support vector machine (svm)

Dust storms pose significant environmental and economic challenges, particularly in arid regions like Sistan-Baluchestan Province, Iran. This study aims to compare the performance of individual models (GRNN and SVM) with a triple hybrid model (GRNN-SVM-LSTM) for forecasting the frequency of dust storm days (FDSD). Using hourly dust data from eight SYNOP codes of the World Meteorological Organization across five synoptic stations, spanning a 40-year period (1980–2020), the models were evaluated based on key performance metrics: Correlation Coefficient (R), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Nash-Sutcliffe Efficiency (NS). The triple hybrid model outperformed all other approaches, achieving the highest predictive accuracy in seasonal combinations 1 and 2. The SVM model ranked second, while the GRNN model performed relatively better in combinations 1 and 2 compared to combination 4. Overall, the GRNN-SVM-LSTM model demonstrated superior predictive performance for FDSD, with RMSE = 0.523–0.501, R = 0.999–0.989, MAE = 0.441–0.421, and NS = 0.907–0.893. These findings highlight the potential of the proposed model for improving dust storm forecasting and developing early warning systems.

Rainfall is one of the complex natural phenomena and one of the most crucial component of the water cycle, playing a significant role in assessing the climatic characteristics of each region. Understanding the amount and trends of rainfall changes is essential for effective management and more precise planning in agricultural, economic, and social sectors, as well as for studies related to runoff, droughts, groundwater status, and floods. Additionally, rainfall prediction in urban areas has a significant impact on traffic control, sewage flow, and construction activities.

The objective of this study is to compare the accuracy of classification models, including Chi-squared Automatic Interaction Detector (CHAID), C5 decision tree, Naive Bayes (NB), Quest tree, and Random Forest, k-Nearest Neighbors (KNN), Support Vector Machine (SVM), and Artificial Neural Network (ANN) in predicting rainfall occurrence using 50 years of data from the synoptic station at Hamedan Airport. In this study, 80% of the data is used for training the models, and 20% for model validation and the results obtained from the model executions are compared using metrics such as confusion matrix, Receiver Operating Characteristic (ROC) curve, and the Area Under the Curve (AUC) index. To create the classification variable for rainfall and non-rainfall data, based on rainfall data, the days of the year are categorized into two classes: days with rainfall (y) and days without rainfall (n). Data preprocessing is performed using Automatic Data Preprocessing (ADP). Then, Principal Component Analysis (PCA) is employed to reduce the dimensions of the variables.

In this study, the PCA method reduces the dimensions of the variables to 5. Also, approximately 80% of the available data corresponds to rainless days, while 20% corresponds to rainy days. The research results indicated that the KNN model with an accuracy of 91.9% for training data and the SVM model with 89.13% for test data exhibit the best performance among the data mining models. The AUC index for the KNN model is 0.967 for training data and 0.935 for test data, while for the SVM algorithm, it is 0.967 for training data and 0.935 for test data. According to the ROC curve for Hamedan rainfall data, the KNN model outperforms other models. Considering the sensitivity index in the confusion matrix, the KNN and SVM models perform better in predicting non-rainfall occurrence for training data. In terms of the precipitation occurrence prediction, the RT and KNN models show better results according to the specificity index.

The results demonstrated that for the RT, C5, ANN, SVM, BN, KNN, CHAID, QUEST, accuracy metrics was obtained 86.82%, 89.78%, 89.55%, 89.96%, 88.06%, 91.9%, 88.29%, 87.46%, 91.9%, respectively for training data. Moreover, for test data, the accuracy metrics for this model was obtained 83.82%, 87.9%, 88.12%, 89.13%, 87.12%, 89.13%, 87.12%, 88.19%, 86.93%, 86.76%, respectively. The AUC index in the training data for RT, C5, ANN, SVM, BN, KNN, CHAID QUEST models was 0.94%, 0.99%, 0.94%, 0.94%, 0.93%, 0.97%, 0.93%, 0.89%, respectively. In addition, for the test data, this metric was evaluated 0.89%, 0.89%, 0.93%, 0.94%, 0.92%, 0.90%, 0.92%, 0.88% respectively. As observed, considering accuracy metric and AUC index for training data KNN model and for test data SVM model were more sufficient in rainfall prediction.

The current study aims to find the best methods of using remote sensing and supervised classification algorithms in long-term salinity monitoring of salinity changes in the Atabieh area with an area of 5000 hectares in the west of Khuzestan province. The procedure is based on the separation of different levels of saline soils utilizing information obtained from Landsat 7 and 8 satellite images (2001 to 2015) along with salinity data taken from the study area, and salinity indices including SI1, SI2, SI3, NDSI, IPVI, and VSSI. The results show the expansion of the saline zone trend in the soils of the study area, among which, soils with EC of more than 16 dS m-1 (very saline) have the highest frequency. The area of saline soils has increased significantly over the past 15 years, with a saline land area increasing by more than 90%. The percentage of salinity class is low (S1). According to this study, the only significant index in soil salinity at a 95% confidence level is the SI3 index, which has been able to have a good estimate of the increasing changes in soils in the region. The results of the supervised classification showed that the support vector machine (with an overall accuracy of 95.78 and a kappa coefficient of 0.89) is more accurate. After the vector machine method, the methods of minimum distance, maximum likelihood, and distance of Mahalanobis have the highest accuracy, respectively. Based on salinity maps obtained in years in 2001, 2005, 2010, and 2015, it can be said that the salinity rate in the whole of the study area was progressing and at the same time the salinity area in the middle and high classes increased decreased and on the other hand, the salinity area in the high class in 2001 gradually increased and distributed in 2015 throughout the region.

Estimating the amount of wood produced in poplar plantations is not possible without determining the area of poplar plants. Estimating the amount of wood produced is very important for the production management of wood industry units. Industry owners need to know how much poplar wood can be extracted in the coming years. Therefore, identifying areas of poplar cultivation in Mazandaran province is one of the main objectives of the present study.

This study was conducted to estimate adult poplar cultivations area using Sentinel2 satellite in Mazandaran Province. Firstly, the general survey carried out on poplar cultivation with full canopy distribution and its cultivation area at the regarding area. Then, in the whole area, the numbers of 202 cultivated areas have randomly been mapped using Global Positioning System (GPS) as Training Samples. In the next step, the relevant images have been identified in term of time and downloaded, and also radiometric, atmospheric, and geometric corrections have been done on them. Also, banding appearance, color compositions, images mosaicking and bands calculations have been performed. The Support Vector Machine (SVM) algorithm has been used in order to extract poplar cultivation in the growth period of 2018 (March to December). Evaluation of the accuracy of the classification results of the generated maps was performed using 183 ground trust points via GPS.

The results showed that in 2018, 705 hectares of Mazandaran area are under poplar plantations, which Mahmoudabad and Neka cities with an area of 167 hectares and 33 hectares, respectively, had the highest and lowest area of poplar cultivation. Poplar cultivation area estimated in this study was excluding young poplar stands. Furthermore, the results of the accuracy assessment demonstrated that the overall accuracy is 95.35. The average area of poplar stands was 800 square meters. But the number of stands less than 500 square meters (7121 plots) includes more than 80% of the stands. Only 0.2% of the total stands (57 stands) make up more than one hectare. These 57 stands occupied about half of the province's poplar plantations.

Despite the favorable climatic conditions in Mazandaran province, the area of poplar cultivations was lower than expected. Therefore, it is suggested that in the future, the necessary measures should be taken to meet the wooden needs of Mazandaran province through wood farming.

Drought prediction is an important item in realm of hydrometeorology and hydrology, and selection of suitable meteorological variables for drought prediction is a goal in recent studies. In this paper, suitable feature selection is investigated with application of Mutual Information (MI) on the predictor’s time series and the well-known statistical machine learning methods, Support Vector Machine (SVM), is proposed to predict drought class based on Standardized Precipitation Index (SPI) in some seasonal scale scenario in the main watersheds of Tehran. In current study, ground weather temperature (at 300, 500, 700 and 850 mi bar) and geopotential height (at 300, 500, 700 and 850 mi bar) was applied in prediction models based on data from 1975 to 2005 in the main watershed of Tehran. Regarding to the amount of predictors, suitable feature selection is investigated with application of Mutual Information (MI) on the predictor’s time series and target time series and the well-known statistical machine learning methods, support vector machine (SVM), is applied to predict SPI class. One of the important issue in this research is use of different variables, for example regarding to selected data points, the effective regions on Tehran precipitation are southern, southwestern and northwestern of Iran in spring, northern and northwestern in autumn and northwestern and western in winter. SVM depicted accurate results in classification and prediction of SPI and it is suitable and applicable. The predicted SPI in winter and autumn are more accurate than the other scenarios.

Managing the production of greenhouse products requires knowledge of controlling many environmental factors and plant nutrition and fight against pests and plant diseases. Recognition pests and fight against them is one of the most important activities in the process of production of greenhouse products. Pre-knowledge of the demographic density of insect pests is one of the effective methods of pesticides and reduce their levels of use, especially for insect pest control toxins. Wireless Sensor Networking Technologies (WSN) is one of the new technologies used to sense the environment and collect and transmit information to the user or the central station to view and respond appropriately to an occurrence or phenomenon. In this study, the use of WSN in monitoring, timely diagnosis of greenhouse white flies, design and mapping of greenhouse contamination was investigated. For this goal, 3750 images of 15 sticky traps with white flies that attached to Melon greenhouse in Isfahan Agricultural Jihad Research Center were provide and transmitted online using a WSN to a computer located at a distance of 900 meters from the greenhouse. The color images of the sticky traps are acquired by using 15 digital cameras were converted to gray colored images using MATLAB software, then after image classification with Support Vector Machine (SVM) classifier based on their features, are divided into two categories of images: whiteflies affected image and whiteflies unaffected image. After identification of the white flies, number of pests was counted and infection maps of Greenhouse with ArcMap10.2 software was drawn up. Assessment of the system showed that accuracy of SVM algorithm for categorizing images of sticky traps was 97.73%, and the average values of statistic parameters of the Confusion matrix for 15 traps including sensitivity, accuracy, specificity and classification accuracy were 98.46%, 83.31%, 99.08% and 97.72% respectively. The overall accuracy of the system for detection and counting Greenhouse whitefly pests is 97.71%. The average root mean square error (RMSE) in estimating of the number of white flight by image processing and direct counting was between 1 and 5.03. Therefore, the system is suitable for detecting and tracing and counting the number of trapped white flies, and it is possible to design appropriate greenhouse poisoning plans to fight this pest.

Study spectral reflectance through spectral indices allows the optimal use of the wide range of spectral wavelengths in hyperspectral data. The purpose of this study was to introduce and evaluate the performance of spectral indices to discriminate dominant orchard species in Chaharmahal Bakhtiari province. In this study, 150 spectral curves were measured in the range of 350 to 2500 mm, from grapes, walnuts and almond trees. After the initial correction, 30 of the most important spectral indices were extracted. Analysis of variance and comparisons of meanings was applied to identify the optimal indices for species discrimination at a 99% confidence level. Then, an artificial neural network (ANN) and support vector machine (SVM) approaches were used to evaluate the performance of indices in species discrimination. ANOVA results indicated that the Moisture Stress Index (MSI), Band ratio at 1,200 nm, normalized phaepophytiniz index (NPQI) and cellulose absorption index (CAI) indices are optimal for discrimination of the studied species. The performance evaluation of the introduced indicators in some of the ANN and SVM enhancement structures has been associated with 100% accuracy in both education and testing, which shows the effectiveness of these studies in distinguishing orchard species. The performance evaluation of the introduced indicators has been validated at 100% in both training and testing stages. This result emphasizes the necessity of performing spectroscopic studies to separate the orchard species before analyzing the hyperspectral images due to their large data volume, high cost and huge data analysis.

In this paper, a smart method is designed in order to classify healthy and illness ducks using their emission voice. For this purpose, firstly, the birds based on their healthy condition are divided into the different categories and then their voices are saved using a microphone and data acquisition card. Gained signals were transformed from time-domain signal to frequency domain using Fast Fourier Transform (FFT). Then, 5 statistical features are extracted from both time and frequency signals namely, mean, standard division, root mean square, variance and kurtosis. Two classifiers which are artificial neuralnetworks (ANN) and support vector machine (SVM) are used, in order to acquire the bird classification in healthy and sick accuracy. The accuracy of ANN classifier in detection of healthy birds within sick and weak birds was determined 75% and 82.1 % based on the time and frequency domain of the sound signals, respectively. The accuracy of SVM classifier in detection of healthy birds within sick and weak birds was determined 85.7 % and 92.8 % based on the time and frequency domain of the sound signals, respectively.

Accurate estimation of river flow can have a significant importance in water resources management. In this study, Genetic programming (GP) and Support Vector Machine (SVM) methods were used to forecast daily discharge of Barandoozchay River. The daily discharge data of Barandoozchay River measured at the Dizaj hydrometric station during 2007 to 2011 was used for modeling, which 80% of the data used for training and remaining 20% used for testing of models. The results showed that in the both of considered methods, the models including discharges of one, two and three days ago had higher accuracy in verification step and the accuracy of models decreased with increasing discharge values. Comparing the performance of GP and SVM methods indicated that, however the accuracy of the GP method with the R=0.978 and RMSE=1.66 (m3/s) was slightly more than SVM method with R=0.976 and RMSE=1.80 (m3/s), but the SVM is easier than GP method. Thus, the SVM method can be used as an alternative method in forecasting daily river discharge.