전정배 Jeon Jeongbae , 김솔희 Kim Solhee , 김태곤 Kim Taegon , 윤성수 Yoon Seongsoo

DOI:10.5389/KSAE.2025.67.6.001 JKWST Vol.67(No.6) 1-12, 2025

---

This study compares one-dimensional (1D) and two-dimensional (2D) structural analysis techniques for cantilever beams using Physics-Informed Neural Networks (PINNs). Structural analysis is a fundamental technology for predicting the safety and performance of structures in various engineering fields, and recently, PINNs utilizing machine learning have emerged as a promising analytical methods. We evaluated the performance of PINNs for 1D Timoshenko beam models and 2D plane stress models. The results demonstrated that the 1D model achieved high accuracy with a maximum deflection error of 0.01% and rotation error of 0.14%, along with excellent computational efficiency, however, it had limitations in capturing stress concentration phenomena. In contrast, the 2D model showed improved accuracy in displacement prediction compared to the 1D model, with 32% improvement in the x-direction and 11% improvement in the y-direction, while accurately capturing stress concentration phenomena around the free end. These results confirmed that the 1D model is suitable for simple displacement prediction or applications where computational efficiency is critical, whereas the 2D model is appropriate for stress concentration analysis or complex stress state analysis. This study is significant in that it provides model selection criteria for effectively utilizing PINNs across various engineering fields, including agricultural engineering.

문영식 Mun Young-sik , 남원호 Nam Won-ho , 신현욱 Shin Hyeon-wook , 이광야 Lee Kwang-ya , 김선국 Kim Seon-kuk , 조영준 Jo Young-jun

DOI:10.5389/KSAE.2025.67.6.013 JKWST Vol.67(No.6) 13-25, 2025

---

Agricultural droughts have been increasing in both frequency and severity in South Korea, necessitating a comprehensive assessment of regional drought response capabilities. This study aims to evaluate agricultural drought response capacity by estimating paddy field areas using spatial analysis techniques. Unlike previous studies that relied on si, gun, and gu-level data, this research utilizes detailed farm map data to improve the accuracy of paddy field area estimation. South Korea was divided into six major river basins—Han, Geum, Nakdong, Yeongsan, Seomjin, and Jeju—for the analysis of paddy field distribution and drought vulnerability. The results show that the Geum Basin has the largest total paddy field area at 286,917 ha, followed by Nakdong (200,108 ha) and Han (158,689 ha), with Jeju having the smallest area at 15 ha. The study classified 1,995 administrative units into four categories based on drought response capacity: high (25%), moderate (35%), low (25%), and very low (15%). The most vulnerable areas were concentrated in the Nakdong (15.2%) and Geum (18.9%) basins, where frequent drought events and limited water resources increase risk. In contrast, the Yeongsan (35.8%) and Seomjin (39.4%) basins demonstrated stronger drought resilience due to well-developed irrigation infrastructure. This research provides a quantitative framework for evaluating agricultural drought vulnerability at a fine spatial scale, offering valuable insights for targeted water resource management and policy development.

이선형 Lee Sun-houng , 김락우 Kim Rack-woo , 이상신 Lee Sang-shin , 최원기 Choi Won-gi , 안수빈 Ahn Su-been , 석희웅 Seok Hee-woong , 민경원 Min Kyung-won , 허윤규 Hur Yoon-ku

DOI:10.5389/KSAE.2025.67.6.027 JKWST Vol.67(No.6) 27-39, 2025

---

Greenhouse gases such as carbon dioxide, ammonia, and methane emitted from the livestock sector accelerate global warming, leading to abnormal weather phenomena like heatwaves and typhoons, which cause significant damage. In line with the 205 0 Carbon Neutrality policy, efforts are required in the livestock sector to manage carbon emissions effectively and minimize greenhouse gas production. Carbon emissions in pig farming facilities are influenced by external environmental factors, internal environmental factors, and manure characteristics. to estimate and predict carbon emissions accurately, the collection of big data on these influencing factors is essential. However, in real farms, it is challenging to measure the factors that affect carbon emissions, and there are significant difficulties in collecting big data. Furthermore, a thorough validation process is required for the practical application of the collected data. In this study, a small-scale smart pig farming engineering experimental and demonstration facility will be established to collect and monitor data on key influencing factors for carbon emissions, such as internal environmental factors and manure characteristics. Additionally, using various ICT devices, data on carbon emissions under specific conditions, such as feed intake control, feed type changes, ventilation, and temperature adjustments, can be collected. Therefore, it is expected that the design of a carbon emission prediction algorithm will be possible through the measurement and analysis of carbon emissions.

안치용 An Chi-yong , 김설민 Kim Seolmin , 송철민 Song Chulmin , 박재성 Park Jaeseong , 박진기 Park Jinki

DOI:10.5389/KSAE.2025.67.6.041 JKWST Vol.67(No.6) 41-54, 2025

---

This study investigates the feasibility of soybean cultivar classification under open-field conditions in South Korea using unmanned aerial vehicle (UAV)-based RGB imagery and evaluates the performance of deep learning models. High-resolution RGB images were acquired at nine different growth seasons for four major soybean cultivars—Daewon, Seonpung, Pungsan, and Aram—and the temporal variations in RGB pixel value distributions were quantitatively analyzed. Among the RGB channels, the green channel showed the highest sensitivity to growth stage changes, making it a key feature for classification. Two types of deep learning models, Attention U-Net and DeepLabv3+ with ResNet-50 and ResNet-101 backbones, were applied to the cultivar classification task. The experimental results demonstrated that DeepLabv3+ with a ResNet-101 backbone achieved the highest average F1-scores and the lowest variance across all cultivars, indicating distinguished performance and stability. Classification performance was particularly high for the Daewon and Aram cultivars, while Pungsan showed relatively lower accuracy and higher variability, likely due to its less distinctive visual traits across growth stages. These findings suggest that UAV-based R GB imagery, combined with robust deep learning models, holds strong potential for reliable and scalable open-field soybean cultivar identification.

석승원 Seok Seungwon , 김솔희 Kim Solhee , 김태곤 Kim Taegon

DOI:10.5389/KSAE.2025.67.6.055 JKWST Vol.67(No.6) 55-67, 2025

---

This study employed the APSIM-Wheat model to optimize key phenological parameters and simulate the growth stages of major wheat cultivars in Korea. Growth stage data for four wheat cultivars―Jogwang, Uri, Keumgang, and Jokyung―were collected from the Barley and Wheat Crop Growth Information Service provided by the National Institute of Crop Science for the period 1974-2022. Weather data from 74 meteorological stations across Korea were used to construct climate scenarios. The phenological parameters, including photoperiod sensitivity (1.5–4.0) and vernalization sensitivity (1.5–3.5), were calibrated by minimizing the root mean square error between observed and simulated growth stages. The calibrated model was then applied to simulate growth stages under varying sowing dates and historical climate conditions, enabling an analysis of the interaction between wheat growth stages and climatic factors. The results showed that shifts in heading and maturity dates were significantly influenced by both historical climate trends and cultivar-specific traits. This study provides a foundation for developing process-based decision support systems to improve wheat management under changing environmental conditions.

홍성구 Hong Seonggu , 쿠사이알쇼타니 Qusay. F. S. Al Shortani

DOI:10.5389/KSAE.2025.67.6.069 JKWST Vol.67(No.6) 69-75, 2025

---

This study investigates the feasibility and efficiency of electrochemical treatment for condensate or wastewater generated during the operation of biomass gasification systems. The condensate, produced as synthesis gas cools after high-temperature partial oxidation of biomass, contains high concentrations of organic acids such as acetate, formic acid, phenolic compounds, and polycyclic hydrocarbons, making it difficult to treat using conventional biological processes. The raw condensate in this study exhibited a COD (Chemical Oxygen Demand) exceeding 40,000 mg/L and an acidic pH of 3.2. Overall, this study confirms that electrochemical treatment is a viable and effective method for degrading both organic acids and toxic phenolic compounds present in biomass gasification wastewater. The combination of NaCl electrolyte and BDD (Boron Doped Diamon) electrodes provided the best results. Unlike previous studies that required alkaline pre-treatment, this study demonstrated effective treatment without altering the natural acidic character of the condensate. Future work should focus on optimizing treatment conditions such as current density and electrode area versus volume ratio, and addressing microbubble related operational challenges to enable direct treatment of undiluted wastewater.

고재준 Gou Jaejun , 남강현 Nam Kanghyun , 박진석 Park Jinseok , 이혁진 Lee Hyeokjin , 장성주 Jang Seongju , 송인홍 Song Inhong

DOI:10.5389/KSAE.2025.67.6.077 JKWST Vol.67(No.6) 77-84, 2025

---

Creating an accurate 3D DTM (Digital Terrain Model) of a stream area is crucial for better understanding the stream hydrodynamic phenomena. The objective of this study was to develop an SfM-based method for generating a DTM of small streams using polarized imagery acquired by UAVs (Unmanned Aerial Vehicles). The target area was a 300 m reach of the Yangjae stream near the Seonbawi Station in Gwacheon, where strong water surface reflection posed challenges to accurate terrain modeling. Polarized images were captured at multiple angles and processed through SfM (Structure from Motion) to create point clouds. Elevation benchmarks measured by EDM (Electronic Distance Measurement) were used to assess the vertical accuracy of the resulting DTMs. Polarized filter and water refraction calibration improved the water depth accuracy of DTM, achieving MAE of 0.242 m and RMSE of 0.276 m, while challenges still remain in addressing randomly occurring underwater conditions and streamside vegetations. The outcome of the study can be used to improve the accuracy of shallow water DTM construction.

김귀훈 Kim Kwihoon , 김마가 Kim Maga , 최진용 Choi Jin-yong

DOI:10.5389/KSAE.2025.67.6.085 JKWST Vol.67(No.6) 85-96, 2025

---

The Keifer and Chu rainfall time distribution model has been used in the inundation analysis for agricultural basins. However, for applying the critical rainfall duration, the Huff series model is needed for the time distribution model. The purpose of this study was to analyze the applicability of critical rainfall duration in agricultural basins according to the rainfall time distribution model and rainfall duration. For this purpose, peak discharge, inundation level, and storage ratio were simulated with different design drainage in four study sites. For the rainfall time distribution model, this study applied the modified Huff model and the Keifer and Chu model. In the simulations, the Keifer and Chu rainfall time distribution model simulated a continuous increase in both peak flow and inundation level with increasing rainfall duration, which made the critical rainfall duration inapplicable. However, the modified Huff model was able to apply the critical rainfall duration. The modified Huff model had 2∼4 hours of critical rainfall duration based on the peak discharge and storage ratio, and had longer than 4 hours based on the inundation depth. As the critical rainfall duration with the inundation level was longer (exceeding 4 hours), less inundation damage will be simulated with shorter rainfall durations. Therefore, it would be more appropriate for agricultural basins to apply critical rainfall duration based on inundation levels rather than peak discharge or storage ratio. The results of this study are expected to serve as a basis for research on applying critical rainfall duration in agricultural watersheds.

최지은 Choi Jieun , 김시호 Kim Siho , 이정은 Lee Jeongeun , 유석철 Yu Seokcheol , 장민원 Jang Min-won , 황세운 Hwang Syewoon

DOI:10.5389/KSAE.2025.67.6.097 JKWST Vol.67(No.6) 97-110, 2025

---

Advanced research is being conducted to develop rainfall-runoff based alternative agricultural water systems for mid-mountain upland farming areas, where stable water supply remains a persistent challenge. This study aims to evaluate the practical applicability of the system by identifying irrigation-vulnerable upland farming areas across the country and estimating daily rainfall-runoff in those areas to assess the feasibility of system implementation. Irrigation-vulnerable upland farming areas cover a total of 3,534.4 ha, comprising 939 agricultural sites. Among these, Gangwon-do was found to have the largest number of sites (237). Even when applying an elevation criterion based on 100 m intervals, the rate of area reduction remained relatively low. Additionally, the average runoff in the target areas was estimated at 1.99 mm/day per unit watershed area and 15.71 mm/day per unit irrigated field area. This study provides baseline data for prioritizing the implementation of alternative agricultural water systems. Future research should assess the applicability of these systems in changing agricultural environments by estimating future water availability under climate scenarios.

박선재 Park Seonjae , 곽지혜 Kwak Jihye , 김지혜 Kim Jihye , 김시내 Kim Sinae , 박성재 Park Seongjae , 안재영 An Jaeyeong , 박수환 Park Suhwan , 이재남 Lee Jae Nam , 강문성 Kang Moon Seong

DOI:10.5389/KSAE.2025.67.6.111 JKWST Vol.67(No.6) 111-122, 2025

---

The purpose of this study was to quantitatively analyze variations in estimated inflow flood volumes caused by differences in rainfall frequency analysis methods and Clark unit hydrograph parameter estimation approaches, focusing on 26 small agricultural reservoirs with watershed areas under 10 km². It assessed the sensitivity of inflow estimates to these methods and identified trends according to watershed characteristics. Three methodological combinations were evaluated: (1) at-site frequency analysis with the Sabol and continuous Kraven formulas, (2) regional frequency analysis with the Sabol and continuous Kraven formulas, and (3) regional frequency analysis with the Seokyeong University formula. Regional frequency analysis generally yielded lower inflow volumes than at-site frequency analysis, with an average decrease of 13.2%. The largest decreases occurred in watersheds larger than 3 km², slopes over 35%, and stream lengths exceeding 3 km. In comparing Clark unit hydrograph parameter estimation methods, the Seokyeong University formula consistently resulted in lower inflow volumes, averaging an 18.6% decrease. The largest decreases occurred in watersheds of 1.5–3 km², slopes of 27–35%, and stream lengths exceeding 3 km. These differences were attributed to limited rainfall gauge coverage and to the larger times of concentration and storage coefficients produced by the Seokyeong University formula. The findings indicated that method selection significantly affected inflow estimation. This study provided baseline data for the design of small agricultural reservoir facilities and flood forecasting, and it highlighted the need for further research on regional distribution, high-resolution rainfall data, and additional hydrological factors.