Mohammad Karimi Firozjaei; Amir Sedighi; Najmeh Neisany Samany
Abstract
Introduction Remote sensing data provide valuable information for the agricultural section and natural resources managers. Nowadays, performance management and estimation via using various methods such as classification and mapping have gained great significance. An example of such data is the mapping ...
Read More
Introduction Remote sensing data provide valuable information for the agricultural section and natural resources managers. Nowadays, performance management and estimation via using various methods such as classification and mapping have gained great significance. An example of such data is the mapping of crops cultivation and orchards at national and regional levels, which is one of the key tools in sustainable agricultural planning and management. These studies appear necessary especially in the field of strategic commodities such as rice and citrus which are among the most important food items for the Iranian people. The spatial information on agricultural lands in the field of agricultural planning and management can help the prevention of the spread of pests, management of the environmental stresses, crop performance estimation and vulnerability assessment in crop production. Field surveys and observations for crops mapping in the growing season in different years are very time-consuming, costly, and only suitable for small-scale studies. In contrast, over the past decades, remote sensing has been recognized as a suitable method for crops mapping for large areas in the shortest time and at low cost. Due to the climatic conditions of the areas in North of Iran, green spaces including vegetation and orchards, and rice fields are located near each other. At the time of the maximum growth of rice products, the spectral characteristics of these land covers are very similar. Therefore, the separation of these two land covers using satellite image classification process faces serious challenges. The aim of this study is to investigate the efficiency of the satellite images and the optimization algorithms for separating green spaces and rice fields from each other at the time of maximum growth. The present study differs from others in this field from two aspects; first, the study compares the capabilities of multispectral and hyperspectral satellite images with each other; additionally, it aims at comparing and evaluating the efficiency of the Particle Swarm Optimization (PSO) and Gravitational Search Algorithm (GSA) so as to determine the optimal features for increasing the separation accuracy of green spaces and rice fields. Materials and methodology This research was carried out based on the two objectives of studying the capabilities of the Hyperion and Landsat images and comparing the efficiency of the PSO and GSA to determine optimal features for the separation of green spaces and rice fields. For this purpose, the two Landsat and Hyperion satellite images as well as ground data sets of the case study in North of Iran were employed. In the first step, preprocessing of the Hyperion and Landsat images was performed. In the second step, various features were extracted from the Hyperion and Landsat images using different spectral indices and transformations. In the third step, the Support Vector Machine (SVM) classifier was applied with two strategies, i.e. the usage of spectral bands and the usage of spectral bands as well as indices as the features in the classification process to extract green spaces and rice fields. In the fourth step, PSO and GSA were employed to extract optimal features from the Hyperion image to distinguish between green spaces and rice fields; then, classification was done with the extracted optimal features; and finally, the efficiency of PSO and GSA were compared to determine the optimal features for the separation of green spaces and rice fields using ground data sets. Results and discussion The results indicate that the use of Landsat image is not effective for the separation of rice fields and green spaces. In other words, due to the high spectral similarity of these land covers, a large percentage of pixels related to the two classes are mistakenly classified in another class. However, the accuracy of the producer and user relating to each class has increased by an average of 10 percent with the addition of spectral indices to the classification process. Using Hyperion image is more effective than Landsat image for the separation of rice fields and green spaces. Moreover, the accuracy for the separation of rice fields and green spaces has increased with the simultaneous consideration of the bands and spectral indices in the classification process. It should be noted that one of the key factors in the efficiency evaluation process of the classification methods is the processing time. The results of using optimization algorithms for determining the optimal features indicate that out of the 150 spectral features (including 140 Hyperion image bands and 10 spectral indices and transformations), using PSO and GSA, only 25 and 31 optimal features were selected for the separation of green spaces and rice fields, respectively.The use of optimal features in the classification increases the accuracy for the separation of green spaces and rice fields more, compared to the use of all features in the classification. Additionally, GSA is superior to PSO when used for extracting optimal features for the separation of green spaces and rice fields. Conclusion The results of this research indicate that the separation accuracy of green spaces and rice fields using Landsat image,is less than that of Hyperion image. With the addition of spectral indices to the classification process, the separation accuracy in both Landsat and Hyperion data increases. Moreover, using an optimization algorithm to determine the optimal features in the classification process will increase the separation accuracy of green spaces and rice fields. Given the overall accuracy values, the efficiency of GSA for separating green spaces and rice fields is higher than PSO.
Mir Reza Ghaffari Razin; Behzad Vosooghi
Abstract
Extended Abstract Introduction Development of reliable models for estimation and prediction of changes inTotal Electron Content (TEC) of the ionosphere is still considered to be a real challenge for geodesists and geophysicists. This ispartly due to the nonlinear behavior of the physical and geophysical ...
Read More
Extended Abstract Introduction Development of reliable models for estimation and prediction of changes inTotal Electron Content (TEC) of the ionosphere is still considered to be a real challenge for geodesists and geophysicists. This ispartly due to the nonlinear behavior of the physical and geophysical parameters affecting the TEC variations, as well as the difficulty in accurate measurement of some of these parameters. Due to its specific nature, as well as its physical and geophysical properties, quantity of TEC hasspatio-temporal variations, which can be attributable to daily, and seasonal variations, various anomalies, or periods of solar activity. Total Electron Content is the quantity which can be used to study ionospheric activities, as well as the spatio-temporal variations in electron density of this layer. In fact, TEC is the total number of free electrons in the path between the satellite and the receiver in a one square meter column. The measurement unit of TEC is TECU, which is equivalent to 1016electrons/m2. Due to inappropriate spatial distribution of GPS receivers and their limited number, as well as observationaldiscontinuity in the time domain, TEC values and electron density obtained from theGPS measurements will be spatiallyand temporallyconstrained. In order to calculate TEC value in areas lacking observation or appropriatestation distribution, TEC value obtained from GPS measurements must be interpolated or extrapolated in a suitable manner. Materials and Methods By combining wavelet localization features with standard neural networks, Wavelet Neural Networks (WNN) have emerged as a new mathematical method for modeling and predicting the behavior of different phenomena.In WNNs, the output parameter is usually calculated by the following equation: (1) wherex is the inputobservations vector, is a the multi-variablewavelet whichcan be calculated by the tensor productof m (basic function of single variable wavelets), ë is the number of neurons in the hiddenlayer, and ù shows the network weight. Unlike the Backpropagation (BP) algorithm, PSO is a global search algorithm that can optimize the initial weights and introduce the appropriate structure for the network. Equations used in this algorithm are as follows: (2) (3) In which, shows the initial weight, represents the particle’s velocity i in repetition t, c1 and c2, indicate the particle acceleration coefficients, is the current position of particle i in repetition t and gbest represents the best particle position. The present study took advantage of a smoothing algorithm to determine STEC observations. Observed STEC values are as follows: (4) To obtain TEC value along the zenith, the following mapping function can be used: (5) Which we will have: (6) Elev. in relation (6) is the satellite’s elevation angle. Results and Discussion Observations of 37 Iranian GeodynamicNetworkson 2012.08.11 (DAY 224) were used to evaluate the efficiency of WNN and PSO training algorithm in modeling and predictingspatio-temporal variations of TEC in Iran. Of the 37 stations, 5 were used as test stations, 2 were used to evaluate the wavelet neural network, and the rest were used to train the network. Four different combinations of input observations are examined in this paper. Number of input observations selected from the Iranian Permanent Geodynamic Network(IPGN) to train the WNN using PSO algorithm was25, 20, 15 and 10, respectively.Table 1 shows the characteristics of different combinations evaluated in this paper. Table 1. Characteristics of the observations used in the different combinationsevaluated To evaluate the accuracy of the results obtained from IRI and WNN model, all results were compared with TEC observations obtained from GPS. Table 2 shows the correlation coefficient for different scenarios. Table 2. correlation coefficient for different scenarios According to Table (2), the first scenario in WNN method with GPS hasthe highest correlation coefficient. Even when the number of observations in the databasedecreases in the third scenario, theWNN method still has a higher correlation coefficient compared to the IRI2012 model. In the fourth scenario, the correlation coefficient for WNN method is reduced to some degree. The average relative and absolute error values at the 5 test stations were calculated for the four different scenarios and presented in Table3. Table 3. Comparison of mean relative error and absolute error values at 5 test stations for four different scenarios. Statistical analysis of relative and absolute error showssuperiority of WNN method in TEC modeling as compared to the IRI2012. Conclusion To model total electron content of the ionosphere, 4 combinations of observations were evaluated. 25, 20, 15 and 10 stations were used to train the wavelet neural network. 300, 240, 180, and 120 observations(latitude and longitude, observation time)were considered in the database, respectively.Results of the analysis indicated that with a decrease in the number of observations in the database, the absolute and relative error increase, while correlation coefficient decreases. This decrease was not evident before 180 observations, but relative and absolute errorreached up to twice their values with 120 observations. It should be noted that even with 120 observations (10 stations for training), results of the wavelet neural network model are more accurate than the results of the IRI2012 model.