Seyed Masoud Soleimanpour; Omid Rahmati; Samad Shadfar; Maryam Enayati
Abstract
Field measurements of soil loss due to gully erosion are very time-consuming and costly, so direct measurement of gully erosion at large scales is a time-consuming, costly, and labor-intensive process. For this purpose, the present study attempted to accomplish this by modeling soil loss due to gully ...
Read More
Field measurements of soil loss due to gully erosion are very time-consuming and costly, so direct measurement of gully erosion at large scales is a time-consuming, costly, and labor-intensive process. For this purpose, the present study attempted to accomplish this by modeling soil loss due to gully erosion using random forest and support vector machine learning models and evaluating their efficiency in the Mahurmilati watershed located in the southwest of Fars province. Field measurements of dimensional parameters of 70 gullies were conducted over four years (2021 to 2024). In the modeling process, 15 environmental factors were considered as independent variables and the rate of soil loss in ditches as the dependent variable, and modeling was performed with a cross-validation approach. The accuracy of the models was evaluated using quantitative criteria such as root mean square error (RMSE), coefficient of determination (R2), root mean square error (RSR), and correlation coefficient (d). The rate of soil loss in gullies during the study period was 15300.94 tons. The results of the model prediction accuracy evaluation showed that the random forest model has better performance than the support vector machine model in terms of evaluation criteria and was introduced as the superior model for predicting the rate of soil loss due to gully erosion. The findings showed that "modeling" can provide valuable services to water and soil conservation management in saving time and money. For this purpose, it is suggested that the use of artificial intelligence-based models and machine learning structures be given more attention in future research.
Alireza Sepahvand; Nasrin Beiranvand; Negar Arjmand
Abstract
Water quality (WQ) is influenced by various variables, including natural ones like rainfall and erosion and human ones like urban, agricultural, and industry operations, that plays a very important role in assessment and determining factors such as environmental conditions, public health, economic and ...
Read More
Water quality (WQ) is influenced by various variables, including natural ones like rainfall and erosion and human ones like urban, agricultural, and industry operations, that plays a very important role in assessment and determining factors such as environmental conditions, public health, economic and social progress and development. Therefore, temporal and spatial trending of water quality is necessary for planning water resource management. In this research, the performance of the six soft computing techniques, including, Random Forest, Reduced Error Pruning Tree (REPt), M5P model, bagging RF, bagging REPt and bagging M5P were compared to estimate the water quality index (WQI) in Khorramabad, Biranshahr and Alashtar sub-watersheds, Lorestan province, Iran. At first, based on water quality data, water quality index (WQI) was calculated and ten distinct water quality parameters (2014 to 2023) were used as input variables and WQI as output. Total data set consists of water quality parameters of three sub-watersheds out of which 70% data used to training and 30% data were used to testing phase. Finally, the models were compared with Correlation Coefficient (C.C.), Root Mean Square Error (RMSE), Maximum Absolute Error (MAE), Taylor diagram and Violin plot box. The obtained results suggest that the BM5P is more accurate to estimate the water quality index (WQI) compared to the M5P, ReepTree and Random Forest (RF) models for the given study area. According to the results of the test part of the BM5P model, it has given us the best result, which are the correlation coefficient, the Root Mean Square Error and the Mean Absolute Error 0.99, 0.2, and 0.15, respectively. Also, the Taylor diagram and violin box plot were concluded that BM5P was the most reliable soft computing technique for the prediction of WQI. Finally, the structure of Artificial Intelligence Techniques (AIT) for modeling is very simple and very less time consumable. Thus, the BM5P model can be useful in the water quality index (WQI) modeling not only for accuracy but also for its time-saving and simple structure compared with other models.
Mohammad Ansari Ghojghar; Masoud Pourgholam-Amiji; Shahab Araghinejad; Banafsheh Zahraie; Saman Razavi; Ali Salajegheh
Abstract
Due to the growing development of meta-models and their combination with optimization algorithms for modeling and predicting meteorological variables, in this research four metaheuristic optimization algorithms of Particle Swarm Optimization (PSO), Genetics Algorithms (GA), Ant Colony Optimization for ...
Read More
Due to the growing development of meta-models and their combination with optimization algorithms for modeling and predicting meteorological variables, in this research four metaheuristic optimization algorithms of Particle Swarm Optimization (PSO), Genetics Algorithms (GA), Ant Colony Optimization for Continuous Domains (ACOR) and Differential Evolutionary (DE) were combined with the adaptive neural-fuzzy inference system (ANFIS) model. The performance of four combined models developed with ANFIS model to predict the Frequency variables of Dust Stormy Days (FDSD) on a seasonal scale in Khuzestan province in the southwest of Iran was evaluated. For this purpose, hourly dust data and codes of the Word Meteorological Organization were used on a seasonal scale with a statistical period of 40 years (1980-2019) in seven synoptic stations of Khuzestan province. The results of good fit indices in the training and testing phase showed that there is no significant difference between the ANFIS method and other combined models used. R and RMSE values of the best combined model (ANFIS-PSO) from 0.88 to 0.97 and 0.10 to 0.19, respectively, and in the ANFIS model from 0.83 to 0.94 and 0.11 to 21, respectively, were variable. The results also showed that the combination of optimization algorithms used with the ANFIS model does not significantly improve the results of the model compared to the individual ANFIS model.
Zhila Ghorbani; Kiomars Sefidi; Farshad Keivan Behjou; Mehdi Moameri; Ali Ashraf Soltani Tolarod
Abstract
The most current way for measuring the soil fragmentation is determination of mean weight diameter (MWD). In this study, the adaptive neuro-fuzzy inference system (ANFIS) was used to predict of range soil fragmentation affected by different grazing intensities, distance from village and sampling depth. ...
Read More
The most current way for measuring the soil fragmentation is determination of mean weight diameter (MWD). In this study, the adaptive neuro-fuzzy inference system (ANFIS) was used to predict of range soil fragmentation affected by different grazing intensities, distance from village and sampling depth. Present study conducted at 2015 in 3 adjacent rural areas (Alvars, Aldashin and Asbe marz) in Darvishchai watershed in Ardabil County. The studied parameters on the soil fragmentation including different grazing intensities in 3 levels (low, medium and high intensity), distance from village in 3 levels (200, 400 and 600 meters) and the soil sampling depths in 2 levels (0-15cm and 15-30cm). Obtained data were transferred to MATLAB software for the development of ANFIS models. For evaluating the models operation, mean squares error (MSE) and correlation (R2) were used. The result of best ANFIS model in prediction of soil fragmentation was compared with results of regression model. The results show that different grazing intensities, distance from village, sampling depth and their combinations had significant effect on the soil fragmentation. Increase of grazing intensity resulted in increment of soil fragmentation. With increment the distance from village from 200 to 400 meters, soil fragmentation decreased but with increment of distance, increased. Soil fragmentation in all conditions was higher at depth of 0-15 cm than depth of 15-30 cm. ANFIS model had more precision in prediction of soil fragmentation (R2=0.96) relative to regression model (R2=0.76).
Maryam Asadi; Ali Fathzadeh
Abstract
Understanding of suspended sediment rate is one of the fundamental problems in water projects which water engineers consistently have involved with it. Wrong estimations in sediment transport cause incorrect design and destruction of hydraulic systems. Due to the difficulty of suspended sediment measurements, ...
Read More
Understanding of suspended sediment rate is one of the fundamental problems in water projects which water engineers consistently have involved with it. Wrong estimations in sediment transport cause incorrect design and destruction of hydraulic systems. Due to the difficulty of suspended sediment measurements, sediment rating curves is considered as the most common method for estimating the suspended sediment load. The main purpose of this research is the capability challenge of this method in comparison to some state of the art models. In this study, we selected some computational intelligence models (i.e. K-nearest neighbor (KNN), artificial neural networks (ANN), Gaussian processes (GP), decision trees of M5, support vector machine (SVM) and evolutionary support vector machine (ESVM)) and compared them with their sediment rating model in 8 basins located in Gilan province. Daily sediment and discharge data considered as the input data for 30-years. Evaluation of the results indicated that the Gaussian process model has the lowest residual sum of squares (RMSE) and the highest correlation coefficient (r) than the other models.
