Performance of Support Vector Machine for Classifying Land Cover in Optical Satellite Images: A Case Study in Delaware River Port Area

Ramayanti, Suci,Kim, Bong Chan,Park, Sungjae,Lee, Chang-Wook The Korean Society of Remote Sensing 2022 大韓遠隔探査學會誌 Vol.38 No.6

The availability of high-resolution satellite images provides precise information without direct observation of the research target. Korea Multi-Purpose Satellite (KOMPSAT), also known as the Arirang satellite, has been developed and utilized for earth observation. The machine learning model was continuously proven as a good classifier in classifying remotely sensed images. This study aimed to compare the performance of the support vector machine (SVM) model in classifying the land cover of the Delaware River port area on high and medium-resolution images. Three optical images, which are KOMPSAT-2, KOMPSAT-3A, and Sentinel-2B, were classified into six land cover classes, including water, road, vegetation, building, vacant, and shadow. The KOMPSAT images are provided by Korea Aerospace Research Institute (KARI), and the Sentinel-2B image was provided by the European Space Agency (ESA). The training samples were manually digitized for each land cover class and considered the reference image. The predicted images were compared to the actual data to obtain the accuracy assessment using a confusion matrix analysis. In addition, the time-consuming training and classifying were recorded to evaluate the model performance. The results showed that the KOMPSAT-3A image has the highest overall accuracy and followed by KOMPSAT-2 and Sentinel-2B results. On the contrary, the model took a long time to classify the higher-resolution image compared to the lower resolution. For that reason, we can conclude that the SVM model performed better in the higher resolution image with the consequence of the longer time-consuming training and classifying data. Thus, this finding might provide consideration for related researchers when selecting satellite imagery for effective and accurate image classification.

Measurement of surface deformation related to the December 2018 Mt. Etna eruption using time-series interferometry and magma modeling for hazard zone mapping

Suci Ramayanti,Arief R. Achmad,정한철,조민정,김상완,박유철,이창욱 한국지질과학협의회 2022 Geosciences Journal Vol.26 No.6

Mount Etna has erupted several times since it was first formed. Recently, Mount Etna began erupting again over 24–27 December 2018. Because it erupts frequently, Mount Etna should be observed on a frequent basis. From June 2018 to October 2019, 34 and 56 synthetic aperture radar (SAR) images were acquired from the ascending and descending tracks of the Sentinel-1 satellite, respectively. We employed the Stanford Method for Persistent Scatterers (StaMPS) and a refined small baseline subset (SBAS) InSAR method to produce a surface deformation time-series map. In the time-series analysis, the phase signal remained unaltered with time. The Okada model was then applied to the result to generate a modeled interferogram, and the Q-LavHA program was run to generate a lava flow prediction model. A direct comparison of the results showed that Persistent Scatterers Interferometry (PSI)-StaMPS and the refined SBAS technique were comparable in terms of the displacement pattern, with slightly different velocity values obtained for individual points. In particular, a velocity range of −25 to 21 cm/yr was obtained from PSI-StaMPS, whereas a range of −30 to 25 cm/yr was obtained from the refined SBAS method. Upon computation of the vertical and east-west displacement components based on ascending and descending track data using both methods, deformation velocities of 51.5 and 52.5 cm/yr in the westerly direction on the western flank of Mount Etna were obtained from PSI-StaMPS and the refined SBAS method, respectively, whereas on the eastern flank, deformation toward the east was estimated to occur at a velocity of 50.1 or 54.2 cm/yr, respectively. PSI-StaMPS estimated a vertical deformation velocity of −5.3 to 18.3 cm/yr, whereas the refined SBAS method produced a velocity range of approximately −7 to 19 cm/yr. The interferogram obtained via Okada modeling showed two fault sources in the 2018 Mount Etna eruption and a total volume change of approximately 12.39 × 106 m3. From the modeling results, a lava flow prediction model was generated using the QLavHA program. The approaches described in this study can be used by government officials, authorities, and other decision-makers to monitor and assess the risk of volcanic activity in the region.

High-resolution imaging coupled with deep learning model for classifying water body of Soyang Lake, South Korea

Suci Ramayanti,박성재,이창욱,박유철 한국지질과학협의회 2023 Geosciences Journal Vol.27 No.6

Precise monitoring of natural phenomena is essential to reduce potential risk when a disaster occurs. The capability of developing high-resolution satellite images has provided advantages in the Earth’s surface observation. High-resolution images, including KOMPSAT-2/3 and PlanetScope images, have enabled more precise observation of the Earth’s surface. Combining the high-resolution KOMPSAT-2/3 and PlanetScope images and a capable image classifier tool can contribute to improving the confidence level of the observation. Recently, deep learning models provided enhanced capability compared to traditional machine learning, especially in image classification. In this study, we aim to investigate performance comparison between deep learning model using U-Net architecture and traditional machine learning represented by support vector machine (SVM) and random forest (RF) model with a focus on identifying water bodies on a lake. Significant differences in surface area of water of Soyang Lake occurred in 2015 and 2022 was the area of interest of this study. Based on confusion matrix analysis, the best classification results were indicated by deep learning model represented by U-Net with overall accuracy ranging from 97% to 100% and a Kappa coefficient of 0.91–1.00. The two traditional machine learning models, SVM and RF, lower up to 11% in overall accuracy and 0.15 for Kappa coefficient values. We also estimated the surface area of water based on the best classification results and found that the changes in water surface area were correlated with the monthly precipitation with a coefficient correlation of 0.89. This study should give new insight into the classification method for high-resolution satellite images. By providing the precise classification result, it could be contributed to providing proper mitigation and design policy related to natural phenomena on the earth’s surface, especially for monitoring the surface area of the lake water.

코로나19 팬데믹 기간의 서울의 사회적 거리두기 단계 변화와 The Suomi National Polar-Orbiting Partnership (S-NPP) 위성 영상을 이용한 Nighttime Light (NTL) 간의 상관관계

( Arip Syaripudin Nur ),이슬기 ( Seulki Lee ),( Suci Ramayanti ),한주 ( Ju Han ) 대한원격탐사학회 2021 大韓遠隔探査學會誌 Vol.37 No.6

한국은 코로나19로 인한 감염전파를 줄이기 위해 4단계의 사회적 거리두기 기준을 설정하고 확진자 발생 비율을 기준으로 단계를 전환하여 시행하고 있다. 이러한 사회적 거리두기는 사람들의 이동 및 모임 등 사회적 접촉을 제한함으로써 시민들의 활동량에 변화를 가져왔다. 이를 직관적으로 확인할 수 있는 데이터 중 하나가 Night Time Light (NTL)이다. NTL은 인공위성에 포착된 불빛을 활용해 측정한 국가경제규모를 측정할 수 있는 변수로, 야간동안 사람의 사회 활동을 파악하는데 활용할 수 있다. NTL 자료는 수오미 위성(Suomi National Polar-orbiting Partnership, S-NPP)에 탑재된 센서인 Visible Infrared Imaging Radiometer Suite (VIIRS)에 포함된 Day-Night Band (DNB)를 통해 얻을 수 있다. 본 연구는 2019년 1월 5일부터 2021년 10월 26일까지 1023개의 Suomi 자료를 수집하고, 서울의 NTL 변화를 시계열로 생성하여 사회적 거리두기 단계와의 상관관계를 분석하였다. 그 결과 사회적 거리두기의 단계가 높아질수록 NTL의 공간적, 시간적 변화가 모두 감소된 것으로 나타났다. 이는 더 높은 단계의 사회적 거리두기 정책이 실행됨에 따라 야간 시간대의 상업 활동 및 모임 인원 제한 등과 같은 사회적 활동의 제한이 실제로 서울의 NTL 감소에 영향을 준 것으로 해석할 수 있다. 본 연구는 향후 코로나19 관련 정부의 정책을 평가하고 개선하기 위한 참고자료로 활용할 수 있을 것이다. In order to reduce the spread of infection due to COVID-19, South Korea has established a four-step social distancing standard and implemented it by changing the steps based on the rate of confirmed cases. The implementation of social distancing brought about a change in the amount of activity of citizens by limiting social contact such as movement and gathering of people. One of the data that can intuitively confirm this is Night Time Light (NTL). NTL is a variable that can measure the size of the national economy measured using lights captured by satellites, and can be used to understand people’s social activities during the night. The NTL visible data is obtained via the Visible Infrared Imaging Radiometer Suite (VIIRS) Day-Night Band (DNB) onboard the Suomi National Polar-orbiting Partnership (S-NPP) satellite. 1023 of Suomi data from 1 January 2019 until 26 October 2021 were collected to generate time series of NTL radiance change over Seoul to analyze the correlation with social distancing policy. The results show that implementing the level of social distancing generally decreased the NTL radiance both in spatial disparities and temporal patterns. The higher level of policy, limiting human activities combined with the low number of people who have been vaccinated and the closure of various facilities. Because of social distancing, the differences in human activities affected the nighttime light during the COVID-19 pandemic, especially in Seoul, South Korea. Therefore, this study can be used as a reference for the government in evaluating and improving policies related to efforts reducing the transmission of COVID-19.

Interferometric SAR observation of the 2021 Mount Nyiragongo Eruption In Congo using Sentinel-1 data

Arip Syaripudin Nur,Muhammad Fulki Fadhillah,Suci Ramayanti,Chang-Wook Lee 대한지질학회 2021 대한지질학회 학술대회 Vol.2021 No.10

On May 22, 2021, Mount Nyiragongo erupted in the Democratic Republic of the Congo, threatening the nearby city of Goma and neighboring towns. As a result, 32 people were reported killed, and more than 400,000 people were forced to be evacuated. Therefore, monitoring the 2021 Nyiragongo eruption is required, including mapping surface deformation, lava flows, and damaged buildings. The use of point-based interferometric synthetic aperture radar (InSAR) time series analysis continues to grow, such as PSI and StaMPS. However, persistent scatter, which is more sensitive to urban and rocky conditions, has limitations in applying mountainous areas with vegetated areas. In this study, we used an enhanced combination method of interferometric scattering with the optimal point scattering method (ICOPS) after the eruption of Mount Nyiragongo to generate time series deformations and mean deformation maps. SAR data from the Sentinel-1 satellite for 2020-2021 were used in this study. For comparison, we used the StaMPS method based on persistent scatterer to study the deformation of Mount Nyiragongo. And the initial results show an increase in the coverage of measurement points (MP) using the ICOPS method. In addition, the use of optimized hotspot analysis (OHSA) on MP can provide additional insight in deformation analysis by exploiting the advantages of spatial clustering on MP. This study is an initial overview showing promising performance for the application of ICOPS in deformation studies. In addition, we generated a damage proxy map (DPM) derived from temporal changes in the InSAR coherence to identify anomalous changes that indicated earthquake damage. The DPM shows a loss of coherence in the urban area near Goma airport, which may have been caused by several earthquakes after the eruption. Compared to image analysis from UNOSAT, DPM revealed areas damaged by lava flows south of Mount Nyiragongo. Rapid assessment after an eruption followed by an earthquake can help with evacuation planning and mitigation to reduce disaster risk.

Surface Deformation Analysis using a Novel Application of Improved Combined Scatterers Interferometry with Optimized Point Scatterers (ICOPS)

Muhammad Fulki Fadhillah,Wahyu Luqmanul Hakim,Suci Ramayanti,Chang-Wook Lee 대한지질학회 2021 대한지질학회 학술대회 Vol.2021 No.10

Interferometric synthetic aperture radar (InSAR) analysis has received much attention recently in the analysis of surface deformation changes. The development of the InSAR time-series continues to progress by maximizing the use of machine learning and statistical methods to obtain satisfactory results. The combination of several methods can maximize the results of time-series analysis which is still plagued by several limitations. The use of time-series analysis based on point scatterers such as Stanford Method for Persistent Scatterers (StaMPS) is more suitable in areas with urban characteristics where there are many areas with stable phases that produce persistent scatterers. However, it has limitations in its application to areas with vegetation cover or mountains. Based on these limitations, we tried to develop an improved combined scatterers interferometry for optimized point scatterers (ICOPS) method. This method combines the use of distributed scatterer (DS) and PS points to increase the coverage of the analysis. In addition, we also apply the use of the SVR algorithm to select the optimum point in the deformation dataset that has been obtained in the previous process. In order to improve the reliability of the deformation mapping, we used the Optimized HotSpot analysis method to obtain a spatially clustered deformation map. The Yellowstone National Park was selected as the study area to observe deformation during 2017-2019 using SAR data from Sentinel-1 satellite. As the result, t In addition, we also apply the use of the SVR algorithm to select the optimum point in the deformation dataset that has been obtained in the previous process. In order to improve the reliability of the deformation mapping, we used the Optimized HotSpot analysis method to obtain a spatially clustered deformation map. The Yellowstone National Park was selected as the study area to observe deformation during 2017-2019 using SAR data from Sentinel-1 satellite. As the result, t In addition, we also apply the use of the SVR algorithm to select the optimum point in the deformation dataset that has been obtained in the previous process. In order to improve the reliability of the deformation mapping, we used the Optimized HotSpot analysis method to obtain a spatially clustered deformation map. The Yellowstone National Park was selected as the study area to observe deformation during 2017-2019 using SAR data from Sentinel-1 satellite. As the result, t The Yellowstone National Park was selected as the study area to observe deformation during 2017-2019 using SAR data from Sentinel-1 satellite. As the result, t The Yellowstone National Park was selected as the study area to observe deformation during 2017-2019 using SAR data from Sentinel-1 satellite. As the result, the CSI method showed an increase in MP density especially in areas not covered by PSI-StaMPS, which showed an improvement in detecting deformation in mountainous areas. We also evaluate the time series result with GPS site data and reveal the accuracy of this method is ~1cm/year in term root mean square error (RMSE). Also, the usefulness of OHSA method give advantages on spatial clustering on the MP to distinguish the deformation. Furthermore, these result still on developing process which is still need a lot improvement and application to achieve the reliable result. The development of this method can be use as the basis for increasing interest in the post-time series optimization process based on persistent distribution as a surface change detection tool.

위성영상을 활용한 2023년 소양호 녹조 현상 관측 및 분석

박성재 ( Sungjae Park ),이슬기 ( Seulki Lee ),( Suci Ramayanti ),박은석 ( Eunseok Park ),이창욱 ( Chang-wook Lee ) 대한원격탐사학회 2023 大韓遠隔探査學會誌 Vol.39 No.5

본 연구에서는 1973년 완공된 소양호에 처음으로 발생한 녹조현상에 대하여 위성영상을 사용하여 분석하였다. 연구자료는 2023년 7월부터 약 2개월간의 광학영상 13장을 사용하였으며, 소양호에 발생한 녹조의 면적을 산출하였다. 정확한 녹조 발생 면적을 산출하기 위하여 support vector machine 알고리즘 기반으로 영상분류를 수행하였다. 그 결과 소양호의 녹조는 녹조 발생을 유발하게 한 불순물이 유입된 지점을 중심으로 발생하였다. 2023년 8월 태풍 카눈의 효과로 일시적으로 감소하는 듯 보였으나 지속된 더위로 인해 다시 녹조가 증가하였다. 본 연구결과는 소양호는 수도권 주요 수원지 중 하나로 반복적인 녹조 발생을 대비해야 하는 점을 시사한다. In this study, we used satellite images to analyze the green algae phenomenon that first occurred in Soyang-ho, which was completed in 1973. The research data used 13 optical images over a period of about 2 months from July 2023, and the area of green algae that occurred in Soyang-ho was calculated. To calculate the exact area where green algae occurred, image classification was performed based on the support vector machine algorithm. As a result, green algae in Soyang-ho occurred around the point where the impurities that caused the green algae were introduced. It seemed to temporarily decrease due to the effects of Typhoon Khanun in August 2023, but green algae increased again due to the continued heat. Soyang-ho is one of the major water sources in the metropolitan area, suggesting that we must prepare for repeated green algae outbreaks.