Application of convolutional autoencoder for spatiotemporal bias-correction of radar precipitation

Sungho Jung; Sungryul Oh; Daeeop Lee; Xuan Hien Le; Giha Lee

doi:10.3741/JKWRA.2021.54.7.453

All Issue

2021 Vol.54, Issue 7 Next Page

Research Article

Application of convolutional autoencoder for spatiotemporal bias-correction of radar precipitation CAE 알고리즘을 이용한 레이더 강우 보정 평가

31 July 2021. pp. 453-462

PDF XML

Abstract

As the frequency of localized heavy rainfall has increased during recent years, the importance of high-resolution radar data has also increased. This study aims to correct the bias of Dual Polarization radar that still has a spatial and temporal bias. In many studies, various statistical techniques have been attempted to correct the bias of radar rainfall. In this study, the bias correction of the S-band Dual Polarization radar used in flood forecasting of ME was implemented by a Convolutional Autoencoder (CAE) algorithm, which is a type of Convolutional Neural Network (CNN). The CAE model was trained based on radar data sets that have a 10-min temporal resolution for the July 2017 flood event in Cheongju. The results showed that the newly developed CAE model provided improved simulation results in time and space by reducing the bias of raw radar rainfall. Therefore, the CAE model, which learns the spatial relationship between each adjacent grid, can be used for real-time updates of grid-based climate data generated by radar and satellites.

Keywords

Radar rainfall

Bias-correction

Convolutional autoencoder

Deep learning

최근 몇 년 동안 국지성 집중호우의 빈도가 증가함에 따라 고해상도 레이더 자료의 중요성 및 활용성이 증가하고 있다. 하지만 여전히 레이더 자료의 경우 시·공간적 편의가 존재하고 이를 보정하는 것이 매우 중요하며 많은 연구에서 레이더 강우의 편의 보정을 위해 다양한 통계적 기법이 시도되었다. 본 연구에서는 시·공간적으로 강우를 추정할 수 있는 이중편파레이더의 편의를 지점 강우와 비교하여 보정하는 것을 목표로 한다. 환경부의 수자원관리 및 홍수 예측에 사용되는 S-밴드 이중편파레이더의 편의 보정을 위하여 합성곱신경망(Convolutional Neural Network, CNN)기반의 Convolutional Autoencoder (CAE) 알고리즘을 구축하여 편의 보정을 수행하였다. CAE 모델의 입력자료는 환경부의 10분 단위 레이더 합성 강우자료와 같은 공간해상도로 보간된 지점 관측 강우자료를 사용하였으며, 자료의 기간은 미호천 유역에 홍수 경보가 발령된 2017년 7월 16일 00시부터 13시까지의 10분 단위 자료를 사용하였다. 그 결과로 지점 강우 대비 원시 레이더 강우의 편의가 줄어듦을 확인할 수 있으며 시·공간적으로 개선된 결과를 보여주고 있다. 따라서 각 인접한 격자 간의 공간 관계를 학습하는 CAE 모델은 레이더 및 위성에서 추정되는 격자형 기후 자료의 실시간 편의 보정에 사용할 수 있을 것으로 분석되었다.

키워드

레이더 강우

편의 보정

합성곱 오토인코더

딥러닝

References

Chumchean, S., Seed, A., and Sharma, A. (2006). “Correcting of real-time radar rainfall bias using a Kalman filtering approach.” Journal of Hydrology, Vol. 317, pp. 123-137. 10.1016/j.jhydrol.2005.05.013

Karimpouli, S., and Tahmasebi, P. (2019). “Segmentation of digital rock images using deep convolutional autoencoder networks.” Computers and Geosciences, Vol. 126, pp. 142-150. 10.1016/j.cageo.2019.02.003

Kim, H.L., Park, H.S., and Ko, J.S. (2014). “Improvement of rainfall estimation according to the calibration bias of dual-polarimetric radar variables.” Journal of Korea Water Resources Association, Vol. 47, No. 12, pp. 1227-1237. 10.3741/JKWRA.2014.47.12.1227

Kim, S.J., Kwon, Y.S., Lee, G.H., and Kim, H.S. (2010). “Radar rainfall adjustment by artificial neural network and runoff analysis.” Journal of the Korean Society of Civil Engineers B, Vol. 30, No. 2B, pp. 159-167.

Kim, T.J., Lee, D.R., Jang, S.M., and Kwon, H.H. (2017). “Generation of radar rainfall data for hydrological and meteorological application (I): Bias correction and estimation of error distribution.” Journal of Korea Water Resources Association, Vol. 50, No. 1, pp. 1-15.

Kwon, S.H., Lee, G.W., and Kim, G.S. (2015). “Rainfall estimation from an operational S-band dual-polarization radar: Effect of radar calibration.” Journal of the Meteorological Society of Japan, Vol. 93, No. 1, pp. 65-79. 10.2151/jmsj.2015-005

Le, X.H., Lee, G.H., Jung, K.S., An, H.U., and Lee, S.S. (2020). “Application of convolutional neural network for spatiotemporal bias correction of daily satellite-based precipitation.” Remote Sensing, Vol. 12, No. 17, 2731. doi: 10.3390/rs12172731 10.3390/rs12172731

Lee, C.K. (2012). “The present and future of radar rainfall estimation.” Water for Future, Vol. 12, No. 8, pp. 15-27.

Masci, J., Meier, U., Cireşan, D., and Schmidhuber, J. (2011). “Stacked convolutional auto-encoders for hierarchical feature extraction.” International Conference on Artificial Neural Networks, Springer, Berlin, Heidelberg, pp. 52-59. 10.1007/978-3-642-21735-7_7

Ministry of Environment (ME) (2019). Press release, assessed 29 October 2019, <http://www.me.go.kr/>.

Na, W.Y., and Yoo, C.S. (2020). “Real-time bias correction of Beaslesan dual-pol radar rain rate using the dual Kalman filter.” Journal of Korea Water Resources Association, Vol. 53, No. 3, pp. 201-214.

Noh, H.S., Kang, N.R., Kim, B.S., and Kim, H.S. (2012). “Flood simulation using Vflo and radar rainfall adjustment data by statistical objective analysis.” Journal of Wetlands Research, Vol. 14, No. 2, pp. 243-254.

Park, N.J., and Ko, H.S. (2020). “Agglomerative hierarchical clustering analysis with deep convolutional autoencoders.” Journal of Korea Multimedia Society, Vol. 23, No. 1, pp. 1-7.

Ryzhkov, A.V., Giangrande, S.E., and Schuur, T.J. (2005). “Rainfall estimation with a polarimetric prototype of WSR-88D.” Journal of Applied Meteorology, Vol. 44, No. 4, pp. 502-515. 10.1175/JAM2213.1

Yang, J.K. (2012). “The present and future of meteorological radar.” Water for Future, Vol. 12, No. 8, pp. 40-43.

You, C.H., Kang, M.Y., Lee, D.I., and Lee, J.T. (2016). “Approaches to radar reflectivity bias correction to improve rainfall estimation in Korea.” Atmospheric Measurement Techniques, Vol. 9, No. 5, pp. 2043-2053. 10.5194/amt-9-2043-2016

Information

Publisher :KOREA WATER RESOURECES ASSOCIATION
Publisher(Ko) :한국수자원학회
Journal Title :Journal of Korea Water Resources Association
Journal Title(Ko) :한국수자원학회 논문집
Volume : 54
No :7
Pages :453-462
Received Date : 2021-04-29
Revised Date : 2021-05-18
Accepted Date : 2021-05-18
DOI :https://doi.org/10.3741/JKWRA.2021.54.7.453

Journal of Korea Water Resources Association 한국수자원학회 논문집 ISSN:2799-8746(Print) 2799-8754(Online)

All Issue