE) of 0.91 and an average Nash-Sutcliffe Efficiency (NSE) value of 0.93 when applied for an intermittent river system of Pamba in Southern Kerala, India. RF follows PROFINN in the imputation of extreme flows as well as long and short gap scenarios. KNN closely follows PROFINN for the imputation of continuous and discontinuous gap scenarios. This study augments the significance of tailored machine learning techniques in enhancing the integrity of hydrological datasets, offering valuable insights for effective decision-making in water resource management and related fields. Furthermore, the success of PROFINN in streamflow data imputation suggests its potential extension to other hydrological research domains, like historical data reconstruction assisting climate change studies and future water resources planning, emphasizing the broader relevance and applicability of this study’s findings.

Sensitivity Analysis of the Inverse Distance Weighting and Bicubic Spline Smoothing Models for MERRA-2 Reanalysis PM2.5 Series in the Persian Gulf Region

Various studies have proved that PM2.5 pollution significantly impacts people’s health and the environment. Reliable models on pollutant levels and trends are essential for policy-makers to decide on pollution reduction. Therefore, this research presents the sensitivity analysis of the Bicubic Spline Smoothing (BSS) and Inverse Distance Weighting (IDW) models built for the PM2.5 monthly series from MERRA-2 Reanalysis collected during January 2010–April 2017 in the region of the Persian Gulf, in the neighborhood of the United Arab Emirates Coast. The models’ performances are assessed using the Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE). RMSE, Mean Bias Error (MBE), and Nash–Sutcliff Efficiency (NSE) were utilized to assess the models’ sensitivity to various parameters. For the IDW, the Mean RMSE decreases as the power parameter increases from 1 to approximately 4 (the optimal beta value) and then stabilizes with a further increase. NSE values close to 1 indicate that the model’s predictions are very efficient in capturing the variance of the observed data. NSE is almost constant as a function of the number of neighbors and the parameter when β > 4. In BSS, the RMSE and NBE plots suggest that incorporating more points into the mean calculation for buffer points leads to a general decrease in model accuracy. Moreover, the MBE plot shows that the mean bias error initially increases with the number of points but then starts to plateau. The increasing trend suggests that the model tends to systematically overestimate the PM2.5 values as more points are included. The leveling-off of the curve indicates that beyond a certain number of points, the bias introduced by including additional points does not significantly increase, suggesting a threshold beyond which further inclusion of points does not markedly change the mean bias. It was also proved that the methods’ generalizability may depend on the dataset’s specific spatial characteristics.