https://doi.org/10.12652/Ksce.2026.46.3.0193

최준성(Choi, Junseong);김상혁(Kim, Sanghyeok);류지원(Ryu, Jiwon);장은경(Jang, Eunkyung);지 운(Ji, Un)

Climate change-induced extreme flooding necessitates complementary flood-management measures in addition to conventional levee-based approaches. Offline retention basins constructed within riparian zones have been considered a viable option, and distributing multiple basins along a river reach can increase effective storage and thereby enhance downstream flood-stage attenuation. Under unsteady hydrograph conditions, however, hydraulic interactions among offline basins may occur, implying that the flood-stage reduction estimated from individual-basin operations may not be directly transferable to integrated multi-basin operation. Therefore, this study performed one-dimensional unsteady-flow simulations using HEC-RAS for the Jangdong District, located in the upstream water-source protection area of Daecheong Dam in the Geumgang River, and comparatively evaluated flood-mitigation performance under individual and integrated operation scenarios by varying the location and length of the lateral weir. For the 100-year design flood hydrograph, the downstream peak-stage reduction obtained by a simple summation of the peak-stage reductions from individual-basin simulations was approximately 45 cm. In contrast, the unsteady-flow simulations explicitly accounting for integrated operation yielded a peak-stage reduction of approximately 41 cm, confirming that the additive estimate does not necessarily reproduce the integrated-operation outcome. Based on these findings, a stepwise unsteady-flow modeling workflow and application framework are presented to support the design of distributed offline retention basins while explicitly considering both individual and integrated operational conditions, as demonstrated for the Geumgang River case.

https://doi.org/10.12652/Ksce.2026.46.3.0205

김하영(Kim, Hayeong);장수형(Jang, Suhyeong);강신욱(Kang, Shinuk);김미은(Kim, Mieun)

The increasing occurrence of localized intense rainfall has emphasized the importance of accurate rainfall?runoff analysis for flood mitigation and dam safety. In general, the accuracy of the HEC-HMS model commonly used in hydrological practice largely depends on the appropriateness of the model parameters. In this study, the HEC-HMS model was developed for 12 water resource dam watersheds managed by K-water using topographic and hydro-meteorological data. To ensure the reliability of the constructed model, validation was conducted focusing on extreme rainfall events during the annual flood season, and the results showed satisfactory performance with NSE > 0.7. Correlation and sensitivity analysis were conducted using watershed physical characteristics and calibrated hydrological parameters. The correlation analysis indicated that the relationships between variables significantly depending on the watershed size of the dam basins. Sensitivity analysis indicated that the time of concentration(Tc) and storage coefficient(K) have significant influences on peak discharge in small to medium-sized watersheds. This study identified the relationships between hydrological parameters and watershed characteristics, establishing parameter standards for different watershed scale. These findings are expected to serve as foundational data for parameter regionalization considering future watershed characteristics.

https://doi.org/10.12652/Ksce.2026.46.3.0219

손상영(Son, Sangyoung)

Tsunamis can severely damage ports, coastal structures, vessels, and urban infrastructure through amplified wave heights and current speeds near coasts; however, conventional hazard assessments based primarily on maximum water level and inundation depth inadequately capture current-driven impacts. This study proposes the Korean-Specific Integrated Tsunami Intensity (K-ITI), a four-class (1?4) index that combines maximum tsunami height and maximum current speed, designed to map directly onto Korea's four-tier national crisis alert system. K-ITI assigns each coastal grid cell the higher of the two class levels determined independently by wave height and current speed, thereby capturing current-driven hazards that conventional water-level-only assessments may overlook. To evaluate the proposed index, limitations of existing domestic velocity observation networks for tsunami-current validation are first examined. The COMCOT model is then validated against water-level and current-speed records at Crescent City Harbor during the 2011 Tohoku tsunami, and subsequently applied to the 1983 Central East Sea, 1993 Hokkaido Nansei-Oki, and 2024 Noto Peninsula tsunamis to derive K-ITI maps along the Korean east coast. Results reveal localized areas, particularly within harbors, where current amplification elevates K-ITI classes beyond what water-level criteria alone would indicate, thereby exposing vulnerabilities overlooked by conventional assessments. These findings demonstrate that K-ITI, when mapped onto existing warning levels and embedded in an operational framework, can serve as a practical integrated tsunami hazard indicator for Korea and contribute to regional tsunami risk assessment across the Northwest Pacific.

https://doi.org/10.12652/Ksce.2026.46.3.0237

양예림(Yang, Yerim);최항석(Choi, Hangseok);권기범(Kwon, Kibeom)

Accurate estimation of rock shear strength parameters is critical for ensuring the long-term stability of geotechnical infrastructures. However, conventional approaches are often constrained by prior assumptions regarding the functional form or by limited interpretability due to the black-box nature of the models. In this study, symbolic regression was employed to derive explicit predictive equations for the shear strength parameters: cohesion (c) and the angle of internal friction (φ). A comprehensive database comprising 199 rock samples was compiled, including P-wave velocity (Vp), density (ρ), uniaxial compressive strength (UCS), and tensile strength(TS) along with the corresponding target parameters (c and φ). Correlation analysis revealed that c is strongly and positively correlated with Vp, UCS, and TS, while φ exhibits the strongest positive correlation with ρ, showing moderate correlations with the other variables. Based on these findings, univariate predictive equations for c were derived using Vp, UCS, and TS as individual input variables, whereas a multivariate predictive equation for φ was developed using Vp and ρ. The proposed equations demonstrated superior predictive performance, achieving R2 values of 0.93-0.94 for c and 0.83 for φ. Furthermore, comparative evaluations against eleven established empirical models confirmed that the proposed equations provide enhanced performance, validating their practical applicability in geotechnical design. These results indicate that symbolic regression is an effective approach for deriving interpretable and accurate predictive equations for rock shear strength parameters.

https://doi.org/10.12652/Ksce.2026.46.3.0249

장영훈(Jang, Young-Hoon);박원영(Park, Wonyoung)

This study examines directions for improving the Korean New Excellent Technology (NET) system in construction by analyzing its structural limitations and comparing it with Japan’s New Technology Information System (NETIS). To this end, the study reviewed relevant literature and institutional documents and conducted a written opinion survey of practitioners to examine the types, procedures, and post-designation management framework of the Korean NET system. The survey was used to identify recurring practical issues that are not easily captured through document analysis alone. The results indicate that the current system relies on a largely uniform procedure that does not sufficiently reflect differences in technology type and application conditions. In addition, the close linkage between technology designation and unit-price preparation constrains cost review, particularly for early-stage technologies with insufficient field data. Although post-evaluation and utilization records are in place, the tracking of productivity and economic performance remains limited. Based on these findings, this study proposes type-specific refinement of evaluation procedures, supplementary field-based verification for cost review, more flexible separation between technology designation and unit-price preparation, and stronger quantitative post-evaluation and economic monitoring. This study contributes by examining the Korean NET system as a continuous structure linking designation, evaluation, unit-price preparation, and post-management.

https://doi.org/10.12652/Ksce.2026.46.3.0259

고은별(Koh, Eunbyul);선종완(Sun, Jong-Wan);박경훈(Park, Kyung-Hoon)

With the rapid surge in bridge management data, the limitations of traditional database query methods in terms of accessibility and efficiency have become apparent. This study develops a hybrid question-answering (QA) system that integrates the natural language understanding of Large Language Models (LLMs) with the computational precision of Python, and validates its effectiveness in suppressing hallucinations. Using South Korea's bridge management data from 2012 to 2025 (474,670 records), we systematically evaluated six models, including cloud-based (GPT-4o, GPT-5, Gemini 2.0 Flash) and local open-source models (GPT-OSS 20B, Qwen3 8B, Gemma3 4B). Experiments were conducted across 20 questions categorized into six types and three levels of complexity. The results showed that GPT-4o and GPT-OSS 20B achieved the highest accuracy of 95 %. While GPT-4o exhibited superior real-time responsiveness (7.61 s), suitable for real-time services, GPT-OSS 20B demonstrated stable performance in large-scale data ranking queries without API token constraints. Error analysis revealed that intent extraction errors (46 %) were the primary cause of failure, and cloud models specifically faced token limit issues (15 %). This study proves that a hybrid architecture separating natural language understanding from numerical computation effectively suppresses hallucinations, and provides model selection criteria based on practical requirements such as cost, performance, and security for bridge management system implementation.

https://doi.org/10.12652/Ksce.2026.46.3.0269

임광균(Lim, Kwang-kyun);윤경철(Yun, Gyeong-Cheol)

This study provides a comprehensive analysis of the structural changes in railway SOC investment and the relationship between private railway investment and key market indicators in Korea over the period 2008?2024. Based on a comparative analysis using OECD data, the level of railway SOC investment in Korea was found to be approximately 0.34 % of GDP on average over the past five years, which is comparable to that of major advanced countries such as France and Italy. To identify the determinants of variations in private railway investment, regression analysis was conducted using macroeconomic indicators (GDP growth rate, Composite Business Cycle Index), financial market indicators (base interest rate, government bond yields), and construction-related indicators (construction cost index, construction business survey index). The results indicate that private railway investment is influenced more significantly by financial and cost-related factors?such as the base interest rate, long-term government bond yields, and construction costs?than by short-term economic fluctuations such as GDP growth. In particular, the base interest rate was found to have a negative effect on investment, while government bond yields and construction costs were positively associated with the scale of investment. In addition, a lagged correlation analysis was conducted to account for the time delay inherent in the implementation of private railway projects. The results show that the relationship between construction costs and investment becomes stronger with a lag of one to two years, suggesting that increases in construction costs are gradually reflected in total project costs and investment during the project execution process. In contrast, interest rates and bond yields exhibited relatively stronger correlations in the contemporaneous or short-term lag periods. These findings imply that private railway investment is shaped not by short-term economic conditions but by medium- to long-term financial environments, cost structures, and time lags in project implementation. Accordingly, future policies on private railway investment should move beyond simple expansion strategies and instead focus on mitigating interest rate risks, addressing rising construction costs, and establishing an appropriate balance between public and private investment from a structural perspective.

https://doi.org/10.12652/Ksce.2026.46.3.0283

신승숙(Shin, Seung Sook);박상덕(Park, Sang Deog);이규송(Lee, Kyu Song);김기홍(Kim, Gihong)

As large wildfire damage increases due to climate change, vegetation degradation and surface disturbance are elevating the risk of sediment disasters in watersheds. This study aims to propose erosion risk criteria based on vegetation changes and soil erosion data in wildfire areas, and to derive the Best Management Practices(BMPs) for wildfire sites using GIS-based SEMMA simulation results. Erosion response varied depending on vegetation type following a fire. Pine-dominated areas showed a high erosion risk due to slow vegetation recovery, whereas broadleaf-dominated areas exhibited low erosion response due to rapid vegetation recovery. In a case study, the Gangneung wildfire area in 2019 exhibited a significant decline in the Normalized Difference Vegetation Index (NDVI) due to excessive logging and surface disturbance, which subsequently led to sediment damage during heavy rainfall event. In contrast, the Uljin wildfire Watersheds in 2022 showed a significant increase in NDVI one year after the fire due to rapid vegetation recovery. Based on global erosion rate standards and domestically measured erosion rates, erosion rate criteria applicable to Korean mountainous areas were proposed and applied in the SEMMA simulation analysis. The estimated erosion rate in the Gangneung wildfire watershed increased by approximately 3.5 times due to artificial disturbance, whereas that in the Uljin wildfire watershed decreased by approximately 50 % as a result of natural vegetation recovery. These results indicate that erosion risk in wildfire-affected areas is significantly influenced not only by wildfire occurrence but also by vegetation recovery and surface disturbance. In this study, SEMMA-based BMPs, which consider erosion risk criteria and vegetation resilience, are expected to be actively utilized in establishing management plans for forest ecosystem restoration and the reduction of sediment disasters in wildfire areas.

https://doi.org/10.12652/Ksce.2026.46.3.0301

송영선(Song, YeongSun);윤공현(Yun, Konghyun)

True orthoimages, which correct relief displacement caused by terrain and above-ground structures such as buildings, have become important geospatial products for digital twins and smart city applications. However, the conventional workflow based on 3D stereo-plotting requires extensive manual work for digital building model generation, limiting its scalability for nationwide production. This study evaluates the applicability of an automated true orthoimage generation workflow based on Dense Image Matching (DIM) meshes and examines its technical validity through comparison with the conventional 3D stereo-plotting approach. Aerial images acquired with 80 %×80 % overlap using Falcon, DMC2, DMC3, and Osprey cameras were processed with Bentley iTwin Capture Modeler, nFrame SURE, and Trimble Inpho, and the results were compared with those from the conventional approach. The DIM mesh-based workflow produced mean planar errors of 0.148-0.359 m at rooftop checkpoints, showing positional accuracy comparable to the conventional method. It also significantly reduced processing effort, although software-specific differences were found in boundary representation and error correction. Based on these results, this study redefines quality inspection items for DIM mesh-based true orthoimages and discusses the differentiated characteristics of automated production approaches in multi-sensor environments.