Monitoring the changes in the appearance of the city and its surroundings based on the analysis of land appearance metrics (Case study: Ardabil city)

Due to the fact that today the settlements are growing more than the population living in it (Regami et al., 2017). Therefore, we can expect that by 2030, the area of settlements will reach 3 times the population living in it (Sun et al., 2018). Naturally, this increase in the size of settlements will lead to land use change, which will lead to the destruction of ecosystems, ecological and anthropological balance, environmental pollution, destruction of agricultural lands, infrastructure changes in the structure and ecological function of the land and…will be. Therefore, this expansion needs proper management.

The present study is of developmental-applied type and its method is descriptive-analytical. In the present study, the Landsat satellite image with the specifications listed in Table (1) and Google Earth software, ENVI4.8, ArcGIS10.2 have been used. Thus, satellite images from 2011 to 2021 were referenced by removing 23 control points from the image surface with an RMS error equal to 0.41 pixels of the earth. In geometric correction, points were selected for ground control that had a good distribution on the image surface to have less error in estimating unknown coefficients in the equation used. In the present study, the method of reducing the numerical value of dark pixels in images has been used for radiometric correction. In this way, a constant value of the total value of the pixels in a given band is reduced so that radiometric corrections can be applied to each satellite image. In the next step, due to the location of Ardabil city and its surroundings in two rows (67-134 and 1367-34), the images were mosaic. Then, using field visits and GPS, educational samples were identified for each user (settlement, seal, agriculture, desert) in the study area. Given that the control points were taken in 1400 and the images used in this study are from 2011 to 2021, there was a possibility of changes in use between this time period. Based on this, the points were visually compared with the images used and some of them that were suspected of changing the user were removed. Some of the harvested points were used for training and others for classification validation. In the present study, the classification was prepared using the support vector machine algorithm. Evaluation criteria (producer accuracy, user accuracy, overall accuracy, kappa coefficient) were used to evaluate the accuracy of image classification. In the next step, a map of land use changes in the study area was prepared and the changed land uses in the study period were identified and introduced. Were analyzed. Figure (2) shows the flow chart of the research stages.

After performing the backup vector classification algorithm on the satellite images of 2011 and 2021, land use maps were prepared (Figures 4 and 5). In the next step, the accuracy of the classifications was examined based on educational samples taken from the area. After applying the training samples on the image surface, the classification error matrix, statistical characteristics of producer accuracy, user accuracy, overall accuracy and kappa coefficient were determined for each of the classes. The results are presented in Table (3). Then the area of each land use class was calculated and is presented in Figure (6). In the next step, a map of land use changes was prepared. Figure (7). After preparing the change map of each period, the area of each user category was calculated. The support vector machine method had high classification accuracy in satellite images due to its overall accuracy and high kappa coefficient.

In this study, first, satellite images (ETM-OLI) were used and the land use map of Ardabil and its surroundings was extracted by supervised classification (support vector machine). The results also show that satellite images have a unique ability to extract land uses. The results of the application of metrics used in this research show the efficiency of class area metrics, number of spots, spot density, margin density, largest spot index, total margin and percentage of appearance in land use change analysis. Based on the research findings, it can be said that the situation of Ardabil city in the current situation, due to improper use of resources and its destruction is irregular and indicates the development of destruction in this area. According to the land use change map obtained from the comparison of land use in 2011 and 2021, it can be seen that in the 11-year period, the use of the settlement has increased by 1242 hectares, while agricultural use has decreased by 859 hectares and dams by 26 hectares. Bayer has reduced 309 hectares. These changes in the present study were quantified by land use metrics. The results show that the values of metrics for each of these classes have changed over the study period. That is, the effects of destruction and conversion of land uses have also affected the shape and size of land uses. In the next step, the accuracy of the classifications was examined based on educational samples taken from the area. After applying the training samples on the image surface, the classification error matrix, statistical characteristics of producer accuracy, user accuracy, overall accuracy and kappa coefficient were determined for each of the classes. Landscape analysis in this study shows the negative effects of human activities on land landscape changes. Given that the large number of spots indicates fragmentation and instability in the appearance of the land; As a result, low NP indicates stability. In the present study, metric analysis of the number of spots indicates that residential, agricultural and barren land use are in an unstable state. Metric analysis of land landscape percentage indicates that in both time periods, the highest land landscape percentage is agricultural land, barren, residential and dam, respectively. Therefore, the analysis of land landscape in this study shows the impact of human activities on land landscape change.