ENGINEERING JOURNAL OF GEOSPATIAL INFORMATION TECHNOLOGY
Year:2024 | Volume:12 | Issue:1|Papers Count:6

In this study, the ability of local basic functions in static modeling of the Moon's gravity field is investigated by using the local harmonic functions and the artificial observations of the Primary and Extended Missions of GRAIL satellite and GL1500E global model. In this modeling, three months of GRAIL observations are used to form the observational gravity data. The observation used in this research is the gravity difference along the line of sight of the satellite pair (LGD). Due to the supposition that there are ice masses distributed on the moon, analysis and modeling of variations in the gravitational field are crucial. Modeling the moon's gravity field is complicated by the presence of the earth's gravity field and its effect on the dynamic orbit of the GRAIL satellite. In this study, the coefficients of the local gravity model have been obtained by forming the normal equation through adjusted spherical cap harmonic basic functions (ASCH) and solving the inverse problem. In this study, in the first scenario, the modeling of the gravity field using primary mission data (an average altitude of 50 km) is discussed. In the second scenario, due to the different altitude layers of the Primary Mission (average altitude of 50 km) and the Extended Mission (average altitude of 20 km), observations from the Extended Mission are used to provide ASCH coefficients. The observations from the primary mission are used to evaluate the accuracy of the model. The results of the constructed model on the primary mission data at control points was 0. 08 micro Gal and the results of the control points on the extended mission observations was 0. 15 micro Gal. As a result, the lunar South Pole local geo-potential model's validation is confirmed. By employing fewer coefficients the method we used showed an acceptable spatial precision of the gravity field of the changes compared to the geo-potential models such as GL1500E.

Agriculture, as one of the main factors in ensuring food security of the society, is of special importance in decision making, especially in making policies related to the import and export of certain agricultural products. Hence, determining crop acreage is essential for each agricultural year. The division of croplands based on the agricultural system during the cropping year can provide us with more accurate area estimation for autumn and spring cultivation. Because the area of lands with double crops (autumn and spring cultivation) is also talking into consideration in in two times. This study uses the time series of sentinel2 vegetation index (NDVI) and a knowledge-based decision tree method for classifying agricultural lands into four classes (autumn, spring, alfalfa cultivation, and double-crop fields). All parts of the method have been implemented in the Google earth engine (GEE) programming interface. The performance of the proposed method is evaluated in a study area in Shahrekord city using ground truth data gathered by extensive field surveys and eventually, the proposed method with an overall accuracy of 97. 27 % has outperformed the Nearest Neighbor (overall accuracy = 93. 76%) and the Decision Tree (overall accuracy = 94. 32%) classifiers. The final result also shows a high similarity of the map produced by the proposed method and the Support Vector Machine (SVM) classifier. Although, the SVM with an overall accuracy of 97. 84% is slightly more accurate than the proposed method, the simplicity, understandablity and the direct use of the crop phenology features in the various crop years without the need for retraining, are the unique advantages of the proposed method.

Over the last few years, deep learning models have received a lot of attention for the spectral-spatial classification of the hyperspectral images. One of the significant advantages of deep learning methods is that they incorporate both spatial and spectral information in classifying hyperspectral images. Although these models produce accurate classified maps, they are computationally complex, and precise settings of their parameters require a large number of training samples. In order to address these issues, a simplified method that can efficiently extract the spectral-spatial information from a hyperspectral image must be developed. Therefore, the current study proposed a new method for generating the spectral-spatial features of the hyperspectral images. The proposed method uses weighted local kernel matrix representation and minimum noise fraction transformation sequentially and repetitively in order to generate deep spectral-spatial features. The proposed network's spectral-spatial features, which show the local nonlinear relationship between the features extracted from the components of the minimum noise fraction transform at different depths, will finally be stacked together and fed into the support vector machine algorithm for classification. Two hyperspectral benchmark images of the Indian Pines and data from Pavia University are used to test the proposed algorithm. The performance of the proposed method is compared to the spectral classification method and four other spectral-spatial classification methods proposed in recent years. Comparisons show that the proposed method is more accurate in the Indian Pines image more than 20% and in Pavia university image more than 10% than the image classification using the spectral features. In addition, the proposed method is 1% more accurate than the other four spectral-spatial classification methods on average.

Remote sensing satellites provide various data in different parts of the electromagnetic spectrum with spectral, temporal and spatial resolution. In order to make full use of the data obtained from different sources, various numerical and analytical techniques of image integration have been developed. Among the existing image integration methods, due to their high efficiency, speed and spatial accuracy, IHS (Intensity Hue Saturation) and Wavelet Transformation are the most widely used algorithms. But generally, these methods are applied to the entire image all together, and basically whatever its characteristics and contents are, they consider the entire image as a unique object. While from a satellite image of different areas we can get different data and contents. In this research, a new process for integrating images using image analysis based on its surface salience is presented. In this way, the image is divided into two prominent and non-prominent sections, and the integration scenario will be different in these two areas. In the prominent areas, which include residential areas, roads, etc., we used the IHS method which was improved by the genetic optimization method, and in the non-prominent areas (forests, pastures, and agricultural fields) we used the wavelet transformation to analyze and extract the features with high frequency. In this research, in order to implement and evaluate the presented method, samples of images related to worldview 2 gauges have been used. The visual results and the spectral and spatial quantitive ones show the improvement of the integration results compared to the conventional and integrated methods (the output of the assessed metrics CC, ERGAS, RASE and RMSE showed better results compared to the other methods). In addition, the processing speed in this method is much higher than the new techniques which are based on the deep learning networks.

Road extraction from high-resolution remote sensing images has been used in a wide range of applications such as traffic management, route planning, and road navigation. Due to their long length and small width, as well as shadows caused by vegetation and buildings, the detection of the roads challenging. As the roads In an area are of different types such as being near short passages, highways and motorways, we face some difficulties in automatic classifying and recognizing different kinds of roads. In order to improve the reliability and accuracy of extraction of the roads with shorter lengths when there are roads of different sizes,a neural network model is proposed in this paper that achieves pixel-accurate segmentation. The proposed network directly processes the input image and uses four specialized convolutional blocks (SCB) during down-sampling which is complemented by a shallow sampling approach to generate a binary mask for the road class. As the common semantic segmentation networks are deep and have various teachable parameters, the proposed network in this research uses shallow sampling which leads to lessen the network depth and as a result the number of the parameters decreases. The performance of the proposed model in this research was evaluated using the Massachusetts dataset, and the evaluation results clearly show the superior performance of the proposed model compared to the other neural networks with fewer parameters. Compared to the other neural networks such as DEEPLAB3+, U_NET and D_LINKNET, the proposed model was able to improve the IOU and F-Score indices in Massachusetts dataset by 1. 98 and 3. 03, respectively.

Ocean navigation relies heavily on the determination of the position of the floating units. In the past, as there was no satellite navigation system, all the floating units in the ocean calculated their position using celestial bodies and the Sun. With the advent of the satellite navigation system, astronomical positioning using celestial bodies has faded and is only limited to the emergency cases where the floating units are unable to use the satellite equipment. On the other hand, we know that one of the biggest weaknesses in determining the astronomical position in the intersection method and the other vector and drawing methods is related to the approximate position or at least the approximate latitude. In this research, the algorithm for determining the position of a moving or stationary observer by measuring two consecutive heights from the Sun has been developed. One of the prominent points in this algorithm is that it does not need the approximate or initial position of the observer to calculate the position. On the other hand, the difference between this research and the other similar researches is how we calculate the Sun position and how accurate our calculation is. The results of this study show that the best results of positioning can be obtained by measuring two consecutive altitudes of the Sun in about 10 to 20 minutes, and the accuracy of this method is about 10-1 minutes. If the applied theoretical method is used and the accuracy of the calculations of the Sun location is increased, it is possible to achieve the optimal accuracy in the astronomical navigation, which paves the way for future research in obtaining the height of the Sun with very accurate methods (except for the sextant).