KSCE Journal of Civil and Environmental Engineering Research (대한토목학회논문집)

Korean Society of Civil Engeneers (대한토목학회)
권호 표지
DOI QR코드

DOI QR Code

Mapping Burned Forests Using a k-Nearest Neighbors Classifier in Complex Land Cover

k-Nearest Neighbors 분류기를 이용한 복합 지표 산불피해 영역 탐지

  • 이한나 (강릉원주대학교 방재연구소) ;
  • 윤공현 (연세대학교 공학연구원) ;
  • 김기홍 (강릉원주대학교 건설환경공학과)
  • Received : 2023.07.27
  • Accepted : 2023.08.19
  • Published : 2023.12.01
Download PDF
⟨ Previous Next ⟩

Abstract

As human activities in Korea are spread throughout the mountains, forest fires often affect residential areas, infrastructure, and other facilities. Hence, it is necessary to detect fire-damaged areas quickly to enable support and recovery. Remote sensing is the most efficient tool for this purpose. Fire damage detection experiments were conducted on the east coast of Korea. Because this area comprises a mixture of forest and artificial land cover, data with low resolution are not suitable. We used Sentinel-2 multispectral instrument (MSI) data, which provide adequate temporal and spatial resolution, and the k-nearest neighbor (kNN) algorithm in this study. Six bands of Sentinel-2 MSI and two indices of normalized difference vegetation index (NDVI) and normalized burn ratio (NBR) were used as features for kNN classification. The kNN classifier was trained using 2,000 randomly selected samples in the fire-damaged and undamaged areas. Outliers were removed and a forest type map was used to improve classification performance. Numerous experiments for various neighbors for kNN and feature combinations have been conducted using bi-temporal and uni-temporal approaches. The bi-temporal classification performed better than the uni-temporal classification. However, the uni-temporal classification was able to detect severely damaged areas.

인간 활동 영역이 산지 곳곳에 퍼져 있는 한국에서는 산불이 주거지역이나 각종 시설물을 위협하는 경우가 잦다. 따라서 산불 이후 대책 마련과 피해 복구를 위해 피해 범위를 빠르게 파악할 필요가 있으며, 이러한 경우 원격탐사가 유용한 도구가 될 수 있다. 본 연구에서는 2019년 4월에 발생한 고성·속초 산불 피해지역에 k-nearest neighbor (kNN) 알고리즘을 적용하여 피해 범위를 탐지하는 실험을 수행하였다. 다양한 인공지물을 포함하는 지표와 숲이 혼재된 지역 특성을 고려하여 적절한 공간 해상도와 시간 해상도를 제공하는 Sentinel-2 multispectral instrument (MSI) 자료를 사용하였다. Sentinel-2 MSI의 여섯 밴드와 정규식생지수(NDVI), 정규탄화지수(NBR)를 분류 특성으로 사용하였다. 산불 피해지역과 비피해 지역에서 무작위로 추출된 2,000개 지점 정보를 이용하여 kNN 분류기를 훈련시켰다. 분류 성능을 높이기 위해 데이터에서 특이값을 제거하고 임상도를 병용하였다. 다양한 이웃(neighbor) 수와 분류 특성 조합을 적용하여 산불 후 데이터를 이용한 실험과 산불 전후 데이터 차이를 이용한 실험을 수행하였다. 산불 전후 데이터 차이를 이용하였을 때 더 우수한 분류 성과를 얻을 수 있었지만, 산불 후 데이터만을 이용한 경우에도 피해지역의 범위를 파악할 수 있었다.

Keywords

Acknowledgement

This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (No. 2021R1A6A1A03044326) and the Ministry of Science and ICT (No. RS-2023-00252622).

References

  1. Alberg, A. J., Park, J. W., Hager, B. W., Brock, M. V. and Diener-West, M. (2004). "The use of 'overall accuracy' to evaluate the validity of screening or diagnostic tests." Journal of General Internal Medicine, Springer, Vol. 19, No. 5, pp. 460-465, https://doi.org/10.1111/j.1525-1497.2004.30091.x. 
  2. Bar, S., Parida, B. R. and Pandey, A. C. (2020). "Landsat-8 and Sentinel-2 based Forest fire burn area mapping using machine learning algorithms on GEE cloud platform over Uttarakhand, Western Himalaya." Remote Sensing Applications: Society and Environment, Elsevier, Vol. 18, 100324, https://doi.org/10.1016/j.rsase.2020.100324. 
  3. Choi, S. P., Kim, D. H. and Lee, S. K. (2006). "The abstaction of forest fire damage area using factor analysis from the satellite image data." Journal of Korean Society for Geospatial Information System, KSIS, Vol. 14, No. 1, pp. 13-19. 
  4. Chung, M. and Kim, Y. (2020). "Analysis on Topographic Normalization Methods for 2019 Gangneung-East Sea Wildfire Area Using PlanetScope Imagery." Korean Journal of Remote Sensing, KSRS, Vol. 36, No. 2_1, pp. 179-197, https://doi.org/10.7780/KJRS.2020.36.2.1.7. 
  5. Cohen, J. (1960). "A coefficient of agreement for nominal scales." Educational and Psychological Measurement, Sage, Vol. 20, No. 1, pp. 37-46, https://doi.org/10.1177/001316446002000104. 
  6. Cover, T. and Hart, P. (1967). "Nearest neighbor pattern classification." IEEE Transactions on Information Theory, IEEE, Vol. 13, No. 1, pp. 21-27, https://doi.org/10.1109/TIT.1967.1053964. 
  7. Dennison, P. E., Brewer, S. C., Arnold, J. D. and Moritz, M. A. (2014). "Large wildfire trends in the western United States, 1984-2011." Geophysical Research Letters, AGU, Vol. 41, No. 8, pp. 2928-2933, https://doi.org/10.1002/2014GL059576. 
  8. European Space Agency (2015). Sentinel-2 User Handbook. European Space Agency. 
  9. Fix, E. and Hodges, J. L. (1989). "Discriminatory analysis. nonparametric discrimination: Consistency properties." International Statistical Review / Revue Internationale de Statistique, ISI, Vol. 57, No. 3, pp. 238-247, https://doi.org/10.2307/1403797. 
  10. Fornacca, D., Ren, G. and Xiao, W. (2018). "Evaluating the best spectral indices for the detection of burn scars at several post-fire dates in a mountainous region of Northwest Yunnan, China." Remote Sensing, MDPI, Vol. 10, No. 8, 1196, https://doi.org/10.3390/rs10081196. 
  11. Gascon, F., Bouzinac, C., Thepaut, O., Jung, M., Francesconi, B., Louis, J., Lonjou, V., Lafrance, B., Massera, S., Gaudel-Vacaresse, A., Languille, F., Alhammoud, B., Viallefont, F., Pflug, B., Bieniarz, J., Clerc, S., Pessiot, L., Tremas, T., Cadau, E., De Bonis, R., Isola, C., Martimort P. and Fernandez, V. (2017). "Copernicus Sentinel-2A calibration and products validation status." Remote Sensing, MDPI, Vol. 9, No. 6, 584, https://doi.org/10.3390/rs9060584. 
  12. Hawbaker, T. J., Vanderhoof, M. K., Beal, Y.-J., Takacs, J. D., Schmidt, G. L., Falgout, J. T., Williams, B., Fairaux, N. M., Caldwell, M. K., Picotte, J. J., Howard, S. M., Stitt, S. and Dwyer, J. L. (2017). "Mapping burned areas using dense time-series of Landsat data." Remote Sensing of Environment, Elsevier, Vol. 198, pp. 504-522, https://doi.org/10.1016/j.rse.2017.06.027. 
  13. Kara, L. Z., Laksaci, A., Rachdi, M. and Vieu, P. (2017). "Data-driven kNN estimation in nonparametric functional data analysis." Journal of Multivariate Analysis, Elsevier, Vol. 153, pp. 176-188, https://doi.org/10.1016/j.jmva.2016.09.016. 
  14. Klebanov, L. B. (2016). "Big outliers versus heavy tails: What to use?" ArXiv:1611.05410 [Math, Stat], ArXiv, http://arxiv.org/abs/1611.05410. 
  15. Knopp, L., Wieland, M., Rattich, M. and Martinis, S. (2020). "A Deep learning approach for burned area segmentation with Sentinel-2 data." Remote Sensing, MDPI, Vol. 12, No. 15, 2422, https://doi.org/10.3390/rs12152422. 
  16. Landis, J. R. and Koch, G. G. (1977). "The measurement of observer agreement for categorical data." Biometrics, International Biometric Society, Vol. 33, No. 1, pp. 159-174, https://doi.org/10.2307/2529310. 
  17. Lee, S. J., Kim, K. J., Kim, Y. H., Kim, J. W. and Lee, Y. W. (2017). "Development of FBI(Fire Burn Index) for Sentinel-2 images and an experiment for detection of burned areas in Korea." Journal of the Association of Korean Photo-Geographers, The Association of Korean Photo-Geographers, Vol. 27, No. 4, pp. 187-202, https://doi.org/10.35149/JAKPG.2017.27.4.012 (in Korean). 
  18. Mithal, V., Nayak, G., Khandelwal, A., Kumar, V., Nemani, R. and Oza, N. (2018). "Mapping burned areas in tropical forests using a novel machine learning framework." Remote Sensing, MDPI, Vol. 10, No. 1, 69, https://doi.org/10.3390/rs10010069.
  19. National Fire agency (2021). 2020 Fire Statistical Yearbook, https://www.nfds.go.kr/bbs/selectBbsDetail.do?bbs=B21&bbs_no=7948&pageNo=1 (in Korean). 
  20. Nichols, T. R., Wisner, P. M., Cripe, G. and Gulabchand, L. (2010). "Putting the kappa statistic to use." The Quality Assurance Journal, Vol. 13, Nos. 3-4, pp. 57-61, https://doi.org/10.1002/qaj.481. 
  21. Nigsch, F., Bender, A., van Buuren, B., Tissen, J., Nigsch, E. and Mitchell, J. B. O. (2006). "Melting point prediction employing k-nearest neighbor algorithms and genetic parameter optimization." Journal of Chemical Information and Modeling, Vol. 46, No. 6, pp. 2412-2422, https://doi.org/10.1021/ci060149f. 
  22. Pinto, M. M., Libonati, R., Trigo, R. M., Trigo, I. F. and DaCamara, C. C. (2020). "A deep learning approach for mapping and dating burned areas using temporal sequences of satellite images." ISPRS Journal of Photogrammetry and Remote Sensing, Elsevier, Vol. 160, pp. 260-274, https://doi.org/10.1016/j.isprsjprs.2019.12.014. 
  23. Roteta, E., Bastarrika, A., Padilla, M., Storm, T. and Chuvieco, E. (2019). "Development of a Sentinel-2 burned area algorithm: Generation of a small fire database for sub-Saharan Africa." Remote Sensing of Environment, Elsevier, Vol. 222, pp. 1-17, https://doi.org/10.1016/j.rse.2018.12.011. 
  24. Roy, D. P., Huang, H., Boschetti, L., Giglio, L., Yan, L., Zhang, H. H. and Li, Z. (2019). "Landsat-8 and Sentinel-2 burned area mapping-A combined sensor multi-temporal change detection approach." Remote Sensing of Environment, Elsevier, Vol. 231, 111254, https://doi.org/10.1016/j.rse.2019.111254. 
  25. Sim, S., Kim, W., Lee, J., Kang, Y., Im, J., Kwon, C. and Kim, S. (2020). "Wildfire severity mapping using sentinel satellite data based on machine learning approaches." Korean Journal of Remote Sensing, KSRS, Vol. 36, No. 5_3, pp. 1109-1123, https://doi.org/10.7780/KJRS.2020.36.5.3.9 (in Korean). 
  26. Story, M. and Congalton, R. G. (1986). "Accuracy assessment: A user's perspective." Photogrammetric Egineering and Remote Sensing, American Society for Photogrammetry and Remote Sensing, Vol. 52, No. 3, pp. 397-399. 
  27. Weaver, J., Moore, B., Reith, A., McKee, J. and Lunga, D. (2018). "A comparison of machine learning techniques to extract human settlements from high resolution imagery." Proceedings of IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium, IEEE, Valencia, Spain, pp. 6412-6415, https://doi.org/10.1109/IGARSS.2018.8518528. 
  28. Weber, K. T., Seefeldt, S., Moffet, C. and Norton, J. (2008). "Comparing fire severity models from post-fire and pre/post-fire differenced imagery." GIScience & Remote Sensing, Taylor & Francis, Vol. 45, No. 4, pp. 392-405, https://doi.org/10.2747/1548-1603.45.4.392. 
  29. Won, M., Jang, K., Yoon, S. and Lee, H. (2019). "Change detection of damaged area and burn severity due to heat damage from gangwon large fire area in 2019." Korean Journal of Remote Sensing, KSRS, Vol. 35, No. 6_2, pp. 1083-1093, https://doi.org/10.7780/KJRS.2019.35.6.2.5 (in Korean). 
  30. Won, M. S., Koo, K. S. and Lee, M. B. (2007). "An quantitative analysis of severity classification and burn severity for the large forest fire areas using normalized burn ratio of landsat imagery." Journal of the Korean Association of Geographic Information Studies, KAGIS, Vol. 10, No. 3, pp. 80-92 (in Korean).