streamflow

In recent years, the use of artificial intelligence methods, such as artificial neural network models, have become increasingly prevalent in simulating complex natural phenomena, including daily streamflow. The streamflow directly correlates with flood occurrences, and mitigating financial and human losses due to floods is crucial. Accurate streamflow simulation is essential for water resource management and river management. Consequently, in hydrology, deep learning methods have emerged as novel tools to address the longstanding challenge of daily streamflow modeling and are widely used in simulations.Advancements in streamflow modeling with Artificial Intelligence (AI): In recent years, the field of hydrology has witnessed a significant shift toward leveraging AI techniques for streamflow modeling. Among these methods, artificial neural network (ANN) models have gained prominence due to their ability to capture complex relationships within hydrological systems. Streamflow, which represents the flow of water in rivers and streams, is a critical variable for understanding water availability, flood risk, and ecosystem health. By accurately simulating streamflow, researchers and water resource managers can make informed decisions regarding water allocation, flood preparedness, and environmental conservation. Hydrological processes are inherently nonlinear and influenced by various factors such as precipitation, temperature, land cover, and soil properties. Traditional hydrological models often struggle to capture these complexities. However, deep learning methods, including convolutional neural networks (CNNs) and recurrent neural networks (RNNs), offer promising solutions. These models can learn intricate patterns from historical streamflow data, adapt to changing conditions, and provide accurate predictions. As a result, they have become indispensable tools for addressing the longstanding challenge of daily streamflow modeling. Researchers continue to explore novel architectures, data augmentation techniques, and hybrid approaches to enhance the performance and robustness of AI-based streamflow simulations. In summary, the integration of deep learning methods into hydrological research has revolutionized streamflow modeling, enabling more accurate predictions and informed decision-making in water management and flood risk assessment.

In this study, we focused on selecting an appropriate input scenario for deep learning models and simulate daily streamflow on the Kashkan River using LSTM (Long Short-Term Memory) and GRU (Gated Recurrent Unit) deep learning methods. Prior to this, deep learning modeling with the GRU approach using native streamflow measurements had not been performed for Kashkan river. The study area is a flood-prone and mountainous region, specifically the western part of Iran, where a hydrological station with a history of flood events is situated on the Kashkan River. We employ four approaches for handling outliers (Mahalanobis, critical interval removal, Z-Score, and no removal) and four different preprocessing techniques for input data to train two models: LSTM and GRU. Ultimately, eight distinct models are generated and validated against historical data. The input features include regional average precipitation, normalized vegetation cover index, surface soil moisture, groundwater flow, and the Kashkan River’s own flow at the hydrological station, with the best features selected using statistical correlation control.

The results demonstrate that among the deep learning models generated with a 10-day time step, the model with the least error and consistent low error retention in error metrics is observed. Furthermore, the best performance is achieved using different approaches, in the following order: the GRU model with Z-Score-corrected inputs, followed by the Mahalanobis removal approach with average RMSE (Root Mean Square Error) and KGE (Kling-Gupta Efficiency) values of 5.41 and 0.99, respectively, and the critical interval removal approach with RMSE of 6.23 and KGE of 0.7.The results showed that among the deep learning models produced with a time step of 10 days in the model, the lowest amount of error and the persistence of low error can be seen in the error statistics, and among the different approaches used, the best performance is the GRU model with input modified by Z-Score elimination of outlier method, Mahalanobis elimination method with average RMSE and KGE values of 5.41, 0.99, 6.23, and 0.7 in the training phase and 8.17, 0.79, 4.21, and 0.81 in the validation phase and 5.01, 0.68, and 7.21 and 0.52 are in the testing phase. The obtained results do not reject the LSTM method in simulating the river flow, but state that the listed scenarios, especially in the GRU method, have a higher power in dealing with the data and recognizing the complex pattern of daily river flow, taking into account the limitation in use They have seven years of regular daily data, and future research will show how the behavior of GRU and LSTM models will differ if data with higher convergence is used.

GRU in future studies can make difference by enhanced flood forecasting accuracy, efficient computation and real-time applications, integration with lag time preprocessing, adaptability to changing climate and urbanization. Future studies will be on data driven method in flood prone areas. There remains ample room for future research and innovation. Here are some directions for further exploration: hydrological data fusion, spatially explicit models, uncertainty quantification, climate change resilience.

Quantifying the impact of climate change and human activities on hydrological processes is of great importance for regional water-resource management. In this study, trend analysis and analysis of the short-term variations in annual streamflow in the Sajasrood Watershed (SW) during the period 1987–2012 were conducted using linear regression and the Pettitt test. The time series for streamflow in SW exhibits declining trends. The results show that its break point is 1998, so that the streamflow can be divided into the baseline period and the evaluation period. The Soil and Water Assessment Tool (SWAT) was employed to simulate the hydrological processes. SWAT was first calibrated at monthly time step. The NSE, PBIAS, and RSR for the calibration period (1988-1992) were 0.7, 7.3, and 0.55, respectively. The NSE, PBIAS, and RSR for the validation period (1993-1995) were 0.65, -9.4, and 0.58, respectively. The results suggest that SWAT performance was good during calibration and validation periods. The results show that the contribution rates of climate change and human activities to runoff are 33% and 67% during 1999-2012. The proposed framework can help policy makers to regulate the policies accordingly so that the river may restore to a better level and ecosystem may get improved.ر

In the midst of extreme events, the drought is occurring gradually and taking its effects during the longer time. Among the drought monitoring methods, Nalbanti proposed SDI method based on the analysis of drought characteristics based upon the volume of the cumulative river. Being easy is of this method’s advantages. This method has been used to analyze the drought characteristics of many countries such as the United States, India, Iran, Iraq, and Greece. The mean of average annual precipitation is 437 mm in Kermanshah province which has been increased in recent years during the humid season and has been decreased during the dry season, for climate changes and temperature increase. The difference in river volume between the humid and dry seasons has led to the drought. This research aims to investigate the continuity and severity of hydrological drought via using Streamflow Drought Index (SDI) based on the 3, 6, 12, 18 and 24-month periods in this province. Daily Dubai data was used from 10 hydrometric stations for a 31-year period (1981-2011) in order to achieve this goal. The result of the river flow index showed that wet years began in 1363 and continued until 1998 at the stations of Totshami, Shah Gozar, Arangharb, Doab, Polchehr, Khers Abad, DuabMerk, and Ghorbaghestan. And the drought began since 1998 at the stations of Totashi, Arangharb, Plochehr, Khers Abad, DuabMerk and Ghorbaghistan and continued till the recent years (2011).

This study evaluates the application of Atmosphere-Land Surface Interaction System (ALSIS) scheme in simulating the streamflow in Karkheh river basin. The Climate Forecast System Reanalysis (CFSR) data for the period 1982-2011 are used as atmospheric forcing data and sub-grid scale heterogeneity of the land-surface is represented by soil-vegetation mosaics. The cascade of linear reservoirs model is used for modelling the base flow and a routing model, linked to the land surface scheme, is used for modelling river discharge. The comparison of simulated and observed streamflow in six hydrometric stations over Karkheh basin reveals the model ability in simulating the monthly streamflow. Moreover, the model has a good ability in simulating the monthly regime of water balance components, spatial distribution of long-term average of components and their relationships.

Nowadays, water scarcity and frequent occurrence of droughts in Middle East countries including Iran create many problems for population of these regions. Drying off Salty Lake Urmia and its consequences is very important form the Iran and Middle East water industries. The main reason of Lake Urmia crisis is streamflow reduction of rivers leading to the lake due to excessive extraction of water resources in recent decades. Since the river-aquifer is a system with two interrelated elements, therefore, investigation of streamflow coefficient versus groundwater table relationship is necessary for streamflow forecasting. In this study, instantaneous runoff coefficient (events) and water table fluctuation relationship was checked out in two hydrometric stations located at the beginning and ending sections of Ghaleh-Chayriver which discharges water to the Lake Urmia. Then, the mechanism of stream flow loss (flow reduction process) due to water level draw down was studied along the river profiles. The results of this study showed a direct relation between the runoff coefficient and the water table level. Correlation coefficient value (R) between these two factors was observed 0.9 for Yengejeh station located at the beginning section of the river and 0.84 for Shishavan station at the ending section of river. Evaluating the stream flow losses trend along the river indicates that this flow has been naturally diverted to saturate the transition zone (the zone between the water table level and stream bed).

In the past decade the different methods have been used to analyze and predict the physical variables, one of which is singular spectrum analysis (SSA) statistical methods. SSA is one of the methods, used in modeling various statistical processes and more recently, its use in various engineering disciplines including water resources, in order to eliminate random components in time series has been expanded. The main objective of this study was to forecast streamflow in the Karkheh basin using singular spectrum analysis. The gage stations in the Karkheh basin (five station) were selected for this study. The high flow period for these gage stations were determined. In order to modeling methods, 70% and 30% of data were used for calibration and validation respectively. The singular spectrum analysis method was used for pre-processing of data and elimination of noise in the time series of streamflow. Then, the recursive algorithm of the singular spectrum analysis model was used to develop forecasts models of streamflow in the Karkheh basin gage stations. To evaluate the performance of the model Normalized root mean square error, mean absolute error and correlation coefficient were used. In the validation the highest and lowest value of the NRMSE and MARE statistics were 0.47 and 0.50 for Pol Chehr station. The lowest value of the NRMSE statistic for Pol Dokhtar and Cham Anjir stations was 0.3 and 0.31 respectively and close to each other and the lowest value of the MARE statistic for Cham Anjir and Pol Dokhtar stations was 0.29 and 0.30 respectively and close to each other. Finally, the best and the weakest results in two stages of calibration and validation were for Cham Anjir and Pol Chehr Stations respectively. The results of this research showed that singular spectrum analysis can be used to forecast streamflow with reasonable accuracy

Streamflows, actual evapotranspiration and crops’ yield are the main variables to estimate agricultural water productivity. Thus, simulation of these variables is of great importance in evaluation of different measures to increase water productivity. For this, application of conceptual models is a relevant approach and SWAT (soil and water assessment tool) is one of the well known models in this regard. The present paper aims to assess SWAT in simultaneous simulation of streamflows, actual evapotranspiration and the main crops’ yield of the Zarineh Rud basin. The reason for selection of this basin as the study area relates to its role to meet the Urmia Lake’s water requirement. The lake faces with serious water shortage in recent years and escalating water inflow depend to increase water productivity in the upper catchments. To setup SWAT, the basin was divided to 11 subbasins and 908 hydrological response units, which enables us to introduce more accurately the basin’s cropping pattern and water resources, which meet the requirements of the agricultural area. For simulation of the river flows, data from 6 gauging stations were used for calibration and validation of the model for periods of 1987 to 1999 and 2000 to 2006 respectively that resulted R2 and RMSE between 0.49 to 0.71 and 3.9 to 44.9 (m3/sec) for calibration period, and values of 0.54 to 0.77 and 2.07 to 55.7(m3/sec) for validation period respectively. There is no observed data for actual evapotranspiration in the basin. So, it was verified in the wet years by maximum evapotranspiration reported in National Water Document that results presented the values of 0.97 and 52.5(mm/year) for R2and RMSE respectively. Finally, the estimated yields of the 7 staple crops by the model were compared with the recorded data that showed very close values(R2=0.9 and RMSE=1.65(ton/ha)).

The application of chaos theory and genetic programming has been gained a special attention in hydrology by abilities of these two models. In this study, a daily streamflow series with 30 years records of Lighvan River has been studied using these models. First in chaos theory, possibility of the existence of deterministic chaos in the daily streamflow series has been investigated by employing the correlation dimension method and after determination of the necessary parameters for reconstruction of phase space; daily streamflow is predicted by the use of local prediction method. To reconstruct the data pertaining to phase space, the time delay and embedding dimension are needed. For this purpose, autocorrelation function and algorithm of false nearest neighbors have been used and the amount of the obtained correlation dimension expresses chaotic behavior in the time series under investigation. Finally, local prediction method was used for prediction of the daily discharge and the results illustrate acceptable accuracy of this theory. The genetic programming has been used for daily discharge modeling and the best input pattern consists of antecedent discharge with four time lags has also been selected. The obtained correlation coefficient being equal to 0.926 from local prediction method and 0.931 from genetic programming indicate good accuracy and similar results obtained from both methods for streamflow prediction. Thus, according to the obtained results, both methods can be used for streamflow prediction and modeling.

The aim of this study is trend analysis of streamflows of east Azarbayejan in three time scales, which are monthly, seasonal and annual using the non-parametric methods. For this purpose the information of discharges of ten hydrometric stations (1983 -2008) were used. The Mann-Kendal test was used after removing the all significant autocorrelation effects from the data. In order to estimate the slope of trend line the Sen,s estimator method was used.Results showed that about 30% of the stations showed statistically significant negative trends in annual time scale. The strongest negative trend line slope belonged to the Vanyar station having the slope of 0.35 (m3/s/year) in annual time scale. In seasonal time scale, negative trends observed at almost all the stations. In monthly time scale streamflow trends were negative for most of the stations and none of the stations experienced positive significant trends (even at 10% level). The strongest negative trend observed in Tazekand (Soofi-Chay), which was significant at 10% level.

The hydrological variables estimation is very important for safety of hydraulic structures and facilities، surface water resources management and planning. The streamflow measurement as one of the important hydrological variables has various limitations. For this reason، application of the hydrological models with available and easily measured variables is the important and necessary approach. The aim of this research is the possibility of available climatic variables e. g. precipitation and temperature for seasonal and annual discharge estimation. Data of precipitation، temperature and discharge related to upstream watersheds of Boukan reservoir dam were used for 16 years period during 1993-2007 in this study. Multiple and stepwise regression models were determined in annual and seasonal scales for each watersheds using SPSS 17 software، then calibration and validation of each models were examined. Root mean squared error (RMSE)، relative error (RE) and efficiency coefficient (CE) were used for models evaluation. The results showed that the most models performance was very good and stepwise regression has more accurate results than multiple regression. Also، the RE، RMSE and CE range criteria were changed between 20. 96 to 44. 69 and 22. 80 to 81. 11 percent، 0. 37 to 7. 37 and 0. 05 to 7. 53 m3 s-1 and 0. 7 to 0. 94 and 0. 31 to 0. 92 for annual and seasonal scale، respectively. Ratio of precipitation influence on discharge was very low in summer in comparison to other seasons because annual model will not be accurate for estimating discharge in summer.