다종 위성자료와 기계학습을 이용한 고해상도 표층 염분 추정

성태준,심성문,장은나,임정호,Sung, Taejun,Sim, Seongmun,Jang, Eunna,Im, Jungho 대한원격탐사학회 2022 大韓遠隔探査學會誌 Vol.38 No.5

해양 염분은 전 지구 규모에서 해수 순환에 영향을 미칠 뿐만 아니라, 연·근해 지역 저염분수가 어족자원 및 수산업에 피해를 줄 수 있는 등 해양 식생환경의 변화를 줄 수 있다. 해수의 표면 특성인 sea surface salinity (SSS)에 따라 마이크로웨이브 영역의 방사율이 달라지며, 이를 통해 Soil Moisture Active Passive (SMAP) 등 위성 센서를 활용한 SSS 산출물이 제공되고 있다. 하지만 마이크로파 위성 센서 기반의 SSS 산출물은 낮은 시공간해상도로 자료를 생산하며, 연안지역과 고위도 지역에서 정확도가 낮다. 이러한 이유로 연·근해 지역 SSS의 상세한 시공간적 변화를 관측하기에는 적합하지 않다. 본 연구에서는 Jang et al. (2022)에서 제시한 기계학습 기반의 개선된 SMAP SSS (SMAP SSS (Jang))를 참조자료로 활용하여, 정지궤도해색센서(Geostationary Ocean Color Imager, GOCI) 영상으로부터 고해상도 SSS를 추정하는 Light Gradient Boosting Machine (LGBM) 기반의 모델을 개발하였다. 3가지 입력변수 조합을 테스트하였고, Multi-scale Ultra-high Resolution Sea Surface Temperature (SST) 자료가 추가된 scheme 3가 가장 높은 정확도를 보였다(R² = 0.60, RMSE = 0.91 psu). 이를 바탕으로 본 연구영역에서 SST가 SSS 모의에 효과적인 환경변수로 작용함을 보였다. 본 연구에서 제시한 LGBM 기반의 GOCI SSS는 SMAP SSS (Jang)와 비슷한 시공간적 패턴을 보였지만, 더 높은 공간해상도를 바탕으로 SSS의 보다 상세한 공간적 분포와 더불어 SMAP SSS (Jang)에서 산출하지 않는 연안 지역의 정보까지 모의하였다. 또한, 중국 남방지역에 대홍수가 발생하였던 2020년 8월을 대상으로 양자강 유출수(Changjiang Diluted Water)의 거동을 분석한 결과, GOCI SSS는 한국 해양수산연구원의 보도자료와 비교하여 일관성 있는 시공간적 변화를 보였다. 본 연구의 결과로 연안 지역의 저염수 뿐 아니라, 원해 지역에서 광학위성 신호를 활용한 고해상도 SSS 산출의 가능성을 제시하였다. Ocean salinity affects ocean circulation on a global scale and low salinity water around coastal areas often has an impact on aquaculture and fisheries. Microwave satellite sensors (e.g., Soil Moisture Active Passive [SMAP]) have provided sea surface salinity (SSS) based on the dielectric characteristics of water associated with SSS and sea surface temperature (SST). In this study, a Light Gradient Boosting Machine (LGBM)-based model for generating high resolution SSS from Geostationary Ocean Color Imager (GOCI) data was proposed, having machine learning-based improved SMAP SSS by Jang et al. (2022) as reference data (SMAP SSS (Jang)). Three schemes with different input variables were tested, and scheme 3 with all variables including Multi-scale Ultra-high Resolution SST yielded the best performance (coefficient of determination = 0.60, root mean square error = 0.91 psu). The proposed LGBM-based GOCI SSS had a similar spatiotemporal pattern with SMAP SSS (Jang), with much higher spatial resolution even in coastal areas, where SMAP SSS (Jang) was not available. In addition, when tested for the great flood occurred in Southern China in August 2020, GOCI SSS well simulated the spatial and temporal change of Changjiang Diluted Water. This research provided a potential that optical satellite data can be used to generate high resolution SSS associated with the improved microwave-based SSS especially in coastal areas.

GOCI 위성영상과 기계학습 기법을 이용한 Ocean Colour-Climate Change Initiative (OC-CCI) Forel-Ule Index의 공간 상세화

성태준 ( Taejun Sung ),김영준 ( Young Jun Kim ),최현영 ( Hyunyoung Choi ),임정호 ( Jungho Im ) 대한원격탐사학회 2021 大韓遠隔探査學會誌 Vol.37 No.5

Forel-Ule Index (FUI)는 자연에 존재하는 담수 및 해수의 색을 남색부터 고동색까지 21 가지의 등급으로 구분하는 지표이다. FUI는 여러 선행연구에서 수계의 부영양화 지수, 수질인자, 광 특성 등과 연관 지어 분석되었으며, 여러 수질인자의 광학적 정보를 동시에 가지고 있는 새로운 수질 지표로써의 가능성이 제시되었다. 본 연구에서는 500 m의 높은 공간해상도를 가지는 정지궤도 해양위성해색탑재체(Geostationary Ocean Color Imager; GOCI) 관측 자료와 Random Forest (RF) 기계학습 기법을 활용하여Ocean Colour-Climate Change Initiative (OC-CCI) 기반의 4 km FUI 자료를 공간 상세화 시켰다. 이를 활용하여 우리나라 연안 해역에 대한 수질인자와 의 상관관계와 주요 해역에 대한 FUI의 공간적 분포 및 계절별 특성 변화를 분석하였다. 검증 결과 RF 기법으로 추정한 RF FUI는 결정계수(R²)=0.81, 평균 제곱근 오차(Root Mean Square Error; RMSE)=0.7784로, Pitarch의 OC-CCI FUI 알고리즘을 적용하여 계산한 GOCI FUI 추정 정확도(R²=0.72, RMSE=0.9708) 대비 향상된 결과를 보였다. RF FUI는 총 질소(Total Nitrogen), 총 인(Total Phosphorus), 클로로필-a(Chlorophyll-a), 총 부유물질(Total Suspended Solids), 투명도(Secchi Disk Depth)를 포함하는 5가지 수질인자와 각각 0.87, 0.88, 0.97, 0.65, -0.98의 상관계수로 강한 상관성을 보였다. 산출된 FUI의 시간적 패턴 역시 여러 수질인자와의 물리적 관계를 반영하며 유의미한 계절적 패턴의 변화를 보였다. 본 연구의 결과로 한반도 연안 수질 관리에서 고해상도 FUI의 활용 가능성을 제시하였다. Forel-Ule Index (FUI) is an index which classifies the colors of inland and seawater exist in nature into 21 grades ranging from indigo blue to cola brown. FUI has been analyzed in connection with the eutrophication, water quality, and light characteristics of water systems in many studies, and the possibility as a new water quality index which simultaneously contains optical information of water quality parameters has been suggested. In this study, Ocean Colour-Climate Change Initiative (OC-CCI) based 4 km FUI was spatially downscaled to the resolution of 500 m using the Geostationary Ocean Color Imager (GOCI) data and Random Forest (RF) machine learning. Then, the RF-derived FUI was examined in terms of its correlation with various water quality parameters measured in coastal areas and its spatial distribution and seasonal characteristics. The results showed that the RF-derived FUI resulted in higher accuracy (Coefficient of Determination (R²)=0.81, Root Mean Square Error (RMSE)=0.7784) than GOCI-derived FUI estimated by Pitarch’s OC-CCI FUI algorithm (R²=0.72, RMSE=0.9708). RF-derived FUI showed a high correlation with five water quality parameters including Total Nitrogen, Total Phosphorus, Chlorophyll-a, Total Suspended Solids, Transparency with the correlation coefficients of 0.87, 0.88, 0.97, 0.65, and -0.98, respectively. The temporal pattern of the RF-derived FUI well reflected the physical relationship with various water quality parameters with a strong seasonality. The research findings suggested the potential of the high resolution FUI in coastal water quality management in the Korean Peninsula.

GK-2A 위성 영상과 기계 학습 기법을 이용한 실시간 산불 탐지

성태준(Taejun Sung),강유진(Yoojin Kang),임정호(Jung-Ho Im) 한국통신학회 2022 한국통신학회 학술대회논문집 Vol.2022 No.6

GOCI 위성영상과 기계학습 기법을 이용한 고해상도 Forel-Ule Index 산출 알고리즘 개발

성태준(Sung, Taejun),김영준(Kim, Young Jun),최현영(Choi, Hyunyoung),임정호(Im, Jungho) 한국측량학회 2021 한국측량학회 학술대회자료집 Vol.2021 No.11

Himawari-8 정지궤도 위성 영상을 활용한 딥러닝 기반 산불 탐지의 효율적 방안 제시

이시현,강유진,성태준,임정호,Sihyun Lee,Yoojin Kang,Taejun Sung,Jungho Im 대한원격탐사학회 2023 大韓遠隔探査學會誌 Vol.39 No.5

As wildfires are difficult to predict, real-time monitoring is crucial for a timely response. Geostationary satellite images are very useful for active fire detection because they can monitor a vast area with high temporal resolution (e.g., 2 min). Existing satellite-based active fire detection algorithms detect thermal outliers using threshold values based on the statistical analysis of brightness temperature. However, the difficulty in establishing suitable thresholds for such threshold-based methods hinders their ability to detect fires with low intensity and achieve generalized performance. In light of these challenges, machine learning has emerged as a potential-solution. Until now, relatively simple techniques such as random forest, Vanilla convolutional neural network (CNN), and U-net have been applied for active fire detection. Therefore, this study proposed an active fire detection algorithm using state-of-the-art (SOTA) deep learning techniques using data from the Advanced Himawari Imager and evaluated it over East Asia and Australia. The SOTA model was developed by applying EfficientNet and lion optimizer, and the results were compared with the model using the Vanilla CNN structure. EfficientNet outperformed CNN with F1-scores of 0.88 and 0.83 in East Asia and Australia, respectively. The performance was better after using weighted loss, equal sampling, and image augmentation techniques to fix data imbalance issues compared to before the techniques were used, resulting in F1-scores of 0.92 in East Asia and 0.84 in Australia. It is anticipated that timely responses facilitated by the SOTA deep learning-based approach for active fire detection will effectively mitigate the damage caused by wildfires.

기계학습을 이용한 광학 위성 영상 기반의 도시 내 수목 피복률 추정

배세정,손보경,성태준,이연수,임정호,강유진,Sejeong Bae,Bokyung Son,Taejun Sung,Yeonsu Lee,Jungho Im,Yoojin Kang 대한원격탐사학회 2023 大韓遠隔探査學會誌 Vol.39 No.5

Urban trees play a vital role in urban ecosystems,significantly reducing impervious surfaces and impacting carbon cycling within the city. Although previous research has demonstrated the efficacy of employing artificial intelligence in conjunction with airborne light detection and ranging (LiDAR) data to generate urban tree information, the availability and cost constraints associated with LiDAR data pose limitations. Consequently, this study employed freely accessible, high-resolution multispectral satellite imagery (i.e., Sentinel-2 data) to estimate fractional tree canopy cover (FTC) within the urban confines of Suwon, South Korea, employing machine learning techniques. This study leveraged a median composite image derived from a time series of Sentinel-2 images. In order to account for the diverse land cover found in urban areas, the model incorporated three types of input variables: average (mean) and standard deviation (std) values within a 30-meter grid from 10 m resolution of optical indices from Sentinel-2, and fractional coverage for distinct land cover classes within 30 m grids from the existing level 3 land cover map. Four schemes with different combinations of input variables were compared. Notably, when all three factors (i.e., mean, std, and fractional cover) were used to consider the variation of landcover in urban areas(Scheme 4, S4), the machine learning model exhibited improved performance compared to using only the mean of optical indices (Scheme 1). Of the various models proposed, the random forest (RF) model with S4 demonstrated the most remarkable performance, achieving R² of 0.8196, and mean absolute error (MAE) of 0.0749, and a root mean squared error (RMSE) of 0.1022. The std variable exhibited the highest impact on model outputs within the heterogeneous land covers based on the variable importance analysis. This trained RF model with S4 was then applied to the entire Suwon region, consistently delivering robust results with an R² of 0.8702, MAE of 0.0873, and RMSE of 0.1335. The FTC estimation method developed in this study is expected to offer advantages for application in various regions, providing fundamental data for a better understanding of carbon dynamics in urban ecosystems in the future.