Farzaneh Parsaie; Ahmad Farrokhian Firouzi; Masoud Davari; Ruhollah Taghizadeh-Mehrjardi
Abstract
Surface soil saturated hydraulic conductivity (Ks), as one of the most important physical properties of soil, plays a key role in the distribution of water and nutrients within the soil environment and holds particular significance in water and soil resource management. This study aimed to digitally ...
Read More
Surface soil saturated hydraulic conductivity (Ks), as one of the most important physical properties of soil, plays a key role in the distribution of water and nutrients within the soil environment and holds particular significance in water and soil resource management. This study aimed to digitally model Ks using machine learning approaches in the Kilanah watershed, located in Kurdistan Province, covering an area of 12,000 hectares. Three machine learning algorithms, including Gradient Boosted Decision Tree (XGBoost), Random Forest (RF), and k-Nearest Neighbors (k-NN), were utilized, incorporating various environmental variables derived from the digital elevation model and Sentinel-2 satellite imagery. These variables included distance from the drainage channel, valley depth, relative slope position, channel base level, brightness index, wind effect index, Normalized Difference Vegetation Index (NDVI), Band 12, greenness index, and surface curvature. Additionally, soil parameters such as organic matter, lime content, bulk density, geometric mean particle diameter, soil texture, and near-soil spectroscopic data (Latent Variable) within the wavelength range of 400–2450 nm were used as proxies for pedogenic factors to model saturated hydraulic conductivity. The results indicated that the XGBoost model exhibited the highest accuracy for predicting Ks, with an R² value of 0.65 and an nRMSE of 0.25, outperforming the other models. Spectral data, topographic variables, and soil parameters, as model inputs, played a significant role in predicting the spatial variability of Ks. The XGBoost model was able to provide highly accurate predictions. The results demonstrated that topographic, physical, and spectral variables influence Ks; organic matter, soil texture, and topographic indices such as slope and relative position had the most substantial impact. The generated maps can be utilized for water and soil resource management and hydrological models.
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.
masoud salari; Fereydoon Sarmadian; Ali Salajegheh
Abstract
Wild sheep (Ovis orientalis) are a critical component of wildlife biodiversity in Iran and are categorized as Vulnerable (VU) on the IUCN Red List. This species plays a crucial role in maintaining the integrity of rangeland ecosystems and contributes to ecological balance within their habitats. ...
Read More
Wild sheep (Ovis orientalis) are a critical component of wildlife biodiversity in Iran and are categorized as Vulnerable (VU) on the IUCN Red List. This species plays a crucial role in maintaining the integrity of rangeland ecosystems and contributes to ecological balance within their habitats. Variations in land characteristics (including climate, topography, soil, vegetation, hydrological factors, and land use) result in distinct habitat suitability classifications for this species. This study involved long-term observational research on wild sheep behavior over a decade, aiming to identify the most influential factors affecting habitat suitability and to generate a habitat suitability map using machine learning algorithms alongside the Analytical Hierarchy Process (AHP) in Khabr National Park. The findings indicate that the region has relatively high suitability for this species, with elevation, slope, vegetation cover, and proximity to water resources emerging as the most significant factors. Validation of the results using the kappa coefficient and the overall accuracy index confirms the high precision of the findings. This underscores the value of integrating machine learning models with AHP in habitat suitability assessments, aiding management in understanding the species’ ecological requirements and identifying priority conservation areas.
Farzaneh Parsaie; Ahmad Farrokhian Firouzi; Masoud Davari; Ruhollah Taghizadeh-Mehrjardi
Abstract
Mechanical properties of soil, such as shear strength and penetration resistance, play a crucial role in optimizing crop productivity and proper soil management. The objective of the research was to produce digital map of soil shear strength and penetration resistance in Kielaneh watershed, located in ...
Read More
Mechanical properties of soil, such as shear strength and penetration resistance, play a crucial role in optimizing crop productivity and proper soil management. The objective of the research was to produce digital map of soil shear strength and penetration resistance in Kielaneh watershed, located in Kurdistan Province, covering an area of 12,000 hectares using Gradient Boosted Decision Trees (XGBoost), Random Forest (RF), and k-Nearest Neighbors (KNN). Soil penetration resistance and shear strength were measured using handheld penetrometers and vane shear devices at 150 observation points from the surface soil layer (0 to 10 centimeters). Spectral data and auxiliary variables derived from the Digital Elevation Model and Sentinel-2 satellite images were used to predict soil shear strength and penetration resistance. These variables include CHND, VD, RSP, CHNBL, Brightness, WE, NDVI, Band12, Greenness, PLC, as well as soil parameters such as organic matter, calcium carbonate, bulk density, geometric mean particle size, soil texture (percentages of clay, sand, silt), and visible near-infrared spectral data as latent variable (LT), representing soil formation factors. The results showed that the XGBoost had higher accuracy compared to other models for predicting shear strength in surface soil layer with an (R2) of 0.61 and an nRMSE of 0.16, as well as for predicting penetration resistance in the surface soil layer with an (R2) of 0.60 and an nRMSE of 0.11. In conclusion, the XGBoost model, using spectral data along with topographic variables and soil parameters, was able to estimate the spatial variability of soil mechanical properties with acceptable accuracy in the study area. The generated maps can be used to make necessary management decisions regarding of the region.
Fatemeh Ebrahimi Meymand; Hasan Ramezanpour; Nafiseh Yaghmaeian; Kamran Eftekhari
Abstract
In recent years, the use of digital soil mapping (DSM) based on machine learning algorithms with the aim of preparing soil maps has become widespread with the basis of soil class prediction with the help of modeling the relationships between them and environmental variables. One of this method's challenges ...
Read More
In recent years, the use of digital soil mapping (DSM) based on machine learning algorithms with the aim of preparing soil maps has become widespread with the basis of soil class prediction with the help of modeling the relationships between them and environmental variables. One of this method's challenges is the imbalanced nature of soil distribution in landscape, which leads to overfitting and underfitting of classes, and as a result, reduces the accuracy of many used models. This study was conducted to evaluate the ability of two machine learning algorithms, including random forests and support vector machines, for the digital mapping of soil classes with an imbalanced data set. This study was conducted on 95 soil profile classes at the family level, in 4000 hectares of land in the Honam sub-basin, Lorestan province. The issue of imbalance in soil classes was investigated by using six data sets, including the original soil data set and five data sets created by several resampling approaches including two manual classifications and three over-sampling, under-sampling, and Synthetic Minority Over-Sampling Techniques in the R software. The results showed that despite the low values of overall accuracy, the Geographical distribution of soils with high frequency in the study area in digital soil map obtained from the random forest and the original data set as well as Synthetic Minority Over-Sampling Technique, with conventional soil map of study area is significant. Therefore, the low observation number of other soil classes and as a result incorrect training of models can be considered as one of the main reasons for the low accuracy of the used models.
Serveh Darvand; Hassan Khosravi; Hamidreza Keshtkar; Gholamreza Zehtabian; Omid Rahmati
Abstract
The purpose of this study was to compare machine learning models including Support Vector Machine, Classification and Regression Tree, Random Forest, and Multivariate Discriminate Analysis to prioritize susceptible areas to dust production. To determine the dust days, hourly meteorological data of Alborz ...
Read More
The purpose of this study was to compare machine learning models including Support Vector Machine, Classification and Regression Tree, Random Forest, and Multivariate Discriminate Analysis to prioritize susceptible areas to dust production. To determine the dust days, hourly meteorological data of Alborz and Qazvin provinces and satellite images of the same days for the period 2000 to 2019 were used. 420 dust collection points were identified and the map of their distribution was prepared. The maps of factors affecting the occurrence of dust, including landuse map, soil orders map, slope map, slope aspect map, elevation map, vegetation map, topographic surface moisture, topographic surface ratio, and geology mam were prepared. Using the mentioned models, the impact of each of the effective factors of dust was determined and prioritization maps of dust harvesting areas were prepared. Models were evaluated using the ROC curve. According to the results, the elevation factor is more important in all models than the other parameters used in the model. The modeling results also showed that the Random Forest )RF( and Multivariate Discriminate Analysis (MDA) models had the highest values of accuracy (0.96), precision (0.94), Probability of Detection (POD) (0.98), and False Alarm Ratio (FAR) (0.051) compared to the others. The performance of the RF and MDA models is better than the other models, followed by the Support Vector Machine (SVM) and Classification and Regression Tree (CART) models, respectively. Also, in evaluating the models using Receiver Operating Characteristic (ROC), the RF model was selected as the best model.
Zahra Barati; Ebrahim Omidvar; Ataollah Shirzadi
Abstract
Landslide susceptibility mapping is considered as the first important step in landslide risk assessment. The main purpose of this study is to compare the performance of a machine learning algorithm (a logistic model tree), and a statistical model (a logistic regression), for landslide susceptibility ...
Read More
Landslide susceptibility mapping is considered as the first important step in landslide risk assessment. The main purpose of this study is to compare the performance of a machine learning algorithm (a logistic model tree), and a statistical model (a logistic regression), for landslide susceptibility modeling in the Sarkhoon watershed, Chaharmahal and Bakhtiari province. For this purpose, at first, a landslide inventory map including a total of 98 landslide locations was constructed using historical landslides, and extensive field surveys. In addition, a total of 100 non-landslide locations were also identified to construct a database. The landslide and non-landslide locations were randomly selected and divided into two groups with a 70/30 ratio for modelling and validation processes. Twenty conditioning factors were selected based on literature review and geo-environmental properties in the study area. Subsequently, the logistic model tree (LMT) and the logistic regression (LR) models were applied to identify the influence of conditioning factors on landslide occurrence. Finally, the performance of the models in landslide susceptibility mapping was investigated using the area under the receiver operating characteristics curve (AUC). The results concluded that the LR model (AUC = 0.797) outperformed and outclassed the LMT (AUC = 0.740) model in the study area. Although both models were reliable tools for spatial prediction of landslide susceptibility; however, the LR model was more accurate that it can be proposed as an alternative tool for better management of areas prone to landslide in the study area.