Volume 15, Issue 1 (7-2024)                   J Watershed Manage Res 2024, 15(1): 94-106 | Back to browse issues page


XML Persian Abstract Print


Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:

Hadian Amri M, Gholami V, Yousefi A. (2024). Prediction of the rainfall-runoff process using field plots and artificial neural network (ANN). J Watershed Manage Res. 15(1), 94-106. doi:10.61186/jwmr.15.1.94
URL: http://jwmr.sanru.ac.ir/article-1-1255-en.html
1- Department of Soil Conservation and Watershed Management, Mazandaran Agricultural and Natural Resources Research and Education Center, Agricultural Research, Education and Extension Organization (AREEO), Sari, Iran & Department of Soil Conservation and Watershed Management, Mazandaran Agricultural and Natural Resources Research and Education Center, Agricultural Research, Education and Extension Organization (AREEO), Sari, Iran
2- Department of Nature Engineering, Faculty of Natural Resources, University of Guilan, Rasht, Iran & Department of Nature Engineering, Faculty of Natural Resources, University of Guilan, Rasht, Iran
3- Rangeland Management, Former Student of the University of Tehran, Tehran, Iran & Former Student of the University of Tehran, Tehran, Iran
Abstract:   (714 Views)
Extended Abstract
Background: Simulation of the rainfall-runoff process for estimating runoff due to rainfall is an important step in planning and management of natural resources and water resources, especially in watersheds without hydrometric stations. However, this process has its own complexities and several affecting factors such as precipitation factors (amount and intensity of rainfall), vegetation cover (type of cover and cover density), soil factors (soil texture, soil initial moisture, and permeability), and land management quality. This study aimed to provide a model for simulating the rainfall-runoff process using artificial neural network (ANN) modeling and runoff data of field plots.
Methods: The study was executed on a slope at Guilan University using clay-loam soil in a repetitive way, with paired plots subjected to different vegetation and land management treatments. The amount of rain was measured after every rainfall using a storage rain gauge. Runoff values were estimated by plots, and from the difference between precipitation and runoff values, the initial loss values at the surface of each plot were calculated for each precipitation event under different soil moisture conditions. The total precipitation of the previous five days was estimated as the previous soil moisture. Regarding the agronomic management method, two patterns of plowing in the slope direction and plowing perpendicular to the slope direction were used and compared to patrolling native rangeland species. The changes in slope, soil texture, and soil properties were negligible due to the limits of the area, thus soil was not used as an input affecting runoff. Since the aim was to evaluate the effect of vegetation cover and agronomic management on runoff production, it was necessary to provide the same conditions for the level of paired plots to neutralize the effect of soil. To model the obtained data, they were divided into two categories: educational and subject data. The parameters of runoff values were considered the output of the model, and the precipitation values, percentage of rangeland and tree canopy, precious soil moisture, and percentage of leaf litter were regarded as the optimal inputs of the model. The land slope was estimated by surveying. Vegetation cover and leaf litter percentages were measured using the ratio of vegetated area to the total plot and microplot area. It is difficult to quantitatively determine the type of coverage, but the height of the vegetation was also considered due to the simulation of the effect of raindrops, rain spraying, and tree foliage.
Results: The amount of reserve was actually very small due to the limited area of plots. Based on statistical analysis, the rainfall and previous soil moisture factors have a positive relationship with runoff production. Vegetation and leaf litter have an inverse relationship with runoff values. Finally, the most important factor in controlling runoff production is vegetation cover (R2 = -0.71). The highest efficiency in controlling runoff production was observed in a plot with a rangeland cover of 100% canopy cover. Vegetation somehow determines the amount of leaf litter and humus in the soil. Tree species have also been limited in controlling runoff reduction, and if rangeland cover is located under the canopy of trees, it will cause double efficiency in reducing runoff production. However, there is generally no maximum pasture cover under trees or forest lands. The results showed that the amount of runoff production can be reduced by up to 10% due to the limited number of trees due to vegetation. In previous studies, this effect of forest cover varied between 40% of dense forests and sparse forests. The results also show that both factors of the type of leaf litter and the amount or percentage of leaf cover are influential in controlling runoff. Regarding the effect of plowing patterns, the results show that plowing and strip cultivation in the direction perpendicular to the slope direction lead to a decrease in the runoff rate, more runoff penetration, more moisture retention in the soil, and better conditions for vegetation growth and development. Finally, these cases will lead to a significant decrease in runoff production. A comparison of the measured runoff in these two plots in different precipitation events has indicated that the plot with plowing and perpendicular cultivation in the direction of slope after the complete establishment of vegetation can be up to 50% more effective in reducing runoff production. The results of the trial-error method in the neural network model indicate that the rainfall values, the type of vegetation cover, and the percentage of vegetation canopy are the optimal inputs for simulating runoff values. The results also show that the hyperbolic tangent transfer function and the LM learning technique are the best options for the optimal structure of networks. Neural network training in two stages showed that the used model was highly efficient in estimating runoff values. Based on the validation results, the MLP is an efficient network for simulating runoff values or the rainfall-runoff process. Moreover, a comparison between the simulated and the observed values of runoff in the experimental phase showed a good agreement between the simulated and the observed values. The values of MSE = 0.97, R2 = 0.004 and MSE = 0.91, R2 = 4.2 were obtained in the training and testing phases of the model, respectively, and, finally, a high-performance model was presented to simulate the rainfall-runoff process. The result of the modeling process showed that rangeland cover had the highest efficiency in controlling runoff.
Conclusion: Vegetation characteristics, such as vegetation type and density, are the most important factors controlling runoff in sloping lands. In addition, land management, cultivation patterns, and plowing methods are other important factors. Therefore, it is possible to estimate the total losses and initial loss based on soil characteristics, land slope, and previous soil moisture and to select suitable cultivation patterns or vegetation types for runoff control or planting and model their yields during the rainfall-runoff process. A tested model based on the neural network can also be a tool for estimating runoff values on a monthly and annual scale based on the precipitation data of meteorological stations. The model can be used to simulate the effect of different vegetation scenarios on runoff production or to estimate runoff based on the precipitation of meteorological stations.


Full-Text [PDF 1266 kb]   (232 Downloads)    
Type of Study: Research | Subject: حفاظت آب و خاک
Received: 2023/06/17 | Accepted: 2023/10/21

Add your comments about this article : Your username or Email:
CAPTCHA

Send email to the article author


Rights and permissions
Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

© 2024 CC BY-NC 4.0 | Journal of Watershed Management Research

Designed & Developed by : Yektaweb