• 한국콘텐츠학회논문지
• /
• 제18권1호
• /
• pp.154-163
• /
• 2018

우리나라뿐만 아니라 전 지구적으로 호우발생 빈도의 증가, 태풍이나 허리케인 세기의 강화 등에 따라 대규모 자연재해의 발생횟수와 피해액은 지속적으로 증가하는 추세이다. 태풍, 홍수, 호우, 강풍, 풍랑, 해일, 조수, 대설, 가뭄, 지진, 황사 등과 같은 자연재해는 발생지점과 규모를 예측하기 어려우며, 전조현상이 명확하게 나타나지 않아 대응에 많은 어려움이 존재한다. 그러나 자연재해의 피해규모를 예측할 수 있다면, 조기대응을 통해 피해를 저감할 수 있을 것이다. 따라서 본 연구에서는 국민안전처에서 발간하는 재해연보('91년~'15년)를 기반으로 서해연안지역의 풍랑피해함수를 개발하였다. 풍랑피해함수는 지역별, 시설별로 구분하여 개발하였으며, NRMSE는 1.94%~26.07%로 분석되었다. 개발된 식을 통해 피해규모를 예측하고, 그에 대한 적절한 대응이 이루어진다면, 피해를 저감할 수 있을 것으로 사료된다.

• 지구물리와물리탐사
• /
• 제17권4호
• /
• pp.242-246
• /
• 2014

지층의 전기화학적인 물성을 이용한 광대역유도분극(SIP) 탐사는 황화광물을 포함한 금속광물탐사에 유용한 기술이다. 탐사자료로부터 IP 물성을 계산하기 위해서는 등가회로 분석을 수행한다. 분석에 사용되는 암석의 SIP 반응을 고려한 등가회로 모델은 해의 비유일성이라는 문제를 가지기 때문에 정확한 분석을 위해 적절한 모델을 설정하는 것이 매우 중요하다. 따라서 이 연구는 SIP 이상반응을 나타내는 광석의 분석에 적합한 새로운 모델을 제안하고자 하였다. 이 모델은 기존의 Dias model과 Cole-Cole model의 비교를 통하여 적합성을 검증하였다. 그 결과, Dias model과 Cole-Cole model을 이용한 분석 결과의 NRMSE 오차는 각각 10.05%와 17.03%를 보였다. 하지만 제안한 새로운 모델의 NRMSE 오차는 0.87%로 상당히 낮았기 때문에 다른 모델보다 SIP 이상 자료의 등가회로 분석에 유용하고 판단하였다.

• 한국토양비료학회지
• /
• 제49권5호
• /
• pp.420-428
• /
• 2016

For more than 50 years, satellite images have been used to monitor crop growth. Currently, unmanned aerial vehicle (UAV) imagery is being assessed for analyzing within field spatial variability for agricultural precision management, because UAV imagery may be acquired quickly during critical periods of rapid crop growth. This study refers to the derivation of growth estimating equation for highland Kimchi cabbage using UAV derived normalized difference vegetation index (NDVI) and agro-meteorological factors. Anbandeok area in Gangneung, Gangwon-do, Korea is one of main districts producing highland Kimchi cabbage. UAV imagery was taken in the Anbandeok ten times from early June to early September. Meanwhile, three plant growth parameters, plant height (P.H.), leaf length (L.L.) and outer leaf number (L.N.), were measured for about 40 plants (ten plants per plot) for each ground survey. Six agro-meteorological factors include average temperature; maximum temperature; minimum temperature; accumulated temperature; rainfall and irradiation during growth period. The multiple linear regression models were suggested by using stepwise regression in the extraction of independent variables. As a result, $NDVI_{UAV}$ and rainfall in the model explain 93% of the P.H. and L.L. with a root mean square error (RMSE) of 2.22, 1.90 cm. And $NDVI_{UAV}$ and accumulated temperature in the model explain 86% of the L.N. with a RMSE of 4.29. These lead to the result that the characteristics of variations in highland Kimchi cabbage growth according to $NDVI_{UAV}$ and other agro-meteorological factors were well reflected in the model.

• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2019년도 학술발표회
• /
• pp.153-153
• /
• 2019

Evapotranspiration (ET) is an important component of hydrological processes. Accurate estimates of ET variation are of vital importance for natural hazard adaptation and water resource management. This study first developed a soil water index (SWI)-based Priestley-Taylor algorithm (SWI-PT) based on the enhanced vegetation index (EVI), SWI, net radiation, and temperature. The algorithm was then compared with a modified satellite-based Priestley-Taylor ET model (MS-PT). After examining the performance of the two models at 10 flux tower sites in different land cover types over East Asia and Australia, the daily estimates from the SWI-PT model were closer to observations than those of the MS-PT model in each land cover type. The average correlation coefficient of the SWI-PT model was 0.81, compared with 0.66 in the original MS-PT model. The average value of the root mean square error decreased from $36.46W/m^2$ to $23.37W/m^2$ in the SWI-PT model, which used different variables of soil moisture and vegetation indices to capture soil evaporation and vegetative transpiration, respectively. By using the EVI and SWI, uncertainties involved in optimizing vegetation and water constraints were reduced. The estimated ET from the MS-PT model was most sensitive (to the normalized difference vegetation index (NDVI) in forests) to net radiation ($R_n$) in grassland and cropland. The estimated ET from the SWI-PT model was most sensitive to $R_n$, followed by SWI, air temperature ($T_a$), and the EVI in each land cover type. Overall, the results showed that the MS-PT model estimates of ET in forest and cropland were weak. By replacing the fraction of soil moisture ($f_{sm}$) with the SWI and the NDVI with the EVI, the newly developed SWI-PT model captured soil evaporation and vegetation transpiration more accurately than the MS-PT model.

• 한국농공학회논문집
• /
• 제61권6호
• /
• pp.123-132
• /
• 2019

This study is to estimate the spatial soil moisture using Terra MODIS (Moderate Resolution Imaging Spectroradiometer) satellite data and machine learning technique. Using the 3 years (2015~2017) data of MODIS 16 days composite NDVI (Normalized Difference Vegetation Index) and daily Land Surface Temperature (LST), ground measured precipitation and sunshine hour of KMA (Korea Meteorological Administration), the RDA (Rural Development Administration) 10 cm~30 cm average TDR (Time Domain Reflectometry) measured soil moisture at 78 locations was tested. For daily analysis, the missing values of MODIS LST by clouds were interpolated by conditional merging method using KMA surface temperature observation data, and the 16 days NDVI was linearly interpolated to 1 day interval. By applying the RNN-LSTM (Recurrent Neural Network-Long Short Term Memory) artificial neural network model, 70% of the total period was trained and the rest 30% period was verified. The results showed that the coefficient of determination ($R^2$), Root Mean Square Error (RMSE), and Nash-Sutcliffe Efficiency were 0.78, 2.76%, and 0.75 respectively. In average, the clay soil moisture was estimated well comparing with the other soil types of silt, loam, and sand. This is because the clay has the intrinsic physical property for having narrow range of soil moisture variation between field capacity and wilting point.

• 한국농공학회논문집
• /
• 제63권6호
• /
• pp.101-115
• /
• 2021

The purpose of this study is to evaluate the compatibility of the vegetation index between the two satellites and the applicability of agricultural monitoring by comparing and verifying NDVI (Normalized Difference Vegetation Index) based on Sentinel-2 and Terra MODIS (Moderate Resolution Imaging Spectroradiometer). Terra MODIS NDVI utilized 16-day MOD13Q1 data with 250 m spatial resolution, and Sentinel-2 NDVI utilized 10-day Level-2A BOA (Bottom Of Atmosphere) data with 10 m spatial resolution. To compare both NDVI, Sentinel-2 NDVIs were reproduced at 16-day intervals using the MVC (Maximum Value Composite) technique. As a result of time series NDVIs based on two satellites for 2019 and compare by land cover, the average R² (Coefficient of determination) and RMSE (Root Mean Square Error) of the entire land cover were 0.86 and 0.11, which indicates that Sentinel-2 NDVI and MODIS NDVI had a high correlation. MODIS NDVI is overestimated than Sentinel-2 NDVI for all land cover due to coarse spatial resolution. The high-resolution Sentinel-2 NDVI was found to reflect the characteristics of each land cover better than the MODIS NDVI because it has a higher discrimination ability for subdivided land cover and land cover with a small area range.

• Investigative Magnetic Resonance Imaging
• /
• 제25권1호
• /
• pp.10-22
• /
• 2021

Purpose: To overcome the difficulty in building a large data set with a high-quality in medical imaging, a concept of 'blended-transfer learning' (BTL) using a combination of both source data and target data is proposed for the target task. Materials and Methods: Source and target tasks were defined as training of the source and target networks to reconstruct cardiac CINE images from undersampled data, respectively. In transfer learning (TL), the entire neural network (NN) or some parts of the NN after conducting a source task using an open data set was adopted in the target network as the initial network to improve the learning speed and the performance of the target task. Using BTL, an NN effectively learned the target data while preserving knowledge from the source data to the maximum extent possible. The ratio of the source data to the target data was reduced stepwise from 1 in the initial stage to 0 in the final stage. Results: NN that performed BTL showed an improved performance compared to those that performed TL or standalone learning (SL). Generalization of NN was also better achieved. The learning curve was evaluated using normalized mean square error (NMSE) of reconstructed images for both target data and source data. BTL reduced the learning time by 1.25 to 100 times and provided better image quality. Its NMSE was 3% to 8% lower than with SL. Conclusion: The NN that performed the proposed BTL showed the best performance in terms of learning speed and learning curve. It also showed the highest reconstructed-image quality with the lowest NMSE for the test data set. Thus, BTL is an effective way of learning for NNs in the medical-imaging domain where both quality and quantity of data are always limited.

• 대한방사선기술학회지:방사선기술과학
• /
• 제44권6호
• /
• pp.655-661
• /
• 2021

Positron emission tomography (PET) images is affected by acquisition time, short acquisition times results in low gamma counts leading to degradation of image quality by statistical noise. Noise2Void(N2V) is self supervised denoising model that is convolutional neural network (CNN) based deep learning. The purpose of this study is to evaluate denoising performance of N2V for PET image with a short acquisition time. The phantom was scanned as a list mode for 10 min using Biograph mCT40 of PET/CT (Siemens Healthcare, Erlangen, Germany). We compared PET images using NEMA image-quality phantom for standard acquisition time (10 min), short acquisition time (2min) and simulated PET image (S2 min). To evaluate performance of N2V, the peak signal to noise ratio (PSNR), normalized root mean square error (NRMSE), structural similarity index (SSIM) and radio-activity recovery coefficient (RC) were used. The PSNR, NRMSE and SSIM for 2 min and S2 min PET images compared to 10min PET image were 30.983, 33.936, 9.954, 7.609 and 0.916, 0.934 respectively. The RC for spheres with S2 min PET image also met European Association of Nuclear Medicine Research Ltd. (EARL) FDG PET accreditation program. We confirmed generated S2 min PET image from N2V deep learning showed improvement results compared to 2 min PET image and The PET images on visual analysis were also comparable between 10 min and S2 min PET images. In conclusion, noisy PET image by means of short acquisition time using N2V denoising network model can be improved image quality without underestimation of radioactivity.

• 한국농림기상학회지
• /
• 제24권2호
• /
• pp.124-132
• /
• 2022

불균일한 미세먼지 관측값으로부터 남한 전체에 대한 공간적 분포를 추정하기 위해서는 적절한 보간 처리가 필수이다. 본 연구에서는 2019년도에 미세먼지 농도가 높았던 1월달과 농도가 낮았던 7월달의 전국의 기상청 및 AirKorea 측정소 자료를 이용하여 IDW, OK, SI, RBF 총 4가지 보간법을 테스트하였다. 각 보간 방법별 세부 인자를 고려한 총 6가지 경우에 대해 보간 처리 및 교차 검증을 진행하였다. 자료 처리속도는 SI, RBF, IDW, OK 순으로 빠르게 나타났다. 교차 검증의 결과, IDW가 상대적으로 제일 낮은 NRMSE 결과를 보였고 OK방법이 가장 큰 NRMSE를 보였다. 이러한 연구의 결과는 사용자가 남한 지역에서 불균일한 미세먼지 관측 자료를 사용하여 전체 수평 공간을 보간할 때 적합한 방법을 단기간에 선택하고 신뢰성과 효과성 있는 분석을 실시하는데 도움이 될 것으로 기대된다.

• 한국해안·해양공학회논문집
• /
• 제34권1호
• /
• pp.11-18
• /
• 2022

본 연구는 대한해협 인근 입자추적 예측 기법의 정확도 개선을 위해서 해수유동 수치모델 결과를 이용하여 만든 입자추적 모델과 현장 관측 자료를 이용한 기계학습 기반 입자 추적 모델을 비교 및 분석하였다. 세부 연구 방법으로는 대한해협에서 관측된 표층 뜰개 이동 궤적 자료, 3개 관측소(가거도, 거제도, 교본초 관측소)의 조위 및 바람자료를 학습시켜 만든 기계 학습(선형 회귀, 의사결정나무) 기반 예측자료, 수치모델 예측자료(ROMS, MOHID)를 3가지 오차평가방법(CC, RMSE, NCLS)을 통해 비교하였다. 최종 결과로서 CC와 RMSE에서는 의사결정나무 모델의 예측 정확도가 가장 우수하였고 NCLS에서는 MOHID 모델의 예측 결과가 가장 우수하였다.