All Issue

2026 Vol.27, Issue 7 Preview Page
Estimation of Random Walk Model Parameters Using Global Optimization Technique

전역최적화 기법을 통한 RWM 최적 매개변수 추정

Chanul Choi1
,
Song Eu2
,
Mingeun Song3
,
Giha Lee4*
최 찬울1
,
유 송2
,
송 민근3
,
이 기하4*
1Ph.D Candidate, Department of Civil and Environmental Engineering, School of Convergence Science and Technology
2Research Official, Landslide Division, Department of Forest Disaster and Environment, National Institute of Forest Science
3Ph.D Candidate, Department of Civil and Environmental Engineering, School of Convergence Science and Technology
4Professor, Department of Civil and Environmental Engineering, School of Convergence Science and Technology
*Corresponding Author
1 July 2026. pp. 31-40
PDF XML Citation
Abstract

This study proposes an automated framework to estimate optimal parameters of the probabilistic Random Walk Model (RWM) for rapid debris-flow runout mapping. The workflow relies on minimal inputs: a 5 m resolution DEM, mapped debris-flow damage and initiation points, and a binary damage/non-damage map for evaluation. The method was applied to Samga-ri, Punggi-eup, Yeongju-si (Gyeongsangbuk-do, Republic of Korea), where a large debris-flow event occurred during heavy rainfall on July 15, 2023. To reduce the time burden and potential local-optimum bias of manual tuning, a Differential Evolution–based global optimization module was developed to calibrate VS, IW, and SS. The optimized parameters were VS=18.509, SS=6.849, and IW=2.822, yielding a mean IoU of 0.3228±0.0047, precision of 0.5942, and recall of 0.4361. The relatively low IoU and recall are attributed to the RWM structure, which evaluates outcomes mainly at final deposition cells rather than the full travel path. The mean deposition pattern was consistent with post-event aerial imagery, and the maximum deposition height was approximately 3.8 m. In addition, a pass-count-based pass probability analysis showed that simulated flows were concentrated along the observed damage corridor, indicating that flow-path tendencies can be inferred even with a simplified probabilistic model.

Keywords
Debris flow
Random walk model
Differential evolution
Hyperparameter tuning

본 연구는 토석류 피해범위를 신속하게 추정하기 위해 확률 기반 Random Walk Model(RWM)의 최적 매개변수(VS, IW, SS)를 전역최적화 기법으로 자동 추정하는 방법을 제안하였다. 입력자료는 5 m 해상도 DEM, 토석류 피해지역 및 시작 지점 자료, 피해/비피해 이진 분류 맵으로 구성하였다. 연구 대상지는 2023년 7월 15일 집중호우로 대규모 토석류 피해가 발생한 경상북도 영주시 풍기읍 삼가리 지역이며, 수동 보정의 시간 소요 및 국소 최적해 불확실성을 줄이기 위해 차분진화(Differential Evolution) 기반 최적화 모듈을 개발하였다. 최적화 결과 VS=18.509, SS=6.849, IW=2.822가 추정되었고, 평균 IoU 0.3228±0.0047, 정밀도 0.5942, 재현율 0.4361을 보였다. IoU와 재현율이 낮은 원인은 이동 경로를 누적하지 않고 최종 퇴적 셀 중심으로 평가되는 RWM 특성으로 해석되었다. 평균 퇴적량은 항공사진 기반 피해 양상과 유사했고 최대 퇴적 높이는 약 3.8 m였다. 또한 통과횟수 기반 통과확률 분석에서 실제 피해 경로를 따라 토석류 이동이 집중되는 경향이 확인되어, 단순 확률 모형에서도 흐름 경향 파악이 가능함을 제시하였다.

키워드
토석류
랜덤워크모델
차분진화
최적 매개변수 보정
References
1

Dalbey, K., Patra, A. K., Pitman, E. B., Bursik, M. I. and Sheridan, M. F. (2008), Input uncertainty propagation methods and hazard mapping of geophysical mass flows, Journal of Geophysical Research: Solid Earth, Vol. 113, pp. 1~16.

2

Hyman, D. M., Bevilacqua, A. and Bursik, M. I. (2019), Statistical theory of probabilistic hazard maps: Hazard boundary location distribution, Natural Hazards and Earth System Sciences, Vol. 19, pp. 1347~1363.

10.5194/nhess-19-1347-2019
3

Imamura, R. and Sugita, M. (1980), Study on simulation of debris depositing based on a random walk model, Journal of the Japan Society of Erosion Control Engineering, Vol. 32, No. 3, pp. 17~26.

4

Iverson, R. M. (1997), The physics of debris flows, Reviews of Geophysics, Vol. 35, No. 3, pp. 245~296.

10.1029/97RG00426
5

Iverson, R. M. and George, D. L. (2014), A depth-averaged debris-flow model that includes the effects of evolving dilatancy: I. Physical basis, Proceedings of the Royal Society A, Vol. 470, No. 2170, pp. 1~31.

10.1098/rspa.2013.0819
6

Iwamoto, Y., Imamura, R., Nakasuji, A. and Sugita, M. (1982), 8-direction Random Walk Model for sediment deposition location simulation, Shin-Sabo, No. 125, pp. 16~21

7

McDougall, S. (2017), Landslide runout analysis—Current practice and challenges, Canadian Geotechnical Journal, Vol. 54, No. 5, pp. 605~620

10.1139/cgj-2016-0104
8

Patra, A. K., Bauer, A. C., Nichita, C. C., Pitman, E. B., Sheridan, M. F., and Bursik, M. (2005), Parallel adaptive numerical simulation of dry avalanches over natural terrain, Journal of Volcanology and Geothermal Research, Vol. 139, No. 1~2, pp. 1~21.

10.1016/j.jvolgeores.2004.06.014
9

Pitman, E. B. and Le, L. (2005), A two-fluid model for avalanche and debris flows, Philosophical Transactions of the Royal Society A, Vol. 363, No. 1832, pp. 1573~1601.

10

Pudasaini, S. P. and Fischer, J.-T. (2020), A mechanical model for phase separation in debris flow, International Journal of Multiphase Flow, Vol. 129, 103292.

10.1016/j.ijmultiphaseflow.2020.103292
11

Spiller, E. T., Bayarri, M. J., Berger, J. O., Calder, E. S., Patra, A. K., and Pitman, E. B. (2014), Automating emulator construction for geophysical hazard maps, SIAM/ASA Journal on Uncertainty Quantification, Vol. 2, pp. 126~152.

10.1137/120899285
12

Storn, R. and Price, K. (1997), Differential evolution—A simple and efficient heuristic for global optimization over continuous spaces, Journal of Global Optimization, Vol. 11, pp. 341~359.

10.1023/A:1008202821328
13

Takahashi, T. (2007), Debris Flow: Mechanics, Prediction and Countermeasures, Taylor & Francis, London, UK.

14

Tsai, Y.-L., Irvin, J., Chundi, S., Ng, A. Y., Field, C. B. and Kitanidis, P. K. (2022), Improving debris flow evacuation alerts in Taiwan using machine learning, arXiv preprint, arXiv:2208.13027.

10.21203/rs.3.rs-2235385/v1
15

Yoon, H. J., Lee, S. W., Woo, C. S., Park, K. H., Kim, K. H., Lee, M. B. and Jeong, Y. H. (2011), A study on pilot production and nationwide expansion of debris-flow hazard maps, Research Report No. 11-24, Korea Forest Research Institute (In Korean).

16

Yuan, L., Liu, W., Zhai, J., Wu, S. F., Patra, A. K. and Pitman, E. B. (2018), Refinement on non-hydrostatic shallow granular flow model in a global Cartesian coordinate system, Computational Geosciences, Vol. 22, pp. 87~106.

10.1007/s10596-017-9672-x
17

Zhao, H., Amann, F. and Kowalski, J. (2021), Emulator-based global sensitivity analysis for flow-like landslide run-out models, Landslides, Vol. 18, pp. 3299~3314.

10.1007/s10346-021-01690-w
Information
  • Publisher :Korean Geo-Environmental Society
  • Publisher(Ko) :한국지반환경공학회
  • Journal Title :Journal of the Korean Geo-Environmental Society
  • Journal Title(Ko) :한국지반환경공학회 논문집
  • Volume : 27
  • No :7
  • Pages :31-40
  • Received Date : 2026-06-04
  • Revised Date : 2026-06-09
  • Accepted Date : 2026-06-25
  • DOI :https://doi.org/10.14481/jkges.2026.27.7.31
Journal Informaiton Journal of the Korean Geo-Environmental Society Journal of the Korean Geo-Environmental Society
Online Submission
  • NRF
  • KOFST
  • KISTI Current Status
  • KISTI Cited-by
  • orcid
Journal Informaiton Journal Informaiton - close