• Proceedings of the Korea Water Resources Association Conference
• /
• 2023.05a
• /
• pp.230-230
• /
• 2023

딥러닝 알고리즘 중 과거의 정보를 저장하는 문제(장기종속성 문제)가 있는 단순 RNN(Simple Recurrent Neural Network)의 단점을 해결한 LSTM(Long short-term memory)이 등장하면서 특정한 유역의 강우-유출 모형을 구축하는 연구가 증가하고 있다. 그러나 하나의 모형으로 모든 유역에 대한 유출을 예측하는 지역화 강우-유출 모형은 서로 다른 유역의 식생, 지형 등의 차이에서 발생하는 수문학적 행동의 차이를 학습해야 하므로 모형 구축에 어려움이 있다. 따라서, 본 연구에서는 국내 12개의 유역에 대하여 LSTM 기반 분포형 지역화 강우-유출 모형을 구축한 이후 강우 이외의 보조 자료에 따른 정확도를 살펴보았다. 국내 12개 유역의 7년 (2012.01.01-2018.12.31) 동안의 49개 격자(4km²)에 대한 10분 간격 레이더 강우, MODIS 위성 이미지 영상을 활용한 식생지수 (Normalized Difference Vegetation Index), 10분 간격 기온, 유역 평균 경사, 단순 하천 경사를 입력자료로 활용하였으며 10분 간격 유량 자료를 출력 자료로 사용하여 LSTM 기반 분포형 지역화 강우-유출 모형을 구축하였다. 이후 구축된 모형의 성능을 검증하기 위해 학습에 사용되지 않은 3개의 유역에 대한 자료를 활용하여 Nash-Sutcliffe Model Efficiency Coefficient (NSE)를 확인하였다. 식생지수를 보조 자료를 활용하였을 경우 제안한 모형은 3개의 검증 유역에 대하여 하천 흐름을 높은 정확도로 예측하였으며 딥러닝 모형이 위성 자료를 통하여 식생에 의한 차단 및 토양 침투와 같은 동적 요소의 학습이 가능함을 나타낸다.

• Korean Journal of Remote Sensing
• /
• v.29 no.5
• /
• pp.517-525
• /
• 2013

Time-series data of Normal Difference Vegetation Index (NDVI) obtained by the Moderate-resolution Imaging Spectroradiometer(MODIS) satellite imagery gives a waveform that reveals the characteristics of the phenology. The waveform can be decomposed into harmonics of various periods by the Fourier transformation. The resulting $n^{th}$ harmonics represent the amount of NDVI change in a period of a year divided by n. The values of each harmonics or their relative relation have been used to classify the vegetation species and to build a vegetation map. Here, we propose a method to estimate the annual amount of carbon absorbed on the forest from the $1^{st}$ harmonic NDVI value. The $1^{st}$ harmonic value represents the amount of growth of the leaves. By the allometric equation of trees, the growth of leaves can be considered to be proportional to the total amount of carbon absorption. We compared the $1^{st}$ harmonic NDVI values of the 6220 sample points with the reference data of the carbon absorption obtained by the field survey in the forest of South Korea. The $1^{st}$ harmonic values were roughly proportional to the amount of carbon absorption irrespective of the species and ages of the vegetation. The resulting proportionality constant between the carbon absorption and the $1^{st}$ harmonic value was 236 tCO2/5.29ha/year. The total amount of carbon dioxide absorption in the forest of South Korea over the last ten years has been estimated to be about 56 million ton, and this coincides with the previous reports obtained by other methods. Considering that the amount of the carbon absorption becomes a kind of currency like carbon credit, our method is very useful due to its generality.

• Korean Journal of Remote Sensing
• /
• v.36 no.5_4
• /
• pp.1179-1194
• /
• 2020

This study investigates the feasibility of three algorithms, K-Nearest Neighbors (K-NN), Random Forest (RF) and Neural Network (NN), for estimating the air temperature of an unobserved area where the weather station is not installed. The satellite image were obtained from Landsat-8 and MODIS Aqua/Terra acquired in 2019, and the meteorological ground weather data were from AWS/ASOS data of Korea Meteorological Administration and Korea Forest Service. In addition, in order to improve the estimation accuracy, a digital surface model, solar radiation, aspect and slope were used. The accuracy assessment of machine learning methods was performed by calculating the statistics of R² (determination coefficient) and Root Mean Square Error (RMSE) through 10-fold cross-validation and the estimated values were compared for each target area. As a result, the neural network algorithm showed the most stable result among the three algorithms with R² = 0.805 and RMSE = 0.508. The neural network algorithm was applied to each data set on Landsat imagery scene. It was possible to generate an mean air temperature map from June to September 2019 and confirmed that detailed air temperature information could be estimated. The result is expected to be utilized for national disaster safety management such as heat wave response policies and heat island mitigation research.

• Journal of Korea Water Resources Association
• /
• v.46 no.11
• /
• pp.1129-1140
• /
• 2013

Precipitation is one of the important factors in the hydrological cycle. It needs to understand accurate of spatial precipitation field because it has large spatio-temporal variability. Precipitation data obtained through the Tropical Rainfall Monitoring Mission (TRMM) 3B43 product is inaccurate because it has 25 km space scale. Downscaling of TRMM 3B43 product can increase the accuracy of spatial precipitation field from 25 km to 1 km scale. The relationship between precipitation and the normalized difference vegetation index(NDVI) (1 km space scale) which is obtained from the Moderate Resolution Imaging Spectroradiometers (MODIS) sensor loaded in Terra satellite is variable at different scales. Therefore regression equations were established and these equations apply to downscaling. Two renormalization strategies, Geographical Difference Analysis (GDA) and Geographical Ratio Analysis (GRA) are implemented for correcting the differences between remote sensing-derived and rain gauge data. As for considering the GDA method results, biases, the root mean-squared error (RMSE), MAE and Index of agreement (IOA) is equal to 4.26 mm, 172.16 mm, 141.95 mm, 0.64 in 2009 and 17.21 mm, 253.43 mm, 310.56 mm, 0.62 in 2011. In this study, we can see the 1km spatial precipitation field map over Korea. It will be possible to get more accurate spatial analysis of the precipitation field through using the additional rain gauges or radar data.

• Korean Journal of Remote Sensing
• /
• v.32 no.3
• /
• pp.201-220
• /
• 2016

In this study, the net solar radiation fluxes at the surface are retrieved by updating an existing algorithm to be applicable for MODerate resolution Imaging Spectroradiometer (MODIS) observations, in which linear relationships between the solar radiation reflected from the top of atmosphere and the net surface solar radiation are employed. The results of this study have been evaluated through intercomparison with existing Clouds and the Earth's Radiant Energy System (CERES) data products and ground-based data from pyranometers at Gangneung-Wonju National University (GWNU) and the Southern Great Plains (SGP) of observatory of Atmospheric Radiation Measurement (ARM) site. Prior to the comparison of the surface radiation energy in relation to the energy balance of the earth, the radiation energy of the upper part of the atmosphere was compared. As a result, the coefficient of determination was over 0.9, showing considerable similarity, but the Root-Mean-Square-Deviation (RMSD) value was somewhat different, and the downward and net solar-radiation energy also showed similar results. The surface solar radiation data measured from pyranometers at Gangneung-Wonju National University (GWNU) and Atmospheric Radiation Measurement (ARM) observatory are used to validate the solar radiation data produced in this study. When compared to the GWNU, The results of this study show smaller RMSD values than CERES data, showing slightly better agreements with the surface data. On the other hand, when compared with the data from ARM SGP observatory, the results of this study bear slightly larger RMSD values than those for CERES. The downward and net solar radiation estimated by the algorithm of this study at a high spatial resolution are expected to be very useful in the near future after refinements on the identified problems, especially for those area without ground measurements of solar radiation.

• Korean Journal of Remote Sensing
• /
• v.34 no.5
• /
• pp.787-799
• /
• 2018

Accurate and effective mapping is critical step to monitor the spatial distribution and change of flood inundated area in large scale flood event. In this study, we try to suggest methods to use low spatial resolution satellite optical imagery for flood mapping, which has high temporal resolution to cover wide geographical area several times per a day. We selected the Sebou watershed flood in Morocco that was occurred in early 2010, in which several hundred $km^2$ area of the Gharb lowland plain was inundated. MODIS daily surface reflectance product was used to detect the flooded area. The study area showed several distinct spectral patterns within the flooded area, which included pure turbid water and turbid water with vegetation. The flooded area was extracted by thresholding on selected band reflectance and water-related spectral indices. Accuracy of these flooding detection methods were assessed by the reference map obtained from Landsat-5 TM image and qualitative interpretation of the flood map derived. Over 90% of accuracies were obtained for three methods except for the NDWI threshold. Two spectral bands of SWIR and red were essential to detect the flooded area and the simple thresholding on these bands was effective to detect the flooded area. NIR band did not play important role to detect the flooded area while it was useful to separate the water-vegetation mixed flooded classes from the purely water surface.

• Korean Journal of Remote Sensing
• /
• v.32 no.2
• /
• pp.73-86
• /
• 2016

In this study, an algorithm to estimate Aerosol Optical Thickness (AOT) over small cities during nighttime has been developed by using the radiance from artificial light sources in small cities measured from Visible Infrared Imaging Radiometer Suite (VIIRS) sensor's Day/Night Band (DNB) aboard the Suomi-National Polar Partnership (Suomi-NPP) satellite. The algorithm is based on Beer's extinction law with the light sources from the artificial lights over small cities. AOT is retrieved for cloud-free pixels over individual cities, and cloud-screening was conducted by using the measurements from M-bands of VIIRS at infrared wavelengths. The retrieved nighttime AOT is compared with the aerosol products from MODerate resolution Imaging Spectroradiometer (MODIS) aboard Terra and Aqua satellites. As a result, the correlation coefficients over individual cities range from around 0.6 and 0.7 between the retrieved nighttime AOT and MODIS AOT with Root-Mean-Squared Difference (RMSD) ranged from 0.14 to 0.18. In addition, sensitivity tests were conducted for the factors affecting the nighttime AOT to estimate the range of uncertainty in the nighttime AOT retrievals. The results of this study indicate that it is promising to infer AOT using the DNB measaurements over small cities in Korea at night. After further development and refinement in the future, the developed retrieval algorithm is expected to produce nighttime aerosol information which is not operationally available over Korea.

• Korean Journal of Agricultural and Forest Meteorology
• /
• v.13 no.1
• /
• pp.41-46
• /
• 2011

Reliable information on the surface solar radiation is indispensable for rebuilding food production system in the famine plagued North Korea. However, transfer of the related modeling technology of South Korea is not possible simply because raw data such as solar radiation or sunshine duration are not available. The objective of this study is restoring solar radiation data at 27 synoptic stations in North Korea by using satellite remote sensing data. We derived relationships between MODIS radiation estimates and the observed solar radiation at 18 locations in South Korea. The relationships were used to adjust the MODIS based radiation data and to restore solar radiation data at those pixels corresponding to the 27 North Korean synoptic stations. Inverse distance weighted averaging of the restored solar radiation data resulted in gridded surfaces of monthly solar radiation for 4 decadal periods (1983-1990, 1991-2000 and 2001-2010), respectively. For a direct application of these products, we produced solar irradiance estimates for each sub-grid cell with a 30 m spacing based on a sun-slope geometry. These products are expected to assist planning of the North Korean agriculture and, if combined with the already prepared South Korean data, can be used for climate change impact assessment across the whole Peninsula.

• Korean Journal of Remote Sensing
• /
• v.31 no.6
• /
• pp.523-530
• /
• 2015

A Multiple Regression Method (MRM) is used for the first time with Ozone Monitoring Instrument (OMI) and Moderate Resolution Imaging Spectroradiometer (MODIS) data to estimate formaldehyde (HCHO) Vertical Column Density (VCD). For a 3.5-year period from January 2005 through July 2008, HCHO VCD estimation is investigated in cities over Asia in two categorized areas: (1) Major cities in Northeast Asia (Beijing, Seoul, and Tokyo), (2) Major cities in Southeast Asia (New Delhi, Dhaka, and Bangkok). In the Major cities in Northeast Asia, there are good agreements between HCHO estimated by the multiple linear regression method ($HCHO_{MRM}$) and HCHO measured by OMI ($HCHO_{OMI}$) (0.78 < $R^2$ < 0.82). However, in Major cities in Southeast Asia, there were poor agreements between $HCHO_{OMI}$ and $HCHO_{MRM}$ (0.24 < $R^2$ < 0.39). In addition, an unbiased assessment of the MRM performance using modeling and validation groups shows that the performance of the MRM based on separate modeling and validation groups is comparable to that using all the data for deriving Multiple Regression Equations (MREs). This study demonstrates that MRM can be an alternative tool for HCHO estimation in certain areas over Asia.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.14 no.2
• /
• pp.109-127
• /
• 2011

Atmospheric aerosols, small particles in the atmosphere, are one of the important parameters in climate change and human health. Additionally, accurate estimates of aerosol species are increasingly important in environmental impact assessment studies. Recent advances in global satellite remote sensing provide powerful tool for air quality monitoring. This study explores the potential usage of satellite derived data such as atmospheric aerosols for air quality monitoring as well as climate change study. The objectives of this study is to understand the general features of the global distribution of type dependent aerosols. A detailed spatio-temporal variability of the each different satellite dataset shows the variation of the global zonal average and specific geographical regions where the strong emission sources are located. Especially, significantly large aerosol amounts are observed in Asia and Africa because of the desert dust storm, anthropogenic and biomass burning emissions.