• 대기
• /
• 제34권2호
• /
• pp.177-185
• /
• 2024

As increasing global interest in renewable energy due to the ongoing climate crisis, there is a growing need for efficient technologies to manage such resources. This study focuses on the predictive skill of daily solar power generation using weather observation and forecast data. Meteorological data from the Korea Meteorological Administration and solar power generation data from the Korea Power Exchange were utilized for the period from January 2017 to May 2023, considering both inland (Daejeon) and coastal (Busan) regions. Temperature, wind speed, relative humidity, and precipitation were selected as relevant meteorological variables for solar power prediction. All data was preprocessed by removing their systematic components to use only their residuals and the residual of solar data were further processed with weighted adjustments for homoscedasticity. Four models, MLR (Multiple Linear Regression), RF (Random Forest), DNN (Deep Neural Network), and RNN (Recurrent Neural Network), were employed for solar power prediction and their performances were evaluated based on predicted values utilizing observed meteorological data (used as a reference), 1-day-ahead forecast data (referred to as fore1), and 2-day-ahead forecast data (fore2). DNN-based prediction model exhibits superior performance in both regions, with RNN performing the least effectively. However, MLR and RF demonstrate competitive performance comparable to DNN. The disparities in the performance of the four different models are less pronounced than anticipated, underscoring the pivotal role of fitting models using residuals. This emphasizes that the utilized preprocessing approach, specifically leveraging residuals, is poised to play a crucial role in the future of solar power generation forecasting.

• International Journal of Computer Science & Network Security
• /
• 제24권9호
• /
• pp.63-76
• /
• 2024

Weather forecasting has become a very popular topic nowadays among researchers because of its various effects on global lives. It is a technique to predict the future, what is going to happen in the atmosphere by analyzing various available datasets such as rain, snow, cloud cover, temperature, moisture in the air, and wind speed with the help of our gained scientific knowledge i.e., several approaches and set of rules or we can say them as algorithms that are being used to analyze and predict the weather. Weather analysis and prediction are required to prevent nature from natural losses before it happens by using a Deep Learning Approach. This analysis and prediction are the most challenging task because of having multidimensional and nonlinear data. Several Deep Learning Approaches are available: Numerical Weather Prediction (NWP), needs a highly calculative mathematical equation to gain the present condition of the weather. Quantitative precipitation nowcasting (QPN), is also used for weather prediction. In this article, we have implemented and analyzed the various distinct techniques that are being used in data mining for weather prediction.

• 환경영향평가
• /
• 제21권5호
• /
• pp.665-681
• /
• 2012

This study investigates the influence of anthropogenic heat (AH) release on urban boundary layer in the Gyeong-In region using the Weather Research and Forecasting model that includes the Seoul National University Urban Canopy Model (SNUUCM). The gridded AH emission data, which is estimated in the Gyeong-In region in 2002 based on the energy consumption statistics data, are implemented into the SNUUCM. The simulated air temperature and wind speed show good agreement with the observed ones particularly in terms of phase for 11 urban sites, but they are overestimated in the nighttime. It is found that the influence of AH release on air temperature is larger in the nighttime than in the daytime even though the AH intensity is larger in the daytime. As compared with the results with AH release and without AH release, the contribution of AH release on urban heat island intensity is large in the nighttime and in the morning. As the AH intensity increases, the water vapor mixing ratio decreases in the daytime but increases in the nighttime. The atmospheric boundary layer height increases greatly in the morning (0800 - 1100 LST) and midnight (0000 LST). These results indicate that AH release can have an impact on weather and air quality in urban areas.

• 대기
• /
• 제22권1호
• /
• pp.47-55
• /
• 2012

The characteristics of flow and pollutant dispersion for fire scenarios in an urban area are numerically investigated. A computational fluid dynamics (CFD) model coupled to a mesoscale weather research and forecasting (WRF) model is used in this study. In order to more accurately represent the effect of topography and buildings, the geographic information system (GIS) data is used as an input data of the CFD model. Considering prevailing wind, firing time, and firing points, four fire scenarios are setup in April 2008 when fire events occurred most frequently in recent five years. It is shown that the building configuration mainly determines wind speed and direction in the urban area. The pollutant dispersion patterns are different for each fire scenario, because of the influence of the detailed flow. The pollutant concentration is high in the horse-shoe vortex and recirculation zones (caused by buildings) close to the fire point. It thus means that the potential damage areas are different for each fire scenario due to the different flow and dispersion patterns. These results suggest that the accurate understanding of the urban flow is important to assess the effect of the pollutant dispersion caused by fire in an urban area. The present study also demonstrates that CFD model can be useful for the assessment of urban environment.

• 한국항행학회논문지
• /
• 제22권5호
• /
• pp.478-484
• /
• 2018

본 연구는 기온, 강수량, 풍향, 풍속, 습도, 운량, 일조, 일사 등 시간별 기상 데이터를 이용하여, 일사 및 일조 그리고 태양광 발전예측을 하였다. 지도학습에서 입출력패턴은 예측에서 가장 중요한 요소이지만 인간이 직접 결정해야하기 때문에, 반복적인 실험에 의해 결정해야 한다. 본 연구는 일사 및 일조 예측을 위하여 4가지 모델의 입출력 패턴을 제안하였다. 또한, 예측된 일조 및 일사 데이터와 전라남도 영암 태양광 발전소의 발전량 데이터를 사용하여 태양광 발전량을 예측하였다. 실험결과 일조 및 일사 예측에서 모델 4가 가장 예측결과가 우수했으며, 모델 1에 비해 일조의 RMSE는 1.5배 정도 그리고 일사의 RMSE는 3배 정도 오차가 줄었다. 태양광 발전예측 실험결과 일조 및 일사와 마찬가지로 모델 4가 가장 예측결과가 좋았으며, 모델 1 보다 RMSE가 2.7배 정도 오차가 줄었다.

• 한국환경과학회지
• /
• 제26권5호
• /
• pp.573-584
• /
• 2017

In order to simulate a typhoon precisely, the satellite observation data has been assimilated using WRF (Weather Research and Forecasting model) three-Dimensional Variational (3DVAR) data assimilation system. The observation data used in 3DVAR was GPS Radio Occultation (GPS-RO) data which is loaded on Low-Earth Orbit (LEO) satellite. The refractivity of Earth is deduced by temperature, pressure, and water vapor. GPS-RO data can be obtained with this refractivity when the satellite passes the limb position with respect to its original orbit. In this paper, two typhoon cases were simulated to examine the characteristics of data assimilation. One had been occurred in the Western Pacific from 16 to 25 October, 2015, and the other had affected Korean Peninsula from 22 to 29 August, 2012. In the simulation results, the typhoon track between background (BGR) and assimilation (3DV) run were significantly different when the track appeared to be rapidly change. The surface wind speed showed large difference for the long forecasting time because the GPS-RO data contained much information in the upper level, and it took a time to impact on the surface wind. Along with the modified typhoon track, the differences in the horizontal distribution of accumulated rain rate was remarkable with the range of -600~500 mm. During 7 days, we estimated the characteristics between daily assimilated simulation (3DV) and initial time assimilation (3DV_7). Because 3DV_7 demonstrated the accurate track of typhoon and its meteorological variables, the differences in two experiments have found to be insignificant. Using observed rain rate data at 79 surface observatories, the statistical analysis has been carried on for the evaluation of quantitative improvement. Although all experiments showed underestimated rain amount because of low model resolution (27 km), the reduced Mean Bias and Root-Mean-Square Error were found to be 2.92 mm and 4.53 mm, respectively.

• 한국해양환경ㆍ에너지학회지
• /
• 제8권1호
• /
• pp.46-52
• /
• 2005

대양을 운항하는 선박의 최적 항로계획 수립에 중요한 해양환경정보를 실시간으로 예보하는 시스템(MEIS)을 개발하였다. 예보정보는 위성관측 대양환경 자료를 기반으로 유럽중기기후예보센터가 처리한 실시간 자료를 바탕으로 하며, 장기 관측자료 데이터베이스에 근거한 통계적 정보와 함께 제공된다. MEIS시스템은 육상 기지국에 설치되어 해양환경정보를 취득하고 처리하는 육상자료처리시스템(MEIS-Center)과 선박에 탑재되어 가공된 해양환경정보를 화상으로 구현하고 최적항로 선정을 돕는 선박탑재화상구현시스템(MEIS-Ship)으로 구성되며, 운항중인 선박과 육상기지국간의 정보 송수신을 위한 위성통신 시스템을 활용한다. 해양환경 요소는 바람, 파랑, 기압, 폭풍을 포함하며, 바람은 풍향과 풍속 정보를 제공하고, 파랑은 너울과 풍파로 구분하여 파고, 파향, 파주기 정보를 제공할 수 있다. 실시간 정보는 0.5°해상도로 5시간 간격의 10일 예보치가 매일 제공되며, 통계적 정보는 1.5° 해상도의 15년 관측자료를 이용하여 월평균 및 재현주기별 최대값이 산정된다. MEIS-Ship은 항로 시뮬레이션 기능을 제공하며, 설정된 항로에 대해 예보 및 통계적 해양환경정보를 그림 또는 표의 형태로 제공한다. MEIS는 예정 항로상의 정확한 실시간 해양환경 예보를 제공하므로 선박운항자가 항로의 위험도와 운항경제성을 고려하여 최적 항로를 선정하는 것이 가능하다.

• 한국농림기상학회지
• /
• 제11권4호
• /
• pp.192-205
• /
• 2009

2009년 2월 9일 화왕산에서는 대보름 행사인 '억새 태우기'가 많은 사람들이 지켜보는 가운데 시작되었지만 예상하지 못한 강풍으로 산불로 확대되어 많은 인명피해가 발생하였다. 본 연구에서는 3차원 계산 유체역학 모형인 CFD_NIMR_SNU 모형을 이용하여 복잡한 산악지역에서 국지적 가열에 따른 바람장을 모사함으로써 이날 발생한 산악 화재의 특성을 분석하였다. 화재가 발생한 지역의 지표 온도는 가열이 없을 때, $300^{\circ}C$ 및 $600^{\circ}C$ 일 때의 3가지 가열 강도조건을 주어 모사하였다. 지표 가열은 화재 발생 지역 중앙에서 수직 바람장을 $0.7m\;s^{-1}(300^{\circ}C)$와 $1.1m\;s^{-1}(600^{\circ}C)$만큼 증가시켰다. 난류운동에너지는 화재의 열에너지 자체 및 열적 순환에 의해 증가된 운동에너지에 의해 증가하였다. 화재로 인한 열은 복잡한 지형과 강한 경계 바람 조건과 함께 화왕산의 예상하지 못한 난류와 강풍 조건을 유도하였다. CFD_NIMR_SNU 모형은 인명피해를 발생시킨 산불을 이해하는데 도움이 되는 귀중한 분석 자료를 제공하였다. 모사 결과에 따르면 화재 발생 지점은 풍상측의 높은 지형으로 인하여 화재 발생 직전까지는 바람이 거의 억제되었던 것으로 보인다. 이러한 바람의 억제는 화재 발생에 따른 뜨거운 공기의 상승과 강한 경계 바람 조건에 의해 쉽게 되돌려졌다. 즉, 강한 경계 바람과 화재로 인한 가열이 함께 작용하여 강한 난류가 만들어졌고, 여러 명의 사상자가 발생한 산악 화재로 확산되었던 것이다. CFD_NIMR_SNU 모형은 중규모 모형과의 결합을 통하여 좁은 영역의 화재로 인한 난류 예보를 생산하는 등 산불 예방을 위해 활용될 수 있을 것이다.

• 대기
• /
• 제26권3호
• /
• pp.357-372
• /
• 2016

Northeast Asian regions have recently become the main source of anthropogenic and natural aerosols. Measurement of aerosols on the sea in these regions have been rarely conducted since the experimental campaigns such as ACE-ASIA (Asian Pacific Regional Aerosol Characterization Experiment) in 2001. Research vessel observations of aerosol mass and chemical composition were performed on the Yellow and south sea around the Korean peninsula. The ship measurements showed six representative cases such as aerosol event and non-event cases during the study periods. On non-event cases, the anthropogenic chemical and natural soil composition on the Yellow sea were greater than those on the south sea. On aerosol event cases such as haze, haze with dust, and dust, the measured mass concentrations of anthropogenic chemical and element compositions were clearly changed by the events. In particular, methanesulfonate ($MSA^-$, $CH_3SO_3^-$), a main component of natural oceanic aerosol important for sulfur circulation on Earth, was first observed by the vessel in Korea, and its concentration on the Yellow sea was three times that on the south sea during the study period. Sea salt concentration important to chemical composition on the sea is related to wind speed. Coefficients of determination ($R^2$) between wind speed and sea salt concentration were 0.68 in $PM_{10}$ and 0.82 in $PM_{2.5}$. Maximum wave height was not found to be correlated to the sea salt concentration. When sea-salt comes into contact with pollutants, the total sea-salt mass is reduced, i.e., a loss of $Cl^-$ concentration from NaCl, the main chemical composing sea salt, is estimated by reaction with $HNO_3$(gas) and $H_2SO_4$(gas). The $Cl^-$ concentration loss by $SO_4^{2-}$ and $NO_3^-$ more easily increased for $PM_{10}$ compared to $PM_{2.5}$. The results of this study will be applied to verifying a dust-haze forecasting model. In addition, continued vessel measurements of aerosol data will become important to research for climate change studies in the future.

• 한국환경과학회지
• /
• 제29권1호
• /
• pp.45-54
• /
• 2020

High-resolution meteorological simulations were conducted using a Weather Research and Forecasting (WRF) model with an Urban Canopy Model (UCM) in the Ulsan Metropolitan Region (UMR) where large-scale industrial facilities are located on the coast. We improved the land cover input data for the WRF-UCM by reclassifying the default urban category into four detailed areas (low and high-density residential areas, commercial areas, and industrial areas) using subdivided data (class 3) of the Environmental and Geographical Information System (EGIS). The urban area accounted for about 12% of the total UMR and the largest proportion (47.4%) was in the industrial area. Results from the WRF-UCM simulation in a summer episode with high temperatures showed that the modeled temperatures agreed greatly with the observations. Comparison with a standard WRF simulation (WRF-BASE) indicated that the temporal and spatial variations in surface air temperature in the UMR were properly captured. Specifically, the WRF-UCM reproduced daily maximum and nighttime variations in air temperature very well, indicating that our model can improve the accuracy of temperature simulation for a summer heatwave. However, the WRF-UCM somewhat overestimated wind speed in the UMR largely due to an increased air temperature gradient between land and sea.