Evaluation of the image classification results is very important in the remote sensing projects. So far, many indices have been presented to assess the accuracy of image classification, though Kappa coefficient and Overall accuracy are the most famous ones. Some researchers have criticized these two parameters, and have presented new parameters for evaluation of the classification results. In this paper, the relation between two new accuracy assessment parameters (presented by Pontius & Millones) and traditional accuracy assessment parameters (Overall accuracy and kappa coefficient) is studied. These two new parameters are called “ Quantity disagreement” and “ Allocation disagreement” which report disagreement between ground truth and classification data. In order to apply the comparative study on the traditional and new disagreement measures, supervised maximum likelihood classification was applied on 31 satellite images with different spatial resolutions. Then, Kappa and Overall accuracy as traditional accuracy parameters and Quantity disagreement and Allocation disagreement as new measures were computed for each classified image and then the correlation coefficients of the both measures were calculated. The results show a high correlation between new parameters and traditional ones in negative direction irrespective the spatial resolution. In this way, the disagreement do not provide new information about the classification results to the user, and only if there is any request for classification error, the new disagreement parameters can be used along with the traditional ones.

Changes in rainfall have significant effects on vegetation of an area, especially in arid and semi-arid regions. Nowadays, the vegetation can be assessed using indices derived from satellite imagery and remote-sensing techniques. The aim of this study was to evaluate the effect of rainfall on vegetation and to compare NDVI and RVI indices. The study area is Malekshahi, a city with an area of 1165 km2, located in the northeast of Ilam Province. The statistical data of 10 rain gauge stations in the region were used to investigate the rainfall fluctuations during the years 2000 and 2014. ETM images of Landsat satellite were used for the years 2000, 2007 and 2013. To evaluate the vegetation, NDVI and RVI were assessed using ENVI 4. 7 software. The results showed that the highest and lowest rainfalls were 600 and 211 mm in 2000 and 2014, respectively. Comparison of the two vegetation indices showed that the NDVI index with the overall accuracy of above 70% has the highest capacity to separate the semi-dense forests from the dense ones. However, the RVI index showed a greater efficiency to separate the thin forests. The NDVI index had the highest correlation with precipitation compared to RVI index. Thus, NDVI is an appropriate parameter to assess the changing process of precipitation in the study area.

Continuous monitoring of agricultural lands is imperative for managing water and soil resources in a watershed, due to its impact on ecosystem health and food security. Global Land Cover (GLC) maps can be used as a proxy for local and regional land use maps because of their availability, variety, and ease of use without complex processing. This study investigates the performance of three GLC products including MCD12Q1 LC, CGLS LC, and CCI LC against a reference land use/ land cover map of the year 2015 in the LUB. First, identical classes between the reference map and the GLC maps were determined based on the main land use/ land cover classes of the reference map of 2015 (rangeland, agricultural land, water, built-up areas, and bare land). To do so, different classes were merged accordingly to match the classes of the reference map. Subsequently, performance (Area and spatial consistency, and classification accuracy) of the GLC products was evaluated based on ground truth points. Results showed that MCD12Q1 LC and CGLS LC outperformed CCI LC in providing an overview of the surface cover of the LUB with 74% and 86% overall accuracy, respectively. Moreover, MCD12Q1 LC and CGLS LC had an acceptable performance in classifying rangeland and agriculture land as the dominant land cover types in the LUB with 81% and 92% classification accuracy, respectively. The CGLS LC can also be used to continuously monitor agriculture areas in practical applications to examine the overall trend of urbanization and agricultural development. Another important finding is that the GLC product with higher spatial resolution does not necessarily provide better classification accuracy for all types of covers. This study can also be used as a methodological reference in the performance evaluation of the GLC products at different scales and other parts of the country.

In this paper, a new histogram-based method is introduced to make object detectors resistant to hostile attacks. In the following, this method was applied to two object detector models, YOLOV5 and FRCNN, and in this way, two models resistant to attacks were introduced. In order to verify the performance of the mentioned models, we performed the adversarial training process of these models with three targeted attacks TOG-vanishing, TOG-mislabeling, and TOG-fabrication and one untargeted attack, DAG. We have checked the efficiency of the introduced models on two data sets MSCOCO and PASCAL VOC, which are among the most famous data sets in the field of object recognition. The results show that this method, in addition to improving the adversarial accuracy, also improves the clean accuracy of the object detector models to some extent. The average clean accuracy of the YOLOv5-n model for the PASCAL VOC dataset, if adversarial attacks are applied to it, in the case where no defense method is applied, is 85.5%, and in the case where the histogram method is applied, the average accuracy is equal to with 87.36%. In the YOLOv5-n model, according to the results, the best adversarial accuracy of this model, which has increased compared to other models, is in TOG-vanishing and TOG-fabrication attacks, which are 48% and 52.36%, respectively.

Mapping soil classes digitally generally starts with soil profile description with observed soil classes at a taxonomic level in a particular classification system. At each soil observation location there is a set of co-located environmental variables and a challenge is to correlate soil classes with environmental variables. The current methodology treats soil classes as ‘labels’ and their prediction only considers the minimization of the misclassification error. Soil classes at any taxonomic level have taxonomic relationships between each others. Using classification trees, we can specify an algorithm that minimises the taxonomic distance rather than misclassification error. Therefore, in this research, we have attempted to develop decision tree model for spatial prediction of soil taxonomic classes in an area covering 720 km2 located in arid region of central Iran. In this area, using the conditioned Latin hypercube sampling method, location of 187 soil profiles were selected, which then described, sampled, analyzed and allocated in taxonomic classes according to soil taxonomy of America. Auxiliary data used in this study to represent predictive soil forming factors were terrain attributes, Landsat 7 ETM+data and a geomorphologic surfaces map. Discriminant analysis was applied to calculate taxonomic distances. Results showed using the taxonomic distances led to achieve overall accuracy up to 70%. Results also showed some auxiliary variables had more influence on predictive soil class model which included: wetness index, geomorphology map and multi-resolution index of valley bottom flatness. General results showed that incorporating taxonomic distance into decision tree model had reliable accuracy. Therefore, it is suggested using of decision tree model with taxonomic distance for spatial prediction of soil classes in the future studies.

Despite research studies regarding collaborative writing (CW), the effect of using L1 as the medium of collaboration has been underexplored in computer-mediated L2 CW. This study investigated the effect of the language that the learners employed for collaboration (L1 vs. L2) on their L2 texts and examined whether learners participating in online CW using Google Docs produced better subsequent individual L2 texts. Participants consisted of 45 Iranian upper-intermediate English-as-a-foreign-language (EFL) male and female learners, with Farsi as their mother tongue and their ages ranging from 19 to 24. They were divided into three groups (two experimental groups and one control group). The essays were analyzed in terms of complexity, accuracy, fluency (CAF), and overall quality of the texts to see which language (L1 or L2) led to superior L2 texts. The researchers employed Mann-Whitney U, Kruskal-Wallis, and multivariate analysis of variance (MANOVA) to analyze the test data. The results showed that collaboratively written L2 texts were superior in terms of accuracy and overall quality compared to those generated by the control group. Furthermore, the L1 group performed better regarding complexity, whereas the L2 interaction group was superior in terms of fluency and overall quality of the texts. Based on the findings, the way collaboration is done may play a more important role than the language utilized for collaboration. The findings promise implications for the use of collaborative-based processes to contribute to EFL learners’ quality writing.

Lack of up-to-date, documented and scientific information on the current situation (area and distribution) of Zanjan poplar plantation is one of the main problems of wood production managers for planning and management of wood supply in the province and the country. In this study, Sentinel-2 satellite data with spatial resolution of 10 m in spectral bands were used and the ground truth map of existing poplar fields with 600 points was plotted in all cities and villages from field surveys. From the beginning to the end of the poplar growing season (first half of March to December 2018), at least 6 time periods of 30 to 40 days were used in the SVM classifier. Post-test and calibration of SVM model based on the phenology of poplar genus and field samples were extracted, populated area distribution map of province was extracted. The results showed that the total area of ​​poplar areas is 2744 hectares which covers 0.12% of total area of ​​Zanjan province. One percent of the total polygons were randomly selected for field control and after field control, the overall mapping error was obtained and calculated. In this study, the exact location and area of ​​poplar mills were estimated with acceptable accuracy (96%). So that using extracted information (distribution map of poplars of the province) can provide studies on comprehensive planning of poplars and sustainable management of wood production from the poplars of the province.

Water is one of most important human needs for life. According to importance of subject, discussion of management and utilization of water resources has become one of the most important global issues. Remote sensing data are often used in water body extraction studies and type of remote sensing data used plays an important role in water body extraction. In this study, ability of Landsat satellite with application of water indices, to extraction of Gamasiab River in Kermanshah and comparing these indices have been investigated. Specific feature of Low width and shallow rivers has increased the complexity of studies of such rivers using available data. Water body extraction from remote sensing images has been over the past two decades. Water indices were first developed using Landsat TM and Landsat ETM. But its better performance in Landsat 8 is well documented by the researchers. In this study, NDWI, MNDWI, AWEI_nsh, AWEI_sh and WRI indices were used. With extracting optimal threshold from histogram of indices and applying this threshold, the study area was classified into two classes of water and non-water. Then overall accuracy and kappa coefficient values were taken from each of the indices. Finally, AWEI index with overall accuracy of 99. 09% and a Kappa coefficient of 0. 98 was the best response among the indices in the study area. The results this study showed that approach can easily extract water from satellite imagery.

The signs and symptoms of the water crisis in Iran appear to lead to the decline in groundwater levels, land subsidence, soil erosion, dust storms, and the drying of wetlands, rivers and lakes. In this situation, the accurate identification and conservation of water resources is essential to reduce the effects of water scarcity. This research was conducted to better identify the water areas and assess the accuracy of aqueous clearances from non-aquatic environments using spectral measurements of distance from the Lar dam in the southwest of Damavand Peak. Four NDVI, NDWI Gao, NDWI McFeeters and MNDWI spectral indices were used to identify the water mass and distinguish it from other natural and artificial effects on Landsat 8 and Sentinel2 images. The results show the highest overall accuracy is related to the MNDWI index and the lowest overall accuracy is related to the NDWI Gao index. The kappa coefficient also represents a better separation of water zones from land in the MNDWI index. It is suggested that similar investigations be carried out in different areas of the country simultaneously to identify the aquifers so that the spectral indexes can be studied and evaluated more appropriately due to changing environmental conditions.

The values of rangeland vary, and there are significant temporal and spatial changes. Since rangeland are constantly changing, they play a crucial role in the economy and in the protecting the land and water. So it is crucial to grasp them and manage them properly. The middle "Kashkan" watershed in the "Lorestan" province was chosen in order to assess the capability of multi-spectral pictures from Sentinel 2 and Spot 5 satellites in creating rangeland density maps. Using ground control points and the region's digital height model, the photos were geometrically adjusted with an accuracy of less than 0.21 pixels. On the primary multispectral image of each satellite as well as the integrated image of Spot 5 and the rangeland density map, supervised classification utilizing the parallelepiped, minimum distance, maximum likelihood, and neural network classification techniques was carried out. Three density classes—5–25, 25–50, and 50% and above—were prepared for it. 117 ground control points were located on the topographic map of the area in question in order to measure the classification's accuracy. The global positioning system (GPS) was then used to pinpoint the locations of the points in the study area, and the ground reality map of the region was created using the determined coordinates. The Spot 5 image with PCA-3-1 band composition and a neural network classification algorithm, which had an overall accuracy of 70.53% and a Kappa coefficient of 0.65 compared to the Sentinel 2 image with PCA-8-2 band composition and a neural network classification algorithm, which had an overall accuracy of 65.72 and a Kappa coefficient of 0.08, produced better results, according to a study that examined the accuracy of classified images. This study showed that Spot 5 satellite photos outperform Sentinel 2 satellite images when creating rangeland coverage maps with three different densities. It is possible to use satellite images with spatial and spectral resolution suitable for creating a map of rangeland density and regulating and trying to prevent the destruction of rangeland in the west of the country over a certain period of time because the distances for aerial photography of rangeland areas in Iran are great.