Mid-infrared (MIR) image is highly valued across various fields, such as national defense and environmental monitoring, due to its capability to capture temperatures of objects and surfaces. Image registration that unifies coordinates between images is a fundamental process for utilizing multi-temporal satellite images. The radiation-invariant feature transform (RIFT) algorithm is a feature-based matching method that extracts robust matching points to non-linear radiometric distortions in day and night MIR images. However, the original RIFT method has a limitation in detecting a small number of matching points because the properties of the image are not sufficiently considered. In this study, we propose an optimization method of RIFT for MIR day-night image registration. First, patch size is selected so the RIFT algorithm can stably acquire multiple matching points. After comparing the results of applying RIFT by setting various patch sizes, we select the final patch size that extracts the most matching points. In addition, the RIFT algorithm’s hyperparameters are optimized to suit the characteristics of the MIR image by comparing the number of matching points and RMSE for each combination. Finally, the image registration is conducted using a transformation model based on extracted inlier points by applying random sample consensus (RANSAC), data snooping, and locality preserving matching (LPM), which are outlier removal algorithms. Based on experiments conducted from KOMPSAT-3 MIR day/night satellite images, the LPM algorithm produced the best quantitative evaluation result with an average RMSE and circular error of 90% (CE90) of 0.984 pixels and 2.076 pixels. From the experiments, it was demonstrated that the proposed method can contribute to improving image registration by effectively extracting matching points that reflect the characteristics of KOMPSAT-3A MIR day-night imagery.

The COVID-19 pandemic significantly reduced human activities globally, leading to various changes in urban environments. Previous studies analyzing the pandemic’s impact on urban heat islands (UHI) have heavily relied on the Moderate Resolution Imaging Spectroradiometer (MODIS) land surface temperature (LST) dataset with a spatial resolution of 1 km. However, such a coarse resolution of MODIS failed to adequately capture the complexity of urban structures, making it difficult to explore the change of UHI accurately. The objective of this study is to analyze the impact of changes in human activities on UHI before and after the COVID-19 pandemic using downscaled MODIS 250 m LST in Daegu Metropolitan City, South Korea. The spatial downscaling process employed a local linear forest model, with Lasso feature selection used to extract input kernels with high spatial correlation. The validation was conducted using reference LST data from Landsat and ECOSTRESS that overlapped with MODIS LST. The downscaled LST250m had a correlation coefficient (r) of 0.790–0.929 and a root mean square error (RMSE) of 0.731–1.333°C during the daytime, and an r of 0.892 and RMSE of 0.771°C at night. Compared to MODIS LST1km (daytime: r of 0.771–0.882, RMSE of 0.990–1.497°C; nighttime: r of 0.857, RMSE of 0.906°C), the downscaled LST250m exhibited better accuracy. The intensity of surface UHI (SUHI) was calculated using the LST250m to analyze its spatiotemporal changes. The average normalized SUHI intensity before and after COVID-19 at the administrative district level revealed varied across regions. Residential areas showed an increase in normalized SUHI intensity during the night, while commercial areas exhibited a decrease, which was associated with the landcover ratio within each district. The results showed that such changes were dominant in residential, commercial, and transportation areas that were highly associated with human activities. The exploration of the impact of changes in human activities by COVID-19 on UHI using downscaled LST data will contribute to a further understanding of urban climate change and our knowledge of urban resilience.

The purpose of this study is to quantitatively and spatiotemporally analyze the effects of cool roof installations on mitigating urban heat island (UHI) phenomena. By utilizing unmanned aerial vehicle (UAV) and thermal infrared sensor (TIR), the reduction in land surface temperature (LST) due to cool roofs, a key heat mitigation measure, was analyzed across different times of the day. The research was conducted in Jangyu Mugye-dong, Gimhae-si, Gyeongsangnam-do, where cool roofs were implemented as part of a pilot project aimed at mitigating UHI effects. High-resolution thermal images were captured at two-hour intervals from 9 AM to 5 PM on a clear day using UAVs, and the spatiotemporal distribution of LST was analyzed in detail using box plots and z-scores. The results revealed that the cool roof exhibited the most significant temperature reduction effect during the morning hours (9 to 11 AM). The time with the greatest temperature difference, based on the second quartile (Q2), was 11 AM, where the cool roof’s LST was 10.52°C lower than that of a conventional roof. Conversely, this difference decreased as the afternoon progressed, reaching the smallest temperature difference of 0.04°C at 5 PM. The spatial trends of LST between cool roofs and conventional roofs were analyzed using a box plot and z-score analysis for each period. Additionally, roof objects with extreme LSTs—classified as those with values beyond the absolute range of 1.65—were identified, and the frequency of such objects was determined for each period. The analysis showed that cool roof areas consistently maintained an LST that was on average 5–10°C lower than that of conventional roofs. The highest LST observed for conventional roofs peaked at 66.02°C at 11 AM, while the lowest LST for cool roofs was 35.22°C, showing a substantial difference of approximately 30.80°C. This study presents a case demonstrating that the application of cool roofs is an effective strategy for mitigating urban heat island effects. By analyzing the temporal LST patterns in the study area and assessing z-scores for individual roof objects, the research highlights the effectiveness of cool roofs, particularly in the morning when solar radiation is lower. The findings of this study can be utilized for the broader application of heat mitigation facilities, optimal installation, and management strategies, as well as further research on effective urban heat island reduction techniques.

The reduction of natural cover and the increase in artificial surface caused by urbanization lead to surface urban heat island. New towns are considered an appropriate study site to analyze surface urban heat island affected by land cover transformations, due to their development characteristics of rapid urbanization. This study aims to monitor surface urban heat island intensity (SUHII) resulting from land cover changes. We study Magok New Town as a representative urban development case in Seoul, South Korea. SUHII is defined as the difference in the mean land surface temperature (LST) between developed and undeveloped areas, with LST measured using Landsat satellite imagery. We build long-term data for the observation period from April 16, 2003, to October 15, 2023, considering the development period of Magok New Town. We gather 101 high-quality data points from approximately 20.5 years to enable detailed and comprehensive observation of land surface temperature changes. We compare changes in Magok’s SUHII (experimental group) with those of four surrounding areas, two undeveloped and two previously developed areas (control groups). The results show that the SUHII of Magok New Town has converged to resemble the surrounding developed areas as urban development has progressed. In particular, the two developed areas in Gangseo-gu exhibit relatively high SUHII variability throughout the observation period, whereas the SUHII variability in Magok is observed to gradually increase over time. The cause of this phenomenon is that SUHII typically shows higher values in the summer and lower values in the winter, and urban areas, the seasonal variation in SUHII is significantly larger. Therefore, as urban development progressed in Magok, the seasonal variation in SUHII widened, deteriorating Magok’s urban microclimate. The spatial distribution of SUHII is relatively high in airports, transportation facilities, and industrial, and construction areas due to the influence of surface environment and urban activity characteristics. In the residential area of Magok, SUHII is lower than in Gayang, due to the underground construction of transportation facilities and the design of parks. The green area in Magok Central Park offsets the low SUHII during the Cool and Cold seasons, but its effect is limited during the Hot and Warm seasons.

In South Korea, earthquakes of magnitude 4.0 and above continue to occur, yet there remains a lack of awareness regarding liquefaction caused by these seismic events. In this context, evaluating liquefaction during seismic events, using geotechnical information from the National Ground Information Portal System consumes considerable human and temporal resources to derive the Liquefaction Potential Index (LPI). Additionally, there are issues concerning the accuracy of the input data. This study designed an LPI prediction model to address these challenges by comparing the calculated LPI with a reference map indicating liquefaction susceptibility. The results showed that while the LPI prediction model requires some improvement in accuracy compared to the actual LPI, it demonstrated high effectiveness in rapidly selecting ground investigation locations during national disaster situations.

Heat wave-related information is important in establishing effective heat wave response policies. The National Disaster Management Research Institute (NDMI) produced a heat distribution map that can observe heat phenomena by estimating the highest daily temperature based on satellite images in 2017 and supports local governments’ heat wave response policies. For effective heat wave management, high-resolution information on areas such as heat distribution maps is also important; however, various information is required to develop national heat wave measures. Currently, information related to heat waves is provided independently by institutions, such as weather, safety, and health, so it is inconvenient to acquire and use them individually. Accordingly, the NDMI developed an information-integrated display service (heat wave comprehensive information system) that allows heat wave managers to understand situation management information in real time. The heat wave comprehensive information system provides data on weather, safety, and health in an integrated, and expresses it in a map type, a numerical type, and a table and graph. This study presented the main information of the comprehensive heat wave information system and future operation plans. It is believed that it will contribute to the reduction of damage by supporting the regional heatwave monitoring system through the developed heatwave comprehensive information system.

This study evaluated the reliability and applicability of the Geostationary Environment Monitoring Spectrometer (GEMS) data for monitoring atmospheric conditions in South Korea. This study analyzed correlations between GEMS satellite products (Aerosol Optical Depth [AOD], NO2, HCHO, Ozone Profile [O3P]), Airkorea ground-based data, Local Data Assimilation and Prediction System (LDAPS) meteorological data, and land cover map across 161 administrative districts from 2021 to 2023. Despite the satellite’s low spatial resolution, analysis revealed significant correlations between GEMS observations and ground-based measurements. GEMS AOD showed strong negative correlations (–0.6) with atmospheric instability indices and positive correlations with ground PM2.5 (0.57) and carbon monoxide (CO) (0.6) measurements in spring, while summer ozone measurements demonstrated high correlations with ground observations (0.6) and temperature (0.59). Particularly strong correlations were observed in spring and fall, with GEMS AOD showing a distinct positive correlation with ground Particulate Matter (PM) concentrations and a negative correlation with atmospheric instability. The study found varying correlation patterns across different land cover types: urban areas demonstrated high positive correlations between GEMS AOD and PM substances, while forest regions showed generally lower pollutant concentrations, confirming their air purification function. Seasonal analysis revealed complex patterns, with spring and fall showing more interpretable correlations between variables compared to summer. Interpreting correlation patterns in summer was difficult due to the unique atmospheric meteorological factors of the Korean Peninsula. Regional analysis showed effective capture of pollutant transport phenomena, particularly along the west coast where spring westerly winds influence pollution patterns. The study also found that meteorological conditions, especially Boundary Layer Height (BLH) and atmospheric instability, significantly influenced pollutant concentrations and their spatial distribution. These findings suggest that GEMS satellite data can effectively complement ground-based monitoring networks for comprehensive air quality assessment in South Korea, particularly in monitoring broad-scale pollution phenomena and tracking pollutant transport patterns. However, there are limitations to local spatial analysis and special meteorological phenomena, so satellite and ground observation data must be integrated to build an optimal air quality monitoring system.

Accurate and efficient tree type classification in urban forests is crucial for effective management, informed policy decisions, and enhancing urban resilience, particularly with increasing urbanization and climate change. This study developed and evaluated a practical methodology for classifying coniferous and broadleaf trees in the Chungbuk National University Arboretum, South Korea. The study utilized drone-acquired, high-resolution RGB imagery and a Support Vector Machine (SVM) classifier. The workflow encompassed drone image acquisition, concurrent ground truth data collection, image preprocessing, feature extraction (including RGB color bands and Gray-Level Co-occurrence Matrix [GLCM] texture features [TFs]), and SVM model training, optimization, and evaluation. Different SVM kernels (Linear, RBF, Polynomial, Sigmoid) and feature combinations were investigated to optimize model performance, with a specific focus on processing time for practical application. Results indicated that RGB color bands were the primary drivers of accurate classification, while most GLCM TFs provided minimal additional benefit in this specific context. The RBF kernel, with optimized hyperparameters (C=10, γ=0.01), achieved the highest overall accuracy (99%) and F1-score (0.99), while the Linear kernel provided similar accuracy but with a longer processing time. Notably, the drone-based classification significantly outperformed the outdated Korea Forest Service forest map in representing the current forest composition, highlighting the limitations of traditional mapping methods for dynamic urban environments. This research contributes a cost-effective and accurate method for urban forest assessment, demonstrating the value of drone technology and readily available RGB imagery. The entire process, from image acquisition to classification, was completed in approximately 12 hours, showcasing its efficiency. Although this study focused on only two tree types in a single season, the developed methodology shows potential for broader application in classifying a wider range of species and informing management practices across different seasons by considering the phenological stages of trees. The proposed approach provides urban planners and forest managers with a valuable tool for enhancing ecosystem services and improving the quality of life in urban areas. This study underscores the potential of drone technology to revolutionize urban forest monitoring and management practices, paving the way for more sustainable and informed decisionmaking, particularly in rapidly urbanizing regions.

This study compares two resampling methods, the k-dimensional tree (kd-tree) and the Python interface for Terra Advanced Fusion Resample/Reprojection functions (PyTAF), using surface albedo data from the GEO-KOMSAT-2A (GK2A) Advanced Meteorological Imager (AMI) geostationary satellite, with and without consideration of Earth’s curvature. Evaluation metrics include Correlation Coefficient (R), Root Mean Square Deviation (RMSD), Relative RMSD (RRMSD), Bias, spatial distribution, and processing time. Additional quantitative analyses were performed based on viewing zenith angle (VZA) intervals, ranging from 0° to 80° at 20° increments. The results showed that both resampling methods exhibited similar performance in terms of quantitative metrics, but differences emerged in processing time and VZA-specific analysis. These differences were primarily attributed to variations in algorithm design. Specifically, as VZA increased, the panoramic effect caused each pixel to cover a larger geographic area, resulting in geometric distortions. Additionally, the influence of reflectance variability between snow-covered and non-snow-covered regions further exacerbated data uncertainty and geometric distortions. These combined factors contributed to reduced accuracy and increased errors during resampling, leading to higher RMSD and RRMSD values. This study provides empirical evidence of the performance differences between the two resampling methods, offering practical insights for selecting the optimal resampling technique based on research objectives and data conditions.

Deep learning technologies have revolutionized image processing and analysis, introducing groundbreaking innovations that significantly improve the accuracy and efficiency of object segmentation, especially in satellite imagery. The increasing availability of high-resolution satellite images has created a demand for advanced models capable of handling the complexities of object detection in diverse environments. This study investigates the potential of the Segment Anything Model for Remote Sensing (SAMRS), a deep learning framework specifically designed for remote sensing applications, to accurately identify and segment a wide range of objects within satellite imagery. The model was trained using prominent datasets such as Dataset for Object Detection in Aerial Images (DOTA), Dataset for Object Detection in Optical Remote Sensing Images (DIOR), Fine-grained Object Detection in Aerial Images for Remote Sensing Version 2.0 (FAIR1M-2.0), and Instance Segmentation in Aerial Images Dataset (iSAID), enabling it to learn diverse object features and complexities. The evaluation of SAMRS was conducted on Northwestern Polytechnical University Very High Resolution 10-Class Dataset (NWPU VHR-10) and Beijing-3B datasets, where it demonstrated impressive results. In vehicle detection tasks, SAMRS achieved an Intersection over Union (IoU) of 0.9175, an F1-score of 0.9570, and an accuracy of 0.9385. These metrics highlight SAMRS’s capability to automate object detection in complex satellite images, overcoming challenges posed by intricate backgrounds and diverse object sizes. Furthermore, SAMRS is optimized to analyze both large and small-scale objects, ensuring robust performance across varying conditions. The findings emphasize the model’s utility not only for current remote sensing applications but also for future extensions involving drone imagery and domestic satellite datasets. By automating object detection and segmentation, SAMRS has the potential to transform practical fields such as urban planning, disaster management, traffic monitoring, and environmental analysis, making it a vital tool in advancing satellite imagery analysis.