Urban areas play a crucial role in carbon absorption, while also producing a considerable amount of carbon emissions. However, there has been a lack of research that has systematically examined the carbon storage and sequestration in green spaces located within urban environments, at a spatial scale. This study analyzes carbon storage and sequestration in Yurim Park, Daejeon, South Korea on a grid basis to fill the research gap. The research compares the variation in sequestration capacity across different grids and provides insights into the development of sustainable urban parks in urban planning. The classification of grids is based on specific site characteristics, such as land cover, tree distribution, type, and density. This results in a total of seven distinct types. The study employs a combination of the I-tree eco model, drone-based modeling, and on-site surveys to estimate carbon storage and sequestration in urban parks. The results show that the average carbon storage per unit area in the entire park was 15.3 tons of carbon per hectare, ranging from a minimum of 5.0 to a maximum of 21.4 tons per hectare. For the planted area, the average carbon storage was 8.6 tons per hectare. Grids with green areas dominated by broad-leaved trees and closed canopy cover had the highest carbon sequestration and storage values. The planting area ratio and the type of trees planted were found to directly influence the carbon sequestration capacity per unit area of urban parks. This study stands out from previous research by conducting a detailed area-based comparison and analysis of carbon sequestration capacity in urban parks using sophisticated measurement techniques. The findings offer direct insights into strategies and policies for securing future urban carbon sinks and can be of practical use in this regard.

In this study, we analyzed the temporal variations in the land surface temperature (LST) in a residential area in Okgye-dong, Jung-gu, Daejeon in South Korea, using unmanned aerial vehicle (UAV) imagery and machine-learning (ML) techniques. We selected an area with a thermally vulnerable environment, high ratio of elderly population, and high potential for heatwaves (during summer). Exploratory data analysis (EDA) was employed to calculate the basic statistics and the relationship between variables. Then, confirmatory data analysis (CDA) was conducted, and a range of nonlinear regression models were employed. The results revealed that the key variables that influenced the thermal environment of the study area varied with time, while highlighting the impact of weather elements. The evaluation of different nonlinear regression models revealed that the alley width, distance from the river, and shadow ratio were importance features that affected the heat vulnerability in the residential areas. Furthermore, an analysis of residuals indicated that with respect to the time-slots wherein the measurements were conducted, the model predictions did not depict any specific pattern. Thus, this study presents an analysis of the thermal environment in urban areas from the perspective of the urban microenvironment. The quantitative assessment and data-analysis methods proposed in this study can be utilized to improve the thermal environment in urban residential areas with high heat vulnerability.

The wildland–urban interface (WUI) and road accessibility are standard explanations of human-caused ignition but are rarely tested where forest and settlement intermix. We analyzed 1043 ignitions recorded in Gyeonggi-do, Republic of Korea, between January 2016 and February 2026. Because coordinates were displaced, each fire was represented by random points within its cadastral parcel, and a weighted case–control model estimated relative ignition intensity. The Radeloff housing-density threshold, applied to households, was exceeded across about three-quarters of the province. Road distance differed between ignitions and controls but, assuming uniform location within parcels, added little information beyond the forest edge and settlement. Among the landscape, access, and settlement variables, proximity to the forest edge had the strongest association. This association arose on non-forest burnable land adjoining forest rather than within forest, persisted when cases and controls were limited to land within 100 m of forest, and partly reflected parcels straddling the forest boundary. Trail proximity was associated with hiker-caused ignition but not detectably with debris burning, although the difference was not distinguishable among small parcels. These findings call for interpreting standard indicators alongside forest-edge land cover and ignition cause.