Search published articles


Showing 6 results for Machine Learning
Digital Soil Mapping of Soil Classes using Conventional Soil Maps in the Arid Region Southeastern Iran
F. Abbaszadeh Afshar, ِ S. Ayoubi, A. Jafari,
Volume 21, Issue 1 (6-2017)
Abstract

Mapping the spatial distribution of soil taxonomic classes is important for useful and effective use of soil and management decisions. Digital soil mapping (DSM) may have advantages over conventional soil mapping approaches as it may better capture observed spatial variability and reduce the need to aggregate soil types. A key component of any DSM activity is the method used to define the relationship between soil observations and environmental covariates. This study aims to compare multiple logistic regression models and covariate sets for predicting soil taxonomic classes in Bam district, Kerman province. The environmental covariates derived from digital elevation models, Landsat imagery, geomorphology map and soil unit map that were divided into two different sets: (1) variables derived from digital elevation models, remote sensing and geomorphology map, (2) variables derived from digital elevation model, remote sensing, geomorphology map and the soil map. Stratified sampling schemes were defined in 100000 hectares, and 126 soil profiles were excavated and described. The results of accuracy model showed that data set 2 increased accuracy of model including overall accuracy, kappa index, user accuracy and reliability of the producer. The results showed that the multiple logistic regression model can promote traditional soil mapping and it can be used to large group of other scientific fields.
 


Performance Evaluation of Three Image Classification Methods (Random Forest, Support Vector Machine and the Maximum Likelihood) in Land Use Mapping
F. Jahanbakhshi, M. R. Ekhtesasi,
Volume 22, Issue 4 (12-2018)
Abstract
Land use/cover maps are the basic inputs for most of the environmental simulation models; hence, the accuracy of the maps derived from the classification of the satellite images reduces the uncertainty in modeling. The aim of this study was to assess the accuracy of the maps produced by machine learning based on classification methods (Random Forest and Support Vector Machine) and to compare them with a common classification method (Maximum Likelihood). For this purpose, the image of the OLI sensor of Landsat 8 for the study area (Sattarkhan Dam’s basin in the Eastern Azerbaijan) was used after the initial corrections. Five land uses including urban, irrigated and rain-fed agriculture, range and water body were considered. For conducting the supervised classification, ground truth data were used in two sets of educational (70% of the total) and test (30%) data. Accuracy indexes were used and the McNemar test was employed to show the significant statistical difference between the performances of the methods. The results indicates that the overall accuracy of Support Vector Machine, Random Forest, and Maximum Likelihood methods was 96.6, 90.8, and 90.8 %, respectively; also the Kappa coefficient for these methods was 0.93, 0.81 and 0.83, respectively. The existence of a significant statistical difference at the 95% confidence between the performances of the Support Vector Machine algorithm and the other two algorithms was confirmed by the McNemar test.

Comparison of Artificial Neural Network, Decision Tree and Bayesian Network Models in Regional Flood Frequency Analysis using L-moments and Maximum Likelihood Methods in Karkheh and Karun Watersheds
K. Ghaderi, B. Motamedvaziri, M. Vafakhah, A.a. Dehghani,
Volume 25, Issue 4 (12-2021)
Abstract
Proper flood discharge forecasting is significant for the design of hydraulic structures, reducing the risk of failure, and minimizing downstream environmental damage. The objective of this study was to investigate the application of machine learning methods in Regional Flood Frequency Analysis (RFFA). To achieve this goal, 18 physiographic, climatic, lithological, and land use parameters were considered for the upstream basins of the hydrometric stations located in Karkheh and Karun watersheds (46 stations with a statistical length of 21 years). The best Probability Distribution Function (pdf) was then determined using the Kolmogorov-Smirnov test at each station to estimate the flood discharge with a return period of 50-year using maximum likelihood methods and L-moments. Finally, RFFA was performed using a decision tree, Bayesian network, and artificial neural network. The results showed that the log Pearson type 3 distribution in the maximum likelihood method and the generalized normal distribution in the L moment method are the best possible regional pdfs. Based on the gamma test, the parameters of the perimeter, basin length, shape factor, and mainstream length were selected as the best input structure. The results of regional flood frequency analysis showed that the Bayesian model with the L moment method (R2 = 0.7) has the best estimate compared to other methods. Decision tree and artificial neural network were in the following ranks.

Predicting and Mapping Soil Organic Carbon Using Remote Sensing and Machine Learning Algorithms
Miss S. Bandak, A.r. Movhedei Naeani, Ch.b. Komaki, M. Kakooei, J. Verrlest,
Volume 27, Issue 3 (12-2023)
Abstract
Soil organic carbon (SOC) is one of the most important components of soil physical and chemical properties that have an important role in sustainable production in agriculture and preventing soil degradation and erosion. Data mining approaches and spatial modeling besides machine learning techniques to investigate the amount of soil organic carbon using remote sensing data have been widely considered. The objective of the present study was the evaluation of SOC using the remote sensing technique compared with field methods in some areas of the Gonbad Kavous and Neli forests of Azadshar. The soil samples were collected from the soil surface (0-10 cm depth) to estimate the SOC. Data were categorized into two categories: 70% for training and 30% for validation. Three machine learning algorithms including Random forest (RF), support vector machine, extra tree decision, and XGBoost were used to prepare the organic soil carbon map. In the present study, auxiliary variables for predicting SOC included bands related to Lands 8 OLI and sentinel 2 measurement images, topography, and climate. The results showed that the extraction of the components related to the bands along with the calculation of indicators such as normalized vegetation difference, wetness index, and the MrVBF index as auxiliary variables play an important role in more correct estimation of the amount of soil organic matter. Comparison of different estimation regressions showed that the Sentinel 2 random forest model and in Landsat8 with the values of coefficient of determination (R2), root mean square error (RMSE), and mean absolute error (MEA) of 0.64, 0.05, and 0.17, respectively, was the best performance ratio compared to other approaches used in the study to estimate the organic carbon content of surface soil in the study area. In general, the results of this study indicated the ability of remote sensing techniques and learning models in the spatial estimation of soil organic carbon. So, this method can be used as an alternative to laboratory methods in determining soil organic carbon.

Estimating the Quality Index of Surface Waters Using Machine Learning Methods (Case Study: Zayandeh-Rood River)
Mohammad Shayannejad, Elham Fazel Najafabadi, Fahimeh Hatamian Jazi,
Volume 29, Issue 3 (10-2025)
Abstract
Regarding the increasing need for water resources and the decline of surface water resources, awareness of these resources is a crucial need in planning, developing, and protecting them. This research was conducted to model the water quality index (the most widely used feature of determining water quality) using machine learning models (Random Forest and Support Vector Machine) in the Zayandehrood River. Regarding the large number of water quality indices, the NSFWQI index was used in this study. First, this index was calculated, and then, input data, including water quality characteristics of 8 stations over 31 years, and the river water quality index were used. In this research, 80% of the data was used in the training stage, and the remaining 20% was used in the evaluation stage. The optimal model was selected based on the evaluation criteria, including R2, CRM, and NRMSE. The results showed that the Support Vector Machine algorithm (0.931 < R² < 0.982, 1.321

Modeling Gully Erosion Development Using Random Forest and Artificial Neural Network Models in the Mahurmilati Watershed of Fars Province
Seyed Masoud Soleimanpour, Omid Rahmati, Samad Shadfar, Maryam Enayati,
Volume 30, Issue 1 (3-2026)
Abstract
Gully erosion is one of the most important types of water erosion. Since the amount of soil loss due to this erosion is directly related to environmental factors, the amount of soil loss due to each gully can be modeled based on environmental conditions. According to the high ability of machine learning models based on artificial intelligence to analyze environmental information, in addition to determining soil loss due to gully erosion, modeling has been carried out using two random forest models, and artificial neural networks and evaluating their efficiency in the Mahurmilati watershed located in the southwest of Fars province in this study. The dimensional parameters of 70 gullies were measured over four years (2021-2024), and the volume and weight of soil lost were calculated. 15 environmental factors were selected as predictive variables, and modeling was performed with a cross-validation approach using these two models, and the accuracy of the models was evaluated using quantitative criteria. The amount of soil loss in gullies during the study period was 15300.94 tons. The accuracy evaluation of the models showed that the random forest model had better performance based on the coefficient of determination (R2=0.66-0.73). Also, this model had the lowest value in terms of the RSR error index evaluation criterion (RSR=0.66-1.03) and the highest accuracy. In terms of the fit evaluation index (D), the random forest model also had the highest fit between the observational and forecast data and had the highest value of this index (D=0.83), and therefore, it was introduced as the superior model for predicting soil loss due to gully erosion in this watershed.

Page 1 from 1     

© 2026 CC BY-NC 4.0 | Journal of Water and Soil Science

Designed & Developed by: Yektaweb

تحت نظارت وف بومی