The research of using synthetic aperture radar (SAR) satellite imagery to estimate soil moisture is an important application of satellite image sets. In this study, a soil moisture estimation study was conducted by integrating Sentinel-1 SAR imagery provided by Google Earth Engine (GEE) and X-band KOMPSAT-5 SAR. Vegetation indices calculated from high-resolution KOMPSAT-3A optical imagery were used as vegetation inputs required by the water cloud model used for soil moisture estimation. Meanwhile, accurate land cover maps provided by the Korea Environmental Information Service were used as objects for soil moisture estimation. The study area is part of the Gimje Plain, a major rice growing area along the west coast of the Korean peninsula. The estimated results were validated using data from the Soil Moisture Active Passive (SMAP) satellite, as a test. The differences between SMAP L-3 data and L-4 data with estimated result were found to be 4.84±6.78%p and 3.09±5.30%p, respectively, showing significant accuracy levels. This experimental result suggests the potential of integrating SAR imagery with GEE for soil moisture estimation applications.

Using multi-temporal KOMPSAT-3/3A high-resolution satellite images, the Normalized Difference Vegetation Index (NDVI) for the area around the Fukushima daiichi nuclear power plant was determined, and the pattern of vegetation changes was investigated. As well, Landsat-8 and Sentinel-2 images were used. The NDVI products obtained from these three sensors were validated and compared with land cover and land use map of the same area for visual interpretation. The main results of this study showed that the NDVI in the mid-2010 was low within area-of-the interests in the vicinity of the nuclear power plant, but the vegetated area continued to expand until 2021. It can be interpreted as a result of a decrease in human activities. In particular, the map of differences between the mid-2010 and 2021 of high-resolution images revealed detailed spatial features rather than those of the mid-resolution images. In disaster areas where access is restricted, such as the areas around the Fukushima nuclear power plant, and where it is difficult to collect field data, obtaining classification products with high accuracy using satellite images is challenging. It helps to analyze them using primary products such as vegetation indices exactly determined from the high-resolution satellite imagery.

We analyze the dynamics of large batch stochastic gradient descent with momentum (SGD+M) on the least squares problem when both the number of samples and dimensions are large. In this setting, we show that the dynamics of SGD+M converge to a deterministic discrete Volterra equation as dimension increases, which we analyze. We identify a stability measurement, the implicit conditioning ratio (ICR), which regulates the ability of SGD+M to accelerate the algorithm. When the batch size exceeds this ICR, SGD+M converges linearly at a rate of $\mathcal{O}(1/\sqrt{\kappa})$, matching optimal full-batch momentum (in particular performing as well as a full-batch but with a fraction of the size). For batch sizes smaller than the ICR, in contrast, SGD+M has rates that scale like a multiple of the single batch SGD rate. We give explicit choices for the learning rate and momentum parameter in terms of the Hessian spectra that achieve this performance.

We propose a new framework, inspired by random matrix theory, for analyzing the dynamics of stochastic gradient descent (SGD) when both number of samples and dimensions are large. This framework applies to any fixed stepsize and on the least squares problem with random data (finite-sum). Using this new framework, we show that the dynamics of SGD become deterministic in the large sample and dimensional limit. Furthermore, the limiting dynamics are governed by a Volterra integral equation. This model predicts that SGD undergoes a phase transition at an explicitly given critical stepsize that ultimately affects its convergence rate, which is also verified experimentally. Finally, when input data is isotropic, we provide explicit expressions for the dynamics and average-case convergence rates (i.e., the complexity of an algorithm averaged over all possible inputs). These rates show significant improvement over classical worst-case complexities.

Open Data Cube (ODC) is an open source platform that allows analysis experts to easily and quickly access and process satellite information that observes the earth. The platform is stored and managed in the form of a data tube based on image and time series information including geographic locations. ODC offers modules to ingest a number of free satellites. Data ingestion means the process of obtaining and importing data for immediate use or storage in a database. Developing and providing an ODC-applicable ingestion module directly from the country that owns and operates the satellite increases the value of its use. Korea possesses and operates the multiple KOMPSAT series, a high-resolution optical and SAR satellite images related to Earth observation. Technical methodologies for processing satellite data are important because it is significantly larger as it is collected. ODC is basically a project designed with these considerations, and it is possible to apply the latest technology in how it operates, as well as highly scalable cloud computing technology. Not only is there a commercial, ready-to-use cloud computing services, but there are many technologies that have been released as an open source. With all these aspects, the processing modules to ODC for applying KOMPSAT images to open source cloud computing services were developed in this study. It is expected that KOMPSAT data will be useful to the users of ODC using these scripts and that this case can be referenced by users to build their own cloud services and operate ODC.

The advance of DRAM manufacturing technology slows down, whereas the density and performance needs of DRAM continue to increase. This desire has motivated the industry to explore emerging Non-Volatile Memory (e.g., 3D XPoint) and the high-density DRAM (e.g., Managed DRAM Solution). Since such memory technologies increase the density at the cost of longer latency, lower bandwidth, or both, it is essential to use them with fast memory (e.g., conventional DRAM) to which hot pages are transferred at runtime. Nonetheless, we observe that page transfers to fast memory often block memory channels from servicing memory requests from applications for a long period. This in turn significantly increases the high-percentile response time of latency-sensitive applications. In this paper, we propose a high-density managed DRAM architecture, dubbed 3D-XPath for applications demanding both low latency and high capacity for memory. 3D-XPath DRAM stacks conventional DRAM dies with high-density DRAM dies explored in this paper and connects these DRAM dies with 3D-XPath. Especially, 3D-XPath allows unused memory channels to service memory requests from applications when primary channels supposed to handle the memory requests are blocked by page transfers at given moments, considerably increasing the high-percentile response time. This can also improve the throughput of applications frequently copying memory blocks between kernel and user memory spaces. Our evaluation shows that 3D-XPath DRAM decreases high-percentile response time of latency-sensitive applications by ~30% while improving the throughput of an I/O-intensive applications by ~39%, compared with DRAM without 3D-XPath.

Trend technologies such as big data, cloud, Internet of things remain to be progressed and applied in many ways both industry and academics. Remote sensing application or geospatial information system have been regarded as important application domains linked to those huge technologies. This study considered the linkage of satellite image processing services to cloud computing on Platform as a Service (PaaS) for easy access to development components. PaaS is a service driven by infrastructure and helps to simplify the use of multiple technology elements as a service unit. The Web Processing Service (WPS), Open GIS standard, supporting online asynchronous processing is also an important element for multiple data processing of large volumes of satellite image sets. This study presents the generalized scheme to develop web-based satellite image processing application with WPS as a service based on PaaS cloud. All computing resources used in this study are fully open source containing GeoServer and ZOO project for geo-processing and OpenStack and Cloud Foundry for cloud computing. This approach will be useful to build cost-effective and service-oriented remote sensing applications with image processing functions in the standpoint of system engineering.

The cloud computing and mobile environments since mid-2000s are regarded as important keywords in ICT fields. Meanwhile, mobile cloud computing is receiving attention in aspect that uses advantage of cloud computing and mobile at the same time. Recently, the service and application in various fields are developed using mobile cloud computing. But in the remote sensing field, the examples of research and development are but little. In this study, the mobile app for satellite image processing was developed using cloud computing, iOS SDK, and various libraries, and is running on tablet PC. The developed mobile app can be one of the examples for using mobile cloud computing in the remote sensing field. Mobile cloud computing is expected to steadily grow. Therefore, the mobile app based on mobile cloud computing in the remote sensing field will require the development.

Recently mobile device is regarded as a popular platform of geo-based applications. Web browser on mobile is also enough to support mobile web services. Demand of mobile web app handling geo-based data is increasing in the many application fields. Among various possible approaches for actual implementation, 3D geo-data visualization service is the most prospective fields related to 3D real world modeling. For this purpose, 3D terrain modeling and fusion rendering using WebGL of HTML5 supporting graphic accelerator without plug-in components is possible. This study presents a test case of mobile web app for 3D fusion themes taking into account of several data types and multiple formats including satellite images.

In the future, most mobile nodes will have multiple radio interfaces, and this feature can be exploited to reduce the transmission cost in wireless data access applications. In this work, we propose cooperative wireless data access algorithms with strong consistency in multi-radio wireless networks. It can be shown that cooperation with neighbor nodes can reduce the expensive transmission cost over wireless links.

A three dimensional discrete wavelet transform (3D DWT) expands planes of the 2D DWT into volume data. The 3D DWT scheme has advantages of analyzing spectral characteristics in the order of frequency and analyzing changes of spatial information and spectral information simultaneously. Nonetheless, few researchers have attempted to classify multi-spectral images using the 3D DWT. This study aims to apply the 3D DWT to the classification of multi-spectral images and synthetic aperture radar image. To classify these images, we employ two numerical values: The 3D wavelet coefficients and the energy of each band. And then we evaluate their results quantitatively with those of traditional classification techniques including the 2D wavelet scheme. Therefore, the results show that the classification technique of the 3D DWT is more effective than that of traditional techniques, especially in complicated imagery. The accuracy of the 3D energy in SAR imagery is higher. This study provides new numerical values to classify images effectively. Furthermore, these values can be extended to the image retrieval and pattern recognition in high-resolution imagery and this scheme can be employed in image dataset obtained at other times.

Currently, the application domains of the commercialized high resolution satellite imagery are in the middle of expansion. These images contain detailed information to investigate targets in the application level. In this paper, we try to demonstrate traffic analysis scheme based on template matching method algorithm. The simple matching algorithm implemented for this purpose is as follows: First, one user determine a type of vehicle or other traffic feature, as a template corresponding to a target feature. In this process, it is possible to choose multiple templates according to vehicle modes. Second, one can automatically detect and count the number of vehicles statistically similar to template given in the previous step through template matching method algorithm. While, if user already knows the geo-referenced location of a specific vehicle, traffic flow analysis in the static status is also possible in this step. Through the results from each counting vehicles, user can summarize statistics of traffic features at a certain specific time in high-resolution satellite imagery. As for the actual test and experiment, satellite imagery of KOMPSAT EOC was used, and KOMPSAT imagery consists of feature sets in an urban region showing complex types of spatial features. Results of template matching method can provide useful ones in detecting the types of vehicle in the remote traffic detection or monitoring system. In conclusion, this approach can be utilized to obtain some information about on-road traffics such as the location of vehicle, traffic volume, and other transportation parameters, related to transportation uses of remote sensing image.

Commercialized uses of high-resolution satellite imagery have attributed new domains to remote sensing. This study focuses on transportation domain using remotely sensed imagery, with new implementations running on GIS (geographic information System)-based environment. In the mean time, GIS has been regarded as one of the effective methodologies in the transportation problem solving. In transportation geography, circuity terms a concept to represent the difference extent between actual nodes and fully connected nodes in the analysis zone, and accessibility is to represent accessibility level in the matrix form by connectivity level between node to node in a zone. In this study, GIS-based application program is first implemented to perform analytical function such as circuity and accessibility estimation with base road layer data as road graph network. Quantitative estimation of circuity and accessibility is composed of several steps with respect to the graph-typed network structure in an arbitrary analysis zone selected by analyst or a given administrative boundary zone. Data of this approach is not required complex database information based on transportation data model, but some layers capable of directly obtaining from normal digital map sets, In practical application using this GIS-based extension program, high-resolution imagery can be effectively used for selection stage of analysis zone as traffic analysis zone. This kind of application scheme is useful one to road planner or transportation planner dealing with geo-spatial imagery on GIS environment.

In these days, the demand on methodologies and processing schemes for urban application or analysis using high spatial resolution imagery is increasing. In fact, most methodologies for intermediate or coarse spatial resolution imagery can also be applied for high-resolution one, and some processing methods among them show practical usefulness over those imagery data. In use of high spatial resolution imagery, change detection based on accurate extraction of target features of interests in a certain region including complex urban area or inaccessible area is of importance. For this change detection approach, two processing techniques such as homomorphic filtering and morphological operations are considered in this study. After applying other basic image rectification processing, homomorphic filtering which combines non-linear mapping function and spatial frequency filtering is helpful to reveal distinct characteristics for some complicated and multi-associated features on high spatial resolution imagery, and then morphological operations are fast and efficient processing method to detect change with homomorphic filtered and multi-temporal imagery datasets. As a result, this approach with both techniques is thought to be one of the applicable schemes for the change detection problem with high spatial resolution imagery, in the aspects of increased accuracy and automatic extraction of target features.

Potential uses of high-resolution satellite imagery such as KOMPSAT (KOrea Multi Purpose SATellite) EOC (Electro Optical Camera) or IKONOS, in the field of urban planning and transportation planning are widely recognized; currently, actual applications in these fields have been tried in the most countries. However, most approaches, in some extents, are urban feature extraction by automatic or semi-automatic methodologies and human interpretation or recognition with other ancillary digital information. In this study, automatic road feature extraction firstly was attempted with newly implementation of GDPA (Gradient Direction Profile Algorithm), proposed by Wang and Zhang (2000). Further, road features extracted from some ortho-rectified KOMPSAT EOC and IKONOS imageries at a case study area in nearby Seoul, Korea, as results of GDPA, were evaluated with actual digital GIS road layers of National Geographic Institute in Korea, by computation of commission and omission error assessment. Second, these road features, after accuracy assessment, were used to extract quantitative indices such as α index, and γ index, in the area of interests. These indices are for providing information on connectivity status of road network, and also related to accessibility of multi-link structure. Conclusively, it is thought that the results and products in this study can be effectively utilized to local government applications such as urban planning and transportation planning for transportation geographic analysis associated with high-resolution remotely sensed imageries.

Now most Internet map services are provided on the Web, but most of those approaches depend on images in graphic format. These types of image format require relatively large-sized files, and have some problems with data transport and storage on the Web. In addition, as the meaning of the data is not captured as parts of the data itself, data translation, when it happens, must then be accompanied by painstaking manual process to restore the meaning. To overcome these problems, GML (Geographic Markup Language) has developed as the Web standard format of geographic information. Although GML focus on describing the geographic information as the type of XML (eXtensible Markup Language), there are few cases applied to digital geologic maps till now. In this study, Internet mapping of a Korean digital geologic map was accomplished using GML as a case study. Digital geologic map sets consist of geologic boundary, geo-text, fault, cross section, lithology and other complex items to represent geo-based attributes. Each layer was described to both spatial and non-spatial information using GML profile proposed by OGC (open GIS Consortium, Inc.). Spatial and non-spatial information were described as geometry of GML and properties of geographic features, respectively. The geographic information described by GML was transformed into SVG (Scalable Vector Graphics), one of the XML standards for graphic format. Vector maps can be visualized via the Internet without proprietary software on the Web browsers. SVG-enabled browsers allow users to control their view with zoom and pan as they do so in GIS software. As the content and the representation of the geographic information were separated in the Internet mapping process of digital map, it is possible to produce multiple representation about the same geographic information and the dynamic customization on user's need, and the spatial analysis on the Web. Conclusively, GML-base geologic map produced in this study can be transfer rapidly without loss of the meaning and degrading on the Web. So it can be effectively utilized to not only Internet map services but also LBS (Location Based Services) such as PDA or a mobile platform.

This study focuses on transportation application using remotely sensed imagery, in GIS-based environment; GIS is regarded as one of the important and effective methodologies in the transportation problem solving. Generally, transportation application is the complicated task, containing GIS-T (GIS for Transportation) database building and specialized analytical processes; therefore, this study aims at accessibility index extraction using high-resolution imagery, applicable to transportation planning considering an urban environment among transportation. Accessibility, defined as measuring the easiness of reaching from one place to other place, is an element in analyzing the efficiency of a transportation system in its planning process. As for attractive factors in transportation accessibility, many variables such as travel cost, travel time, job opportunity, and the number of facilities or population can be applied. Among them, some parts are associated with visually recognizable features revealed on remotely sensed imagery, and accessibility indices can be utilized to road network or by-pass design problem, and public transit planning. In this study, the gravity-typed accessibility measures are implemented in ArcView-GIS environment with Image Analysis extension. Some ortho-rectified KOMPSAT imageries and IKONOS imageries in local area nearby Seoul-city are used in this study, with digital GIS road layers of the National Geographic Institute in Korea. Conclusively, it is thought that thematic information extraction from high-resolution imagery with GIS databases contributes to expand application fields in remote sensing.

This paper discusses the application of remote sensing techniques and GIS-based prediction modeling for landslide hazard mapping. First, we aim at identifying the past landslide locations by comparing various remote sensing data (KOMPSAT EOC, IRS, SPOT). Using the detected landslides locations and spatial database, probabilistic prediction model based on likelihood ratio function is constructed to generate a prediction map for landslide hazard mapping. After generating the prediction results, to provide the measures of significance of prediction results, a validation of the prediction results is carried out. Through the validation procedure, the prediction results can provide meaningful interpretation with respect to the future landslides. A case study from Jangheung area in Korea is carried out to illustrate above schemes.

In this study, at Haenam area, west-southern part in Korean peninsular, hydraulic properties such as specific discharge and groundwater level measures were obtained at over 2000 well sites. As for remotely sensed data, LANDSAT TM data in Dec., 1986 and Dec., 1998 was processed and classified with 5 classes including some agricultural ones and water zone; besides, these remote sensing data was used to extract geologic or geomorphic lineaments related to fracture zones. Moreover, geologic alteration zones using remotely sensed imagery were also extracted and compared with actual geologic map sets, in the emphases on hydraulic aspects. While surfacial features such as DEM-based geomorphic variables, drainage pattern, water drainage density were extracted; furthermore, water zones, vegetation index, and temporally changed zones were estimated using Landsat TM data. Then they were statistically analyzed with specific water capacity at 11 sub watershed zones. As for additional field information for water resources characterization, the aquifer test using hydraulic properties such as Transmissitivity (T) and Storage capacity (S) was performed. As results, drainage pattern and hydraulic measures at each watershed zone were presented and analyzed, to characterize potential water resources.

Recently, spatial information technologies of GIS (Geographic Information Systems) using remotely sensed have been widely utilized to transportation-related systems, as a part of ITS (Intelligent Transportation System); however, application of remotely sensed imagery to this field is at the beginning stage, in general. Application of remotely sensed imagery in consideration to analysis with vector-typed digital data sets in the node-link structure can effectively used in transportation planning, being very large and complex task using lots of data sets in various fields. In this study, this reference modeling with Landsat 7 ETM+, IKONOS 1 m color imagery, and KOMPSAT EOC in 2000 was carried out in a local governments, Namyangju-city, nearby metropolitan Seoul, Korea. While, thematic Land-use digital data sets in 1:5 K and 1:1 K scale were used to estimate urban growth and transportation-related land-use status. Meanwhile, some algorithms of traffic flow estimation and urban-rural indexing using land-use and DEM (Digital Elevation Model), as well as multiple remote sensing data with different resolution mentioned above were implemented as ArcView/spl trade/-GIS extensions, and applied actual data sets in this case study areas. It is concluded that remotely sensed imagery at transportation planning problems would be provide useful integrated information with multiple data sets.