Accurately estimating and predicting solar photovoltaic (PV) power generation is essential for ensuring a stable renewable energy supply. This study proposes a machine learning-based approach to estimate and predict solar PV power generation. A linear model and three machine learning models were trained using hourly datasets from the Yeongam F1 solar power plant from 2019 to 2022. To ensure model stability, a year-based 4-fold cross-validation was performed. Of the tested models, XGBoost model performed the best, achieving a normalized mean absolute error (nMAE) of 4.16% with a standard deviation of 0.07%. Furthermore, in a practical forecasting scenario where models were trained on data from 2019 to 2021, XGBoost model successfully predicted next-day solar PV power generation for 2022 using LDAPS forecasts initialized every 2100 KST, achieving a nMAE of 7.89%. These results demonstrate that combining short-term numerical weather forecasts with machine learning algorithms can enable reliable prediction of solar PV power generation one day in advance.
This study develops and evaluates an optimized subdivision of severe weather alert zones for Busan and Ulsan, using meteorological, high-impact weather, socioeconomic, and land surface data. Based on these variables, a multivariate cluster analysis was conducted at the census output area level. The optimal number of subdivision zones was determined to be three for Busan and two for Ulsan by several statistical methods. From multiple zoning combinations, three candidate plans were selected, and their operational effectiveness was evaluated based on recent occurrences of heatwaves and heavy rainfall. Among them, the combination of high-impact weather and socioeconomic characteristics showed the highest efficiency. This approach is expected to reduce unnecessary alerts, improve the spatial focus of warnings, and it also aligns well with administrative boundaries. The results demonstrate that a single large severe weather alert zone cannot adequately capture local risk patterns. This work provides a scientific basis that supports the redesign of severe weather alert systems and offers a practical framework that could be adapted for other metropolitan areas. In the context of increasing localized extreme weather events, these findings may help optimize resource allocation, and strengthen public responsiveness.
This study evaluates the subseasonal-to-seasonal (S2S) prediction skill of Madden- Julian Oscillation (MJO) and its East Asian teleconnections using the Korea Meteorological Administration's Global Seasonal Forecasting System version 6 (GloSea6) hindcasts. The MJO prediction skill of the GloSea6 model, in terms of the bivariate correlation of the Real-time Multivariate MJO (RMM) index, is about three weeks in winter, extending to four weeks for MJO phases 2-3. However, the prediction skill of MJO teleconnections to East Asia is limited to about two weeks. This discrepancy arises from prediction errors in MJO spatial structure and magnitude, which drastically increase after forecast week two. In particular, the centers of divergence and convergence become latitudinally biased with weaker magnitudes. These errors are attenuated by latitudinal averaging when computing the RMM index. The bivariate correlation metric also underrepresents magnitude errors. Together, they lead to an overestimation of MJO prediction skills. These results highlight the importance of considering the detailed structure and magnitude of MJO circulation and convection when assessing the prediction skills of MJO and its teleconnections.
Within the quality control (QC) procedures, this study reassesses the threshold of the persistence test from the perspective of sub-daily timescale analysis, using 1-min mean windspeed time series from seven Automatic Weather System (AWS) sites in the Ulsan region in 2024. First, we performed preprocessing steps to prevent distortions in spectral analysis arising from temporal discontinuities or nonphysical signals. Next, to quantify the impact of flat-line segments on the spectrum, we randomly selected 21 out of 286 days with no missing values and flat lines shorter than 5 min, and inserted and moved a 0 m s(-1) segment of varying length (6, 10, 15, 20, 30, 120, 240 min) within each time series. As a result, the spectral relative error increased with segment length; in particular, the increase in error was markedly larger for 10 similar to 15 min than for 6 similar to 10 min. In the 240-min case, the largest error appeared at a timescale of about 8 h, and consistency with the spectrum of the original time series deteriorated substantially across the entire frequency range. In addition, the longer the flat-line segment, the more the daily mean wind speed was underestimated and the more the variance tended to be overestimated. These results suggest that the 240-min threshold used in the KMA persistence test imposes limitations on analyses at sub-daily timescales. Considering the need to minimize spectral and statistical distortions, this study proposes 10 min as a realistic and physically sound threshold for the persistence test applied to 1-min mean wind-speed data.
This study examined net biome exchange (NBE) data from various institutions to evaluate the reliability and characteristics of NBE estimates on a global, continental, and national scale from 2015 to 2020. All three NBE datasets-CT2022, CMS-Flux, and CAMS- consistently showed that the global terrestrial biosphere acts as a net carbon sink. The mid-to high-latitude regions of the Northern Hemisphere, which are rich in vegetation, mainly absorb CO2. However, the datasets show substantial discrepancies in the regional magnitude and variability of NBE. Large discrepancies in NBE exist among the datasets in the low-latitude Northern and Southern Hemispheres, where observational data are scarce. At the national scale, China, Canada, the United States, and Russia, while significant anthropogenic CO2 emitters, are consistently identified as carbon sinks by the terrestrial biosphere across all datasets, indicating relatively low uncertainty in NBE estimates. However, countries such as India and Brazil show large inconsistencies in NBE estimates, highlighting significant uncertainties among the data-sets. South Korea, Japan, and the United Kingdom exhibit near-zero NBE values across all data-sets. These results underscore the necessity of multi-model comparisons, cross-validation, and expanded observational networks in underrepresented regions, as well as the importance of integrating biosphere carbon flux considerations into national carbon neutrality strategies.
This study investigates the physical mechanisms of a downslope windstorm that occurred in the Gangneung region on April 11, 2023, using numerical simulation and the shallow water theory. In shallow water theory, the mechanism of a downslope windstorm involves the flow transition from subcritical to supercritical due to changes in the fluid thickness and velocity, which is identified by the Froude number (Fr = U/root gH). While previous studies calculated the Fr using a fixed characteristic depth, this study applied the spatially varying flow thickness for the Fr calculation, assuming that the lower tropospheric inversion layer with the maximum Brunt-Vaisala frequency could be considered as the free surface in shallow water theory. Numerical simulation conducted using the Weather Research and Forecasting (WRF) model revealed that the new Fr calculated with the varying flow thickness aligns better with shallow water theory and effectively explains the physical mechanisms of the downslope windstorm. On the upwind side, the westerly flow is well identified as subcritical until it reaches the crest, several kilometers west of Daegwallyeong, where its Fr transitions to supercritical by decreased flow thickness and increased wind speed. The flow further accelerated while descending the lee side under supercritical conditions. The subcritical-to-supercritical transition made the westerly flow keep accelerating, causing the downslope windstorm at Gangneung Airport. The new spatially varying Fr revealed clear distinctions between the downslope windstorm and turbulent regions closely matching observations at Gangneung Airport.
The Arctic ozone depletion in 2020 spring drew significant scientific attention due to its rare occurrence in the Northern Hemisphere. Although relatively small in amplitude compared to the Antarctic ozone hole, the 2020 Arctic ozone depletion is the most extensive and markedly prolonged event in the Northern Hemisphere. This study investigates the 2020 Arctic ozone depletion within the context of the background conditions in the stratosphere by comparing it to the 2019 conditions when the stratospheric ozone concentration was high. In early spring 2020, the enhanced meridional temperature gradient strengthened the polar vortex and maintained a cold Arctic condition in the stratosphere, which is markedly different from the 2019 spring and climatological conditions. These background fields create favorable conditions for forming polar stratospheric clouds, which could efficiently cause chemical ozone destruction. Wave activity analysis reveals that the vertical Rossby wave propagation from the high-latitude troposphere and associated stratospheric drag were noticeably weaker than normal in the preceding winter. This contributed to the strong polar jet and weaker Brewer-Dobson circulation in the stratosphere. The strong positive Arctic Oscillation in early 2020 indicates a weaker Rossby wave drag in the high-latitude regions, and 3D stationary wave activity flux (Plum flux) manifests a clear decrease in vertical wave activity over the high-latitude Atlantic. However, its physical causes remain unclear.
Methane, despite having a relatively shorter atmospheric lifespan than carbon dioxide, is a potent greenhouse gas with a global warming potential over 80 times higher on a 20-year basis. Therefore, reducing methane emissions can significantly contribute to effective climate change mitigation. In the energy sector, methane emissions primarily originate from leaks during fossil fuel production, processing, transport, and usage. This study highlights the need for and the potential benefits of expanding Leak Detection and Repair (LDAR) programs based on strengthened Measurement, Monitoring, Reporting, and Verification (MMRV) systems. If LDAR is expanded to power generation facilities in Korea, it's possible to prevent methane leakage equivalent to about 2.57% of the nation's annual power generation. This could result in a saving of over 620,000 tons of LNG, achieving a cost reduction of approximately 530 billion KRW. This is further estimated to save social value costs equivalent to 5 trillion KRW and is expected to reduce greenhouse gases by about 80,000 tons of CO2eq from the fuel combustion process. Mitigating methane emissions not only contributes to climate change mitigation but also reduces tropospheric ozone formation, thereby improving air quality and public health. Furthermore, preventing unintended energy loss yields mid-and long-term economic savings. Given that South Korea is one of the largest importers of natural gas, enhancing methane management in industrial complexes and metropolitan areas is essential for achieving national carbon neutrality goals. Moreover, Expanding LDAR programs can practically contribute to the country's Nationally Determined Contributions (NDCs).
Urban heat islands (UHIs) can intensify during high-temperature weather events such as tropical nights, leading to even greater thermal stress in cities. This study investigates the interactions between UHIs and tropical nights in Busan, South Korea, using 52 years (1973 similar to 2024) of observational data. In Busan, the frequency and intensity of tropical nights are 48.8% and 0.19 degrees C higher than in rural areas, respectively, and these differences have become more pronounced as urbanization has progressed. On average, the UHI intensity is higher on tropical nights than on non-tropical nights, as warmer, drier, and less cloudy nighttime conditions on Busan's tropical nights lead to a positive interaction between UHIs and tropical nights. However, the UHI-tropical night interaction shows considerable variability, being either strongly positive or strongly negative. This is caused by contrasting characteristics of tropical nights in Busan, which result from their different mechanisms and synoptic patterns. Specifically, tropical nights that exhibit strong positive interactions with UHI are relatively calm, dry, and hot with clear skies, driven by strong insolation and subsidence under the dominance of a persistent anticyclone. In contrast, tropical nights that exhibit strong negative interactions with UHI are much more windy, cloudy, and moist. They are driven by reduced nighttime radiative cooling and warm advection when warm moist air is transported by southwesterlies at the periphery of the western North Pacific subtropical high.
Climate change has a pronounced impact on plant phenology, which is closely linked to ecosystem productivity, carbon cycling, and biodiversity. Analyzing phenological shifts using long-term monitoring data is therefore essential for understanding ecosystem responses to climate change. In this study, we quantitatively assessed four key phenophases (budburst, flowering onset, leaf unfolding onset, and 90 similar to 100% fall foliage) based on observations of 20 deciduous broadleaf species collected at 10 arboreta across South Korea from 2009 to 2024. Phenological records were further combined with climatic variables from the Automated Synoptic Observing System (ASOS) to evaluate correlations between phenological events and climate factors. The results showed that budburst (-0.94 d yr(-1)), flowering (-0.83 d yr(-1)), and leaf unfolding (-0.79 d yr(-1)) advanced consistently, while peak (90 similar to 100%) fall foliage was delayed (+0.33 d yr(-1)), leading to an extension of the average growing season by more than 17 days. A distinct seasonal transition point was also identified between fruit set and fruit maturity, where DOY (Day of year) values increased sharply. Regional analysis indicated consistent advancement of spring events across most regions, whereas fall foliage tended to be delayed. Correlation analysis revealed that spring phenophases advanced in response to winter-early spring air and surface temperatures, while fall foliage showed strong positive correlations with late-summer air temperature, surface temperature, and dew point temperature. Stepwise multiple-regression models further showed that leaf unfolding and flowering showed very high explanatory power from climatic predictors (Adjusted R-2: 0.96 and 0.95, respectively), confirming them as the phases most sensitive to inter-annual climatic variability. This study integrated nationwide long-term phenological observations updated through 2024 to present species-and region-level recent changes, and applied rolling 2to 3-month analysis windows to systematically identify, for each phenophase, the sensitive periods and the direction of climatic influences. Moving beyond a simple confirmation of growing-season extension, we quantified stage-specific meteorological influence and sensitivity windows, thereby enhancing the comparability and reproducibility of the results. Our findings can serve as a scientific basis for predicting ecosystem responses to climate change and informing the development of climate crisis adaptation strategies.
This study classified tropical cyclones (TCs) passing near Daegu as compound, rain-dominant, wind-dominant, or weak based on observed daily precipitation and maximum wind speed at the Daegu Automated Surface Observing System (ASOS). TCs of the compound type had the closest tracks to the Daegu ASOS and exhibited the highest maximum sustained wind speed (MSWS). These TCs maintained their intensity due to weak vertical wind shear and strong vertical velocity (VV) around the Korean Peninsula during their influence period. Rain-dominant TCs exhibited greater intensity in terms of mean sea level pressure (MSLP) while passing east of the Daegu ASOS. During the period influenced by rain-dominant TCs, high VV was observed because the Korean Peninsula was located south of the jet stream entrance. Wind-dominant TCs exhibited higher MSWS intensity while passing west of the Daegu ASOS, placing Daegu within the dangerous semicircle of TCs. Weak TCs had the farthest tracks from the Daegu ASOS and exhibited the lowest intensity in terms of MSWS and MSLP. The multiple linear regression model developed in this study performed poorly in predicting rainfall and maximum wind speed at Daegu, as well as determining the type of influence although it performed moderately well in predicting whether Daegu is influenced by a TC. This is because simple linear regression cannot capture nonlinear relationships, and the meteorological variables are at a synoptic scale much larger than that of Daegu. Addressing these limitations in future work could improve TC impact warnings for Daegu.
The general circulation in mid-latitudes is characterized by planetary geostrophic motion, where heat and momentum fluxes from synoptic eddies play a crucial role. The poleward heat flux, produced by baroclinic eddy growth, reduces the meridional temperature gradient and is parameterized as eddy diffusion. Similarly, eddy momentum flux, driven by barotropic wave breaking, is assumed proportional to the horizontal shear of the zonal mean zonal wind, intensifying upper-level westerlies. Incorporating turbulent eddy parameterizations into the planetary-scale heat equation, a balance is achieved between the two fluxes, resulting in eddy-driven circulations in mid-latitudes akin to the Farrell cell. The meridional domain is governed by a fourth-order characteristic equation, exhibiting two primary features related to anomalous potential temperature. The first feature is a linear decline in anomalous potential temperature, inducing westerly winds in midlatitudes. The second feature, represented by a trigonometric function, corresponds to jet streams generated by eddy momentum flux. The meridional structure of the circulation is influenced by three factors. The first factor is a structural number D/SM, where D represents eddy diffusivity, S signifies dry static stability, and M is the proportionality constant for momentum flux, summarizing the life cycle of synoptic eddies. The second factor relates to planetary size relative to the external Rossby deformation radius. Finally, a vertical structure of the atmosphere, associated with eigenvalues of the vertical mode in the main heat equation, constitutes the third factor. The combination of these three factors within the characteristic equation determines the location and number of eddy-driven circulations in mid-latitudes.
Understanding the observational characteristics of precipitation detectors and selecting appropriate sensors for specific purposes are essential for securing meteorological data and ensuring reliable observations. This study compares the detection characteristics and response times of four types of precipitation detectors through field experiments. The objective is to provide foundational data for establishing sensor selection criteria suitable for meteorological observation. The tested sensors employ capacitive, impedance, radar, and optical detection methods. The impedance and domestic capacitive sensors showed stable detection performance, albeit with occasional missed rainfall events. The foreign capacitive sensor responded later and ceased detection earlier than other sensors, yet demonstrated high accuracy. The optical sensor rapidly detected rainfall onset but continued to report rainfall even after cessation and failed to detect snowfall. The radar-based sensor was effective for snow detection due to its detection principle but occasionally produced false positives under non-precipitating conditions due to atmospheric particles. Based on statistical and characteristic analyses, it is concluded that impedance, capacitive, and radar sensors are generally suitable for meteorological applications, with radar sensors being optimal for snow detection.
This study aims to measure carbon dioxide (CO2) in Busan to enhance understanding of the urban atmospheric environment. Two sites with distinct environmental settings-Second College of Education Building (SEB; natural surroundings) and Mechanical Engineering Building (MEB; urban-adjacent)-were established at Pusan National University, and CO(2 )measurements were carried out from March 2024 to February 2025 to analyze diurnal and seasonal variations. The SEB site, positioned lower than MEB, was more directly affected by local emission, resulting in an average summer CO2 concentration of 458.2 ppm, being 12.6 ppm higher than MEB. It is noted that these differences reflect the combined effects of topography, instrument height, wind direction, and the planetary boundary layer height (PBLH). During summer, CO2 concentrations at the SEB were higher than at MEB, with both sites showing increased levels during the night. SEB maintained elevated concentrations continuously from 6:00 p.m. to 8:00 a.m. the following day. Meteorological analysis revealed that humid air advected from the ocean in summer lowered PBLH and stabilized the near-surface atmosphere, creating favorable conditions for CO2 accumulation. Furthermore, southwesterly wind-which accounted for about 30% of summer wind in two sites- suggests that nocturnal CO2 emissions from vegetation respiration in nearby forests contribute to the observed summer CO2 concentration increases. Overall, this study presents a comparative, observation-based analysis of CO2 variability across two urban sites with contrasting environmental conditions, offering insights into the interplay between local emissions and meteorological factors. The results provide a basis for developing localized air quality management strategies.
Recent advances in artificial intelligence (AI) technologies are broadly integrated across all stages of climate prediction systems, driving significant innovations. Deep learning-based weather prediction models are rapidly progressing worldwide, demonstrating performance surpassing that of traditional numerical model-based integrated forecasting systems. In ocean modeling, deep learning simultaneously learns local detailed structures and global ocean configurations, precisely simulating complex ocean dynamics from mesoscale eddies to extensive current patterns, thereby substantially enhancing prediction ability. For land surface modeling, deep learning adopts hybrid forms combined with physics-based approaches to more accurately reproduce intricate land responses, with the integration of vegetation water stress modules enabling realistic depictions of land-atmosphere interactions. In data assimilation, AI contributions are prominent, where deep learning methods utilizing automatic differentiation, diffusion models, and image restoration techniques offer superior computational efficiency and expressiveness compared to conventional data assimilation approaches. Recent technologies like Variational AutoEncoders and Score-based Diffusion effectively incorporate high-dimensional nonlinear characteristics, continuously improving ocean and atmospheric initial field performance. Various deep learning methods exhibit potential for high-resolution enhancement of climate prediction outputs, while deep learning-based generative models are primarily employed for post-processing techniques to correct model biases. Beyond technical advances, AI-based technologies provide a strategic pathway toward operational implementation and climate services, enhancing both forecast accuracy and computational efficiency. Collectively, these devel opments point toward an integrated, next-generation climate prediction framework that bridges physical modeling, data-driven methods, and practical applications.
A wind shear detection system using WISSDOM (WInd Synthesis System using DOppler Measurements) a 1 km resolution analysis field incorporating radar radial velocities through 3D-VAR was developed and evaluated to detect wind shear along aircraft glide paths at domestic airports. This system addresses limitations of existing LLWAS, which detects low-level wind shear only within a narrow area near runways at altitudes below 30 m. Vertical and horizontal wind shear detection techniques were applied in parallel, with performance verified using LLWAS alarm data and IATA aircraft observations from Incheon, Jeju, and Yangyang airports. Wind shear values from the system were generally lower than the ICAO standard (5 kt/100 ft). Optimal thresholds maximizing TSS were derived: vertical 1.0 similar to 1.5 kt/100 ft and horizontal 1.6 similar to 2.4 kt/km. Performance was highest at Incheon (horizontal wind shear PODY 0.82, AUC 0.71), followed by Jeju, and lower at Yangyang. Case analyses using AMDAR data showed spatial similarity between observed and detected wind shear events, supporting the system's operational capability. Despite resolution limits for sub-kilometer phenomena, the system effectively detects mesoscale wind shear and offers a practical means to enhance aviation safety and operational efficiency.
This study quantitatively analyzed the regional characteristics of heavy precipitation and its associated damages in South Korea from May to October during the period 2016-2022. Additionally, spatial clustering based on the K-means method was performed using rainfall frequency, damage frequency, and topographic elevation. The analysis of cumulative rainfall and damage frequency showed that, in most regions, rainfall amounts associated with damage were comparable to or slightly higher than current heavy rain warning thresholds, generally within an operationally acceptable range. In the inland central regions (e.g., Chungcheongbuk-do, Chungcheongnam-do, and Gyeonggi-do), the proportion of damage relative to heavy rainfall occurrence was relatively high under intense rainfall conditions, whereas Jeju Island exhibited a high frequency of heavy rainfall but low damage occurrence. The clustering classified the country into four clusters, with consistent cluster configurations maintained under both the 3-hour and 12-hour cumulative rainfall criteria. Notably, the cluster encompassing the Seoul Metropolitan Area and southeastern coastal regions exhibited higher frequency of damage despite lower frequency of heavy rainfall, suggesting significant influences from socio-geographical factors. In contrast, the mountainous areas of Jeju, characterized by highest rainfall frequency and elevation, reported almost no damage, indicating that social exposure and physical vulnerability are critical determinants of damage rather than rainfall amount alone. These findings are expected to provide as foundational data for improving heavy rainfall warning systems and developing tailored response strategies.
Mid-latitude cyclones are vortices ranging from meso-scale to synoptic-scale, characterized by rising and converging air, and are commonly associated with low-pressure systems and fronts. These systems can cause significant damage due to strong winds and heavy precipitation. While weather forecasters typically monitor the mid-latitude cyclones by analyzing precipitation intensity and movement from radar echoes, this approach becomes challenging when precipitation areas shift spatially or vary regionally. To address this issue, we propose a novel technique for tracking mid-latitude cyclones by detecting their cyclonic circulation centroid from radar wind fields. Our method uses high-resolution three-dimensional wind field data derived from radar radial velocity observations provided by the Korea Meteorological Administration. These data are processed using the "WInd Synthesis System using Doppler Measurements (WISSDOM)". By analyzing wind direction, speed, and vorticity, we objectively identify the cyclonic rotation center in real-time. The proposed technique consists of three steps: 1) identifying potential cyclonic center grid points based on wind speed and vorticity, 2) analyzing cyclonic vorticity area using wind direction pattern, and 3) determining the final cyclonic center points. We applied this method to 50 extratropical cyclone events between 2023 and 2024, successfully detecting cyclonic center in 46 cases. Future improvements will focus on optimizing and refining the algorithm. This research enhances our understanding and predictive capabilities for mid-latitude cyclones and associated severe weather events. Additionally, we aim to extend this technique to detect the centroids of tropical cyclones and mesocyclones, contributing to early monitoring of torrential rainfall.
This study aims to harmonize the long-term data from the Total Lightning Detection System (TLDS; 2002--2015) and the Lightning Detection Network (LINET; 2015--2023), producing the first consistent 22-year lightning climatology over the Korean Peninsula. Spatially, flash density maxima are found over the Yellow Sea (--125.5oE) and the western inland, associated with low-level convergence zones in upslope areas of the Taebaek and Sobaek mountain ranges. Interannually, a statistically significant long-term decreasing trend in lightning activity is newly revealed over the Yellow Sea, South Sea, and adjacent inland regions. Seasonally, lightning follows a unimodal distribution with a distinct August peak (35.8%), which is surprisingly different from the bimodal pattern of seasonal heavy rainfall frequency. On the other hand, the lightning data exhibits a bimodal structure in diurnal distribution. 1) A late afternoon maximum over inland regions (1600--1700 LST), which is likely associated with thermally driven convection and topographic effect. 2) An early morning peak (0500--0600 LST) over the Yellow and South Seas, which is likely influenced by the interaction of nocturnal low-level jets with land breezes. A Hovm & ouml;ller analysis highlights a prominent dawn peak in July over the eastern Yellow Sea and western inland areas, linked to an elongated low-level jet from inland China and enhanced integrated vapor transport in nighttime. These conditions favor the development of the elevated nocturnal convection propagating to the inland with large amount of lightning. This study eventually provides new insights into the spatiotemporal variability of lightning climatology in Korea.
While many previous studies have been conducted on the downslope windstorms (DWs) in the lee of Taebaek Mountains in Korea, no such research has been focused on the lee of Sobaek Mountains at which the DWs often caused wild fires and wind gusts with flight delays. We selected 1,950 cases of DWs in the lee of Sobaek Mountains and classified the mean sea-level pressure patterns into eight clusters using Self-Organizing Map. We found that there are two distinct types: low-passing (Type 1 in Spring, clusters 1 and 5) and west-high and east-low (Type 2 in Winter, clusters 4 and 8) patterns. Others are found to be the transitional patterns between these two types. Using reanalysis and upstream sounding data, we identified that the Type 1 had a more pronounced Foehn effect on the lee of Sobaek Mountains. And, the hydraulic jump was found to be the main mechanism for the DWs in Type 2 with an inversion layer on the windward side especially in clusters 6, 7 and 8 in the case of stratified flow. The partial reflection mechanism was dominant in most types with some differences depending on horizontal wavelength of the mountain wave. Critical-level reflection mechanism was also important in Type 1. DWs in the lee of Taebaek Mountains happened more under the south-high and north-low pattern, which was not the case in the lee of Sobaek Mountains. Type 2 with one case showed a similar mechanism to the Bora wind with hydraulic jump and no critical-level reflection.
