DOI QR코드

DOI QR Code

Application study of random forest method based on Sentinel-2 imagery for surface cover classification in rivers - A case of Naeseong Stream -

하천 내 지표 피복 분류를 위한 Sentinel-2 영상 기반 랜덤 포레스트 기법의 적용성 연구 - 내성천을 사례로 -

  • An, Seonggi (Deepartment of Land, Water and Environment Research, Korea Institute of Civil Engineering and Building Technology) ;
  • Lee, Chanjoo (International Cooperation & Public Relations Division, Korea Institute of Civil Engineering and Building Technology) ;
  • Kim, Yongmin (Geography Major of Department of Social Studies Education, Seoul National University) ;
  • Choi, Hun (Social Education, Chonnam National University)
  • 안성기 (한국건설기술연구원 수자원하천연구본부) ;
  • 이찬주 (한국건설기술연구원 대외협력홍보본부) ;
  • 김용민 (서울대학교 사회교육학과 지리전공) ;
  • 최훈 (전남대학교 사회교육학과)
  • Received : 2024.03.18
  • Accepted : 2024.05.03
  • Published : 2024.05.31

Abstract

Understanding the status of surface cover in riparian zones is essential for river management and flood disaster prevention. Traditional survey methods rely on expert interpretation of vegetation through vegetation mapping or indices. However, these methods are limited by their ability to accurately reflect dynamically changing river environments. Against this backdrop, this study utilized satellite imagery to apply the Random Forest method to assess the distribution of vegetation in rivers over multiple years, focusing on the Naeseong Stream as a case study. Remote sensing data from Sentinel-2 imagery were combined with ground truth data from the Naeseong Stream surface cover in 2016. The Random Forest machine learning algorithm was used to extract and train 1,000 samples per surface cover from ten predetermined sampling areas, followed by validation. A sensitivity analysis, annual surface cover analysis, and accuracy assessment were conducted to evaluate their applicability. The results showed an accuracy of 85.1% based on the validation data. Sensitivity analysis indicated the highest efficiency in 30 trees, 800 samples, and the downstream river section. Surface cover analysis accurately reflects the actual river environment. The accuracy analysis identified 14.9% boundary and internal errors, with high accuracy observed in six categories, excluding scattered and herbaceous vegetation. Although this study focused on a single river, applying the surface cover classification method to multiple rivers is necessary to obtain more accurate and comprehensive data.

하천 공간의 지표 피복 현황 파악은 하천 관리 및 홍수 재해 예방에 필수적이다. 기존 조사 방법은 전문가에 의한 식생 판독을 통한 식생도 작도 방법이나 식생지수를 활용하는 방법이 활용되어 왔으나, 역동적으로 변화하는 하천 환경을 반영하기에 한계가 있다. 이러한 배경에서 본 연구는 내성천을 대상으로 위성영상 자료를 활용한 랜덤 포레스트 기법을 활용하여 다수 연도의 하천 내 식생 분포를 파악하고, 적용성을 검토하였다. 원격탐사 자료 Sentinel-2 위성 영상을 사용하였으며, 지상 참값(ground truth)은 2016년 내성천 지표 피복 자료를 활용하였다. 랜덤 포레스트 머신러닝 알고리듬을 활용하여 미리 선정된 10개 샘플링 영역으로부터 분류군 별로 1,000개의 표본을 추출하여 훈련 및 검증하였으며, 민감도 분석, 연도별 지표 피복 분석, 정확도 분석을 통하여 적용성을 평가하였다. 연구 결과, 검증 자료 기반의 정확도는 85.1%로 나타났다. 트리 수, 샘플 수, 하천 구역에 대한 민감도 분석 결과, 각각 30개, 800개, 하류에서 효율성이 높았다. 지표 분류 유형은 6개 항목에서 높은 정확도를 보여 지표 피복 분류 결과가 실제 하천 환경을 잘 반영하는 것으로 나타났다. 정확도 분석 결과, 전체 샘플 중 14.9%의 경계오류와 내부오류를 확인하였으며, 지표 피복 분류 중 산발 식생과 초본 식생을 제외한 항목들은 높은 정확도를 보였다. 본 연구에서는 단일 하천을 대상으로 적용하였지만, 보다 정확하고 많은 자료의 구축을 위해서는 다수의 하천에 대해 지표 피복 분류 기법의 적용이 요구된다.

Keywords

Acknowledgement

이 논문은 한국건설기술연구원 주요사업(디지털뉴딜 기반 통합물관리 기술 융합 플랫폼(IWRM-K) 개발(202401-66))의 연구비 지원으로 수행되었습니다.

References

  1. An, H.Y., Na, S.I., Park, C.W., So, K.H., and Lee, K.D. (2018). "Atmospheric correction effectiveness analysis of reflectance and NDVI using multispectral satellite image." Korean Journal of Remote Sensing, Vol. 34, No. 6-1, pp. 981-996. 
  2. Auda, Y., Lundin, E.J., Gustafsson, J., Pokrovsky, O.S., Cazaurang, S., and Orgogozo, L. (2023). "A new land cover map of two watersheds under long-term environmental monitoring in the swedish arctic using Sentinel-2 Data." Water, Vol. 15, No. 18, 3311. 
  3. Breiman, L. (2001). "Random forests." Machine Learning, Vol. 45, pp. 5-32. doi: 10.1023/A:1010933404324. 
  4. Choi, S.U., Yoon, B.M., and Woo, H.S. (2005). "Effects of daminduced flow regime change on downstream river morphology and vegetation cover in the Hwang River, Korea." River Research and Applications, Vol. 21, No. 2-3, pp. 313-325. 
  5. European Space Agency (ESA) (2023). Sentinel-2 MSI Revisit and Co verage, accessed 12 October 2023, .
  6. Geerling, G., Penning, E., Rogers, C., Nic, C.V.D.V.S., Harezlak, V., Wilson, S., and Donchyts, G. (2021). Operational moni toring of floodplain vegetation using remote sensing. Authorea, Hoboken, NJ, U.S., pp. 1-16. 
  7. Gurnell, A. (2014). "Plants as river system engineers." Earth Surface Processes and Landforms, Vol. 39, No. 1, pp. 4-25. 
  8. Harezlak, V., Geerling, G.W., Rogers, C.K., Penning, W.E., Augustijn, D.C., and Hulscher, S.J.M.H. (2020). "Revealing 35 years of landcover dynamics in floodplains of trained lowland rivers using satellite data." River Research and Applications, Vol. 36, No. 7, pp. 1213-1221. 
  9. Johnson, W.C. (1994). "Woodland expansions in the Platte River, Nebraska: Patterns and causes." Ecological Monographs, Vol. 64, No. 1, pp. 45-84. 
  10. Jung, W.S., Kim, S.E., and Kim, Y.D. (2021). "Analysis of influential factors of cyanobacteria in the mainstream of Nakdong River using random forest." Journal of Wetlands Research, Vol. 23, No. 1. pp. 27-34.  https://doi.org/10.17663/JWR.2021.23.1.27
  11. Kim, J.Y., and Jang, D.H. (2015). "An applications of remote sensing data for the land surface characteristics analysis of ecological stream restoration area in Daejeon Stream." Journal of the Association of Korean Geographers, Vol. 4, No. 2, pp. 231-340. 
  12. Korea Institute of Construction Technology (KICT) (2013). Development of floodplain maintenance technology for enhancement of waterfront value, No. 2013-128, p. 197. 
  13. Lee, C.H., and Cho, K.H. (2023). "Development of a method for tracking sandbar formation by weir-gate opening using multi-spectral satellite imagery in the Geumgang River, South Korea." Ecology and Resilient Infrastructure, Vol. 10, No. 4, pp. 135-142.  https://doi.org/10.17820/ERI.2023.10.4.135
  14. Lee, C.H., Kim, H.R., and Cho, K.H. (2022). "Spatial distribution and successional changes of riparian vegetation on sandbars exposed after watergate-opening of weirs in the Geumgang River, South Korea." Ecology and Resilient Infrastructure, Vol. 9, No. 3, pp. 194-205.  https://doi.org/10.17820/ERI.2022.9.3.194
  15. Lee, C.J., and Kim, D.G. (2018). "Analysis of the changes of the vegetated area in an unregulated river and their underlying causes: A case study on the Naeseong Stream." Ecology and Resilient Infrastructure, Vol. 5, No. 4, pp. 229-245. 
  16. Lee, C.J., Choi, H., Kim, D.G., Van Oorschot, M., Penning, E., Geerling, G.W. (2023). "Bio-geomorphic alteration through shifting flow regime in a modified monsoonal river system in Korea." River Research and Applications. Vol. 39, pp. 1639-1651. 
  17. Lee, C.J., Go, D.H., Kim, D.G., and Choi, H. (2022). "Quantitative analysis on the lateral migration of an alluvial river - Case study on Gopyeong site of Naeseong Stream." Journal of the Korean Geo-morphological Association, Vol. 29, No. 2, pp. 25-39. 
  18. Lee, C.J., Kim, D.G., Hwang, S.Y., Kim, Y.J., Jeong, S.J., Kim, S.N., and Cho, H.J. (2019a). "Dataset of long-term monitoring on the change in hydrology, channel morphology, landscape and vegetation along the Naeseong Stream (II)." Ecology and Resilient Infrastructure, Vol. 6, No. 1, pp. 34-48. 
  19. Lee, C.J., Kim, D.G., Ji, U., and Kim, J.S. (2019b). "Dataset of long-term monitoring on the change in hydrology, channel morphology, landscape and vegetation along the Naeseong Stream (I)." Ecology and Resilient Infrastructure, Vol. 6, No. 1, pp. 23-33.  https://doi.org/10.17820/ERI.2019.6.1.023
  20. Lee, G.R. (2010). "A comparative analysis on channel forms and landscapes at Naeseongcheon River and Wicheon River in Gyeongpook Province." Journal of the Korean Geomorphological Association, Vol. 17, No. 1, pp. 1-16. 
  21. Nguyen, T.T.H., Doan, M.T., Tomppo, E., and McRoberts, R.E. (2020). "Land use/land cover mapping using multi-temporal Sentinel-2 imagery and four classification methods - A case study from Dak Nong, Vietnam." Remote Sensing. Vol. 12, No. 9, 1367. 
  22. Park, E.J., Park, J.H., and Kim, H.H. (2019). "Mapping speciesspecific optimal plantation sites using random forest in Gyeongsangnam-do Province, South Korea." Journal of Agriculture & Life Science, Vol. 53, No. 6, pp. 65-74. 
  23. Rood, S.B., Kaluthota, S., Philipsen, L.J., Slaney, J., Jones, E., Chasmer, L., and Hopkinson, C. (2019). "Camo-maps: An efficient method to assess and project riparian vegetation colonization after a major river flood." Ecological Engineering, Vol. 141, 105610. 
  24. Son, S.Y., Kim, J.H., and Kim, Y.J. (2022). "Google Earth Engine-based satellite image processing and ground vegetation index extraction technique." Journal of Korea Water Resources Association, Vol. 55, No. 9, pp. 46-53. 
  25. Tang, H., Lu, S., Ali Baig, M. H., Li, M., Fang, C., and Wang, Y. (2022). "Large-scale surface water mapping based on landsat and sentinel-1 images." Water, Vol. 14, No. 9, 1454. 
  26. The Korea Institute of Public Administration (KIPA) (2023). ISSUE PAPER-2023 analysis of disaster management issues in concentrated rainfall: Focusing on the on-site situation propagation system, No. 2023-132. 
  27. Williams, G. P. (1978). The case of the shrinking channels the North Platte and Platte Rivers in Nebraska. U.S. Geological Survey Circular,781, U.S. Geological Survey, Reston, VA, U.S. pp.1-48. 
  28. Woo, H.S., Park, M.H., Cho, K.H., Cho, H.J., and Jeong, S.J. (2010). "Recruitment and succession of riparian vegetation in alluvial river regulated by upstream dams - Focused on the Nakdong River downstream Andong and Imha Dams -." Journal of Korea Water Resources Association, Vol. 43, No. 5, pp. 455-469. 
  29. Zanetti, C., Macia, J., Liency, N., Vennetier, M., Meriaux, P., and Provensal, M. (2016). "Roles of the riparian vegetation: the antagonism between flooding risk and the protection of environment." 3rd European Conference on Flood Risk Management (FLOODrisk 2016), Lyon, France, p. 13015.