Forest fires are caused by various climatic and anthropogenic factors. In Republic of Korea, forest fires occur frequently during spring when the humidity is low. During the past decade, the number of forest fire incidents and the extent of the damaged area have increased. Satellite imagery can be applied to assess damage from these unpredictable forest fires. Despite the increasing threat, there is a lack of comprehensive analysis and effective strategies for addressing these forest fires, particularly considering the diverse topography of Republic of Korea. Herein, we present an approach for the automated detection of forest fire damage using Sentinel-2 images of 14 areas affected by forest fires in Republic of Korea during 2019–2023. The detection performance of deep learning (DL), machine learning, and spectral index methods was analyzed, and the optimal model for detecting forest fire damage was derived. To evaluate the independent performance of the models, two different burned areas exhibiting distinct characteristics were selected as test subjects. To increase the classification accuracy, tests were conducted on various combinations of input channels in DL. The combination of false-color RNG (B4, B8, and B3) images was optimal for detecting forest fire damage. Consequently, among the DL models, the HRNet model achieved excellent results for both test regions with intersection over union scores of 89.40 and 82.49, confirming that the proposed method is applicable for detecting forest fires in diverse Korean landscapes. Thus, suitable mitigation measures can be promptly designed based on the rapid analysis of damaged areas.

Seafloor mapping is essential for effective management and sustainable development of marine resources. Various attempts have been made to map the seafloor using single beam echo sounders, multi beam echo sounders, and side scan sonars. The purpose of this study is to map the sea floor using backscatter and bathymetry based on multi-beam echo sounders. For seafloor mapping, seafloor cover was defined as rock, gravel, sand, and mud according to the folk structure, and 135 grab data were collected for seafloor mapping and accuracy evaluation. For seafloor mapping, bathymetry depth and depth-based secondary products (aspect, curvature, slope, roughness, eastness, northness, mean, standard deviation) and backscatter intensity and secondary products that can be produced from intensity (mean, variance, roughness) was established. In addition, the output of the GLCM algorithm (angular second moment, contrast, dissimilarity, energy, entropy, homogeneity, max, mean, standard deviation) was constructed to extract various features of backscatter intensity. For seafloor cover, a random forest model, a machine learning technique that shows high performance in various fields, was selected, and the ratio of training and test datasets was selected as 8:2. To improve the performance of the random forest model, a hyperparameter was selected by applying a 5-fold cross validation and grid-search method, and the overall accuracy was 0.83.

For mapping of seabed substrate, it is essential to extract and analyze seabed substrate characteristics from measured data such as bathymetry and backscatter.The number of studies have been conducted on seabed substrate classification based on backscatter data obtained from side-scan-sonar, single-beam-echo-sounder, multi-beam-echo-sounder (MBES).These studies presented limitation of low accuracy by mapping of seabed substrate through statistical methods or principal components analysis.The purpose of this study is mapping of seabed substrate using machine learning methods like random forest (RF), support vector machine, and k-nearest neighbor and MBES-based variables.MBES-based variables were built total 23 variables from bathymetry, backscatter, and backscatter-based GLCM statistical methods.Seabed substrate classification schemes were defined 4 classification scheme according to the ratio of gravel, mud, and sand.In the case of RF model and case 4 classification scheme, the F1 score showed the highest accuracy at 82.0%.Through this study, machine learning and MBES-based variables were used to confirm the possibility of high-accuracy mapping of seabed substrate, and it is believed that it can be sued to mapping of seabed substrate in other regions.

Jeongmin Seo, Sanghun Son, Jaegu Bae, Doi Lee, Seonghyeok Lee, Soryeon Park, Chaean Lee, Moonsoo Lim, Dongju Seo, Jinsoo Kim. Korean J. Remote Sens. 2025;41:1021-36. https://doi.org/10.7780/kjrs.2025.41.8

Air pollutants, such as large amounts of carbon dioxide produced by fire, are at risk of promoting global warming, causing more frequent and more wildfires all over the world. Large-scale wildfires cause air pollutants to spread, resulting in a significant increase in fine dust concentration. And human damage, property damage, and natural ecosystem damage are increasing. Rapid and accurate information of fire damaged the forest areas using high-resolution satellite imagery is effective in preparing wildfire prevention measures and monitoring burned areas. In this regard, the many research is underway to estimate fire burn damage using the difference spectral feature between healthy forests and burned forests. There are many spectral indices including well-known indices such as NDVI(Normalized Difference Vegetation Index) and NBR(Normalized Burn Ratio). It is possible to estimate burned area by computing a difference image representing a change in the spectral wavelength of the image between pre- and post-fire. In the case of South Korea, it is difficult to estimate wildfires because the spectral characteristics of vegetation vary from each season according to climate. And Compared to other countries, Forest fires are occurring small scale of forest fires less than 2,000ha. In our study, the accuracy was compared by applying various spectral indices to the estimation and evaluation of burned area in South Korea, using one of the deep learning models U-Net. As a result of the IoU value of 0.80 or more, it was confirmed that it was possible to calculate the forest fire damage site.Acknowledgments:This research was supported by a grant (2021-MOIS37-002) of "Intelligent Technology Development Program on Disaster Response and Emergency Management" funded by Ministry of Interior and Safety (MOIS, Korea).This work was supported by the "Graduate school of Particulate matter specialization" of Korea Environmental Industry & Technology Institute grant funded by the Ministry of Environment, Republic of Korea.