نوع مقاله : مقاله پژوهشی
نویسندگان
1 استادیار، گروه مهندسی احیاء مناطق خشک و کوهستانی، دانشکده منابع طبیعی، دانشگاه تهران، کرج، ایران.
2 دانشجو دکتری، گروه مهندسی احیاء مناطق خشک و کوهستانی، دانشکده منابع طبیعی، دانشگاه تهران، کرج، ایران.
چکیده
کلیدواژهها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
This study aims to assess and compare the performance of the Extreme Learning Machine (ELM) rapid learning algorithm with the open-source numerical model XBeach across ten synoptic stations in Lorestan Province (including Nurabad, Alishtar, Borujerd, Kuhdasht, Khorramabad, Pol-e Dokhtar, Nojian, Dorud, Azna, and Aligudarz) over a 50-year period (1971–2020), with the objective of proposing a hybrid XBeach-ELM method for predicting the Frequency of Dust Storm Days (FDSD) index. The findings revealed a statistically significant and meaningful difference in the modeling results when using the hybrid XBeach-ELM approach compared to other methods examined. The hybrid XBeach-ELM method outperformed the ELM and XBeach models, showing the lowest values for the NRMSE and MAPE error metrics. A t-test comparison of the observed and predicted mean values confirmed the acceptance of the null hypothesis, indicating no significant difference between the observed and predicted time series for the FDSD index when applying the hybrid XBeach-ELM method in Lorestan Province. This equivalence was not observed with the individual ELM or XBeach models. These results suggest that only the hybrid model effectively preserved the mean of the observed time series in predicting the FDSD index. The outcomes of this study have substantial implications for the enhancement of early warning systems, enabling more accurate dust storm forecasting, reducing human and economic losses, supporting urban and infrastructure planning to bolster resilience in high-risk areas, informing local and national policy-making on dust storm management, and advancing the development of sustainable solutions to improve the accuracy of predictive models.
کلیدواژهها [English]
EXTENDED ABSTRACT
Dust storms are recognized as one of the most serious environmental hazards in arid and semi-arid regions worldwide. These storms not only affect human health but also have harmful impacts on wildlife and vegetation. The human health effects of dust storms include increased air pollution, the spread of diseases such as asthma and other respiratory issues, disruption of economic activities, damage to machinery and equipment, contamination of water resources, and gastrointestinal illnesses. Wind speed is the main factor responsible for the formation of dust storms. When wind speed exceeds the threshold for soil erosion, a significant amount of particles is lifted from deserts and dry areas into the atmosphere, causing problems for surrounding regions. Therefore, modeling dust storms is essential for understanding the complex mechanisms behind their formation, spread, and consequences. Numerical models and advanced analytical methods play an important role in simulating the dynamic behavior of dust storms and are crucial for predicting their temporal and spatial variations. While many individual, dual, and triple hybrid models have been used to accurately model the Frequency of Dust Storm Days (FDSD) index, no studies have combined a numerical model with artificial intelligence models for dust storm prediction. Thus, this study, for the first time, combines a rapid learning algorithm with an open-source numerical model to model dust storms and compares the performance of the Extreme Learning Machine (ELM) algorithm with the open-source XBeach model.
This study compares the performance of the Extreme Learning Machine (ELM) rapid learning algorithm with the open-source XBeach model to propose a hybrid XBeach-ELM method, applied to ten synoptic stations in Lorestan Province (including Nurabad, Alishtar, Borujerd, Kuhdasht, Khorramabad, Pol-e Dokhtar, Nojian, Dorud, Azna, and Aligudarz), as shown in Figure 1. The analysis covers a 50-year long-term period (1971–2020). Hourly data on horizontal visibility and World Meteorological Organization (WMO) codes were used for this purpose. A day with a dust storm is defined as a day when, in at least one of the eight reporting periods, one of the dust-related codes (06, 07, 08, 09, 30, 31, 32, 33, 34, 35, or 98) appears in the weather report. This is assuming that the corresponding horizontal visibility data for the reported code is less than 1000 meters. In this study, a horizontal visibility threshold of ≤1000 meters was used for all dust-related codes to identify dust storm days. Extreme Learning Machine (ELM) is an advanced variant of feedforward neural networks, recognized for its fast training speed, which significantly reduces computational time compared to traditional methods. Unlike conventional approaches that rely on iterative optimization of weights (such as backpropagation), ELM randomly initializes the hidden layer weights and determines the output layer weights by solving a matrix equation. This method is widely employed in various practical fields such as classification, prediction, and pattern recognition, thanks to its efficiency, minimal tuning requirements, and satisfactory accuracy. XBeach is an open-source numerical model initially developed to simulate the effects of hydrodynamic processes on sandy coastlines. It has since been extended to other coastal types and applications. The model is capable of simulating coastal bed changes using bed update equations, which are based on sediment changes and the movement of sand driven by currents and waves. This allows XBeach to simulate the effects of erosion and deposition caused by storms and strong waves, as well as predict morphological changes in the coastal bed.
The evaluation indices R, NRMSE, MAPE, and NS were used to assess and compare the performance of the XBeach, ELM, and XBeach-ELM models in predicting the frequency of dust storm days in Lorestan province over a 50-year period. The results indicate that the hybrid XBeach-ELM method provides a much higher accuracy in estimating the frequency of dust storm days across ten stations in Lorestan compared to the ELM and XBeach models. The results from the ELM and XBeach models were almost identical, as there was no significant difference between the two. The results demonstrate that the ELM and XBeach models, when used individually, were not particularly effective in predicting the FDSD index, as their predicted values deviated considerably from the observed values. However, the hybrid XBeach-ELM method succeeded in reducing the gap between the predicted and observed values, significantly improving the accuracy of dust storm predictions. This method has proven to be a valuable tool for simulating and forecasting dust storms in Lorestan province. These findings emphasize the significance of employing hybrid approaches in modeling dust storms—complex and nonlinear phenomena—and could play a crucial role in enhancing disaster management and mitigating the impact of dust storms.
All authors contributed equally to the conceptualization of the article and writing of the original and subsequent drafts.
Data available on request from the authors.
The authors would like to thank the reviewers and editor for their critical comments that helped to improve the paper. The authors gratefully acknowledge the support and facilities provided by the Department of Reclamation of arid and mountainous regions, Faculty of Natural Resources, University of Tehran, Iran.
The authors avoided data fabrication, falsification, plagiarism, and misconduct.
The author declares no conflict of interest.
Abouhalima, M., das Neves, L., Taveira-Pinto, F., & Rosa-Santos, P. (2024). Machine Learning in Coastal Engineering: Applications, Challenges, and Perspectives. Journal of Marine Science and Engineering, 12(4), 638.
Al-Haija, Q. A., Altamimi, S., & AlWadi, M. (2024). Analysis of Extreme Learning Machines (ELMs) for intelligent intrusion detection systems: A survey. Expert Systems with Applications, 124317.
Ansari ghojghar, M. and Pourmohammad, P. (2024). Dust Storm Modeling Using the Triple Hybrid Model GRNN-SVM-LSTM (Case Study: Sistan and Baluchestan Province). Iranian Journal of Soil and Water Research, 55(10), 1961-1979. doi: 10.22059/ijswr.2025.385882.669844. (In Persian).
Anselin L, Syabri I, Kho Y (2009) An Introduction to Spatial Data Analysis. In Fischer MM, Getis A (eds) Handbook of applied spatial analysis, Berlin, Heidelberg and New York, Springer.
Atay H, Ahmadi F (2010) Study of dust as one of the environmental problems of the Islamic world. Case study of Khuzestan province. The Fourth International Congress of Geographers of the Islamic World. Zahedan 138.
Berard, N. A., Mulligan, R. P., da Silva, A. M. F., & Dibajnia, M. (2017). Evaluation of XBeach performance for the erosion of a laboratory sand dune. Coastal Engineering, 125, 70-80.
Bolle, A., Mercelis, P., Roelvink, D., Haerens, P., & Trouw, K. (2011). Application and validation of XBeach for three different field sites. Coastal Engineering Proceedings, (32), 40-40.
Buttò, S., Faraci, C. L., Corradino, M., Iuppa, C., Colica, E., & Pepe, F. (2024). Evaluating longshore sediment transport: A comparison between empirical formulas and XBeach 2DH numerical model. Marine Geology, 107471.
Dargahian, F. , Jalili, A. , Balalan Fard, A. and Aghazadeh, T. (2024). Determining the potential of dust production in the inner centers of Khuzestan province using the WRF numerical model and the WRF-Chem coupled model. Journal of Natural Environmental Hazards, 13(42), 123-140. doi: 10.22111/jneh.2024.47865.2026. Eng. 56 (11–12) (2009) 1133–1152. (In A comparison of the ELM algorithm and XBeach model Persian)
Gelvez-Almeida, E., Mora, M., Barrientos, R. J., Hernández-García, R., Vilches-Ponce, K., & Vera, M. (2024). A Review on Large-Scale Data Processing with Parallel and Distributed Randomized Extreme Learning Machine Neural Networks. Mathematical and Computational Applications, 29(3), 40.
Harter, C., & Figlus, J. (2017). Numerical modeling of the morphodynamic response of a low-lying barrier island beach and foredune system inundated during Hurricane Ike using XBeach and CSHORE. Coastal Engineering, 120, 64–74. https://doi.org/10.1016/j.coastaleng.2016.11.005.
Hayawi, H., Al-Hashimi, M., & Alawjar, M. (2024). Machine learning methods for modelling and predicting dust storms in Iraq. Statistics, Optimization & Information Computing.
Kolokythas, G. A., Raquel, S., & Blanco, M. R. D. (2016, June). Morphological evolution of a bed profile induced by a storm event at the Belgian coast predicted by Xbeach model. In ISOPE International Ocean and Polar Engineering Conference (pp. ISOPE-I). ISOPE.
Maleki, S. , Adjami, M. and Ahmadi, A. (2024). Numerical Modeling of Swash Zone Morphological Processes in Coarse-Grained Beaches with XBeach Open-Source Model. Sharif Journal of Civil Engineering, (), -. doi: 10.24200/j30.2024.63280.3266. (In Persian).
Mei, S., You, W., Zhong, W., Zang, Z., Guo, J., & Xiang, Q. (2024). Optimizing the Numerical Simulation of the Dust Event of March 2021: Integrating Aerosol Observations through Multi-Scale 3D Variational Assimilation in the WRF-Chem Model. Remote Sensing, 16(11), 1852.
O’Loingsigh, T., McTainsh, G. H., Tews, E. K., Strong, C. L., Leys, J. F., Shinkfield, P., & Tapper, N. J. (2014). The Dust Storm Index (DSI): a method for monitoring broadscale wind erosion using meteorological records. Aeolian Research, 12, 29-40.
Roelvink, D., & Costas, S. (2019). Coupling nearshore and aeolian processes: XBeach and duna process-based models. Environmental Modelling & Software, 115, 98-112.
Roelvink, D., Reniers, A., Van Dongeren, A. P., De Vries, J. V. T., McCall, R., & Lescinski, J. (2009). Modelling storm impacts on beaches, dunes and barrier islands. Coastal engineering, 56(11-12), 1133-1152.
Tofighi, F. , Armani, P. , Chehrazi, A. and Alimoradi, A. (2023). Comparison of the function of ELM and RBF models for estimating the porosity of the Asmari Formation, in one of the offshore fields of the northwest Persian Gulf. Journal of Stratigraphy and Sedimentology Researches, 39(2), 45-58. doi: 10.22108/jssr.2023.137083.1256. (In Persian).
Turkashvand M, Kayani Gh (2017) Analysis of Air Pollution Situation Due to the Climatic Effects of Fragrances and Hurricanes in the Southern Regions of Hamadan Province. Journal of Environmental Science and Technology, 19 (4).
Yarmohamadi, M., Alesheikh, A. A., & Sharif, M. (2025). Using Hybrid Deep Learning Models to Predict Dust Storm Pathways with Enhanced Accuracy. Climate, 13(1), 16.
Yarmohamadi, M., Alesheikh, A. A., Sharif, M., & Vahidi, H. (2023). Predicting dust-storm transport pathways using a convolutional neural network and geographic context for impact adaptation and mitigation in urban areas. Remote Sensing, 15(9), 2468.
Zhang, R., Su, J., & Feng, J. (2024). An extreme learning machine model based on adaptive multi-fusion chaotic sparrow search algorithm for regression and classification. Evolutionary Intelligence, 17(3), 1567-1586.
)