DOI QR코드

DOI QR Code

기계학습을 활용한 하절기 기온 및 폭염발생여부 예측

Prediction of Temperature and Heat Wave Occurrence for Summer Season Using Machine Learning

  • 투고 : 2020.04.15
  • 심사 : 2020.06.29
  • 발행 : 2020.06.30

초록

최근 국내에서는 다양한 이상기후들이 발생하고 있으며 이로 인해 인명피해, 재산피해와 같은 큰 피해들이 발생하고 있다. 그 중에서도 폭염으로 인한 피해는 점점 증가하는 경향을 보인다. 이에 대처하기 위해서는 빠르고 정확한 기온 및 폭염발생여부 예측이 필수적이다. 현재 기상청에서는 폭염에 대한 정보를 단기예보를 통해 제공하는데, 단기예보를 위한 기온예측은 수치예보모델을 통해 수행된다. 과거 15년간(1998~2012년) 인구대비 폭염 사망률이 가장 높았던 ◯◯군에 대하여 2019년도 기온 예보자료와 관측 자료를 비교한 결과, 평균제곱근오차가 1.57℃ 발생하였고, 관측 값이 33℃이상에 해당하는 데이터만 비교한 결과, 평균제곱근오차가 1.96℃ 발생하였다. 예보시간은 4시간이고 예보과정에는 약 3~4시간이 소요된다. 이에 본 연구에서는 소요시간과 예측 정확도를 고려하여, 기계학습방법의 일종인 LSTM을 이용한 기온 및 폭염발생 예측 방법론을 제시한다. 기계학습모델을 이용한 4시간 기온예측결과 1.71℃의 평균제곱근오차가 발생하였고, 관측 값이 33℃ 이상에 해당하는 데이터만 비교한 결과 1.39℃의 평균제곱근오차가 발생하였다. 전 범위의 오차는 수치예보모델이 더 작은 값을 가지지만, 33℃이상의 경우에는 기계학습모델 예측의 정확도가 더 높았다. 또한 수치예보를 이용한 경우 예상 소요시간이 4시간가량인 반면 기계학습을 이용한 기온예측에는 평균 9분26초의 시간이 소요되어 경제적이라 판단하였다. 향후 공간적인 범위를 확대하거나 대상 지역을 변경하는 일반적인 방안에 대해서 연구를 수행하고자 한다.

Climate variations have become worse and diversified recently, which caused catastrophic disasters for our communities and ecosystem including economic property damages in Korea. Heat wave of summer season is one of causes for such damages of which outbreak tends to increase recently. Related short-term forecasting information has been provided by the Korea Meteorological Administration based on results from numerical forecasting model. As the study area, the ◯◯ province was selected because of the highest mortality rate in Korea for the past 15 years (1998~2012). When comparing the forecasted temperatures with field measurements, it showed RMSE of 1.57℃ and RMSE of 1.96℃ was calculated when only comparing the data corresponding to the observed value of 33℃ or higher. The forecasting process would take at least about 3~4 hours to provide the 4 hours advanced forecasting information. Therefore, this study proposes a methodology for temperature prediction using LSTM considering the short prediction time and the adequate accuracy. As a result of 4 hour temperature prediction using this approach, RMSE of 1.71℃ was occurred. When comparing only the observed value of 33℃ or higher, RMSE of 1.39℃ was obtained. Even the numerical prediction model of the whole range of errors is relatively smaller, but the accuracy of prediction of the machine learning model is higher for above 33℃. In addition, it took an average of 9 minutes and 26 seconds to provide temperature information using this approach. It would be necessary to study for wider spatial range or different province with proper data set in near future.

키워드

참고문헌

  1. Ahn, J. and Jeong, C. (2018). Numerical Simulation of the Flood Event Induced Temporally and Spatially Concentrated Rainfall - On August 17, 2017, the Flood Event of Cheonggyecheon. Journal of Korean Society of Disaster and Security. 11(2): 45-52. https://doi.org/10.21729/KSDS.2018.11.2.45
  2. Ahn, S. (2016). Deep Learning Architectures and Applications. Journal of Intelligence and Information Systems. 22(2): 127-142. https://doi.org/10.13088/jiis.2016.22.2.127
  3. Choi, J. (2019). Proposal of Early-Warning Criteria for Highway Debris Flow Using Rainfall Frequency (1): Proposal of Rainfall Criteria. Journal of Korean Society of Disaster and Security. 12(2): 1-13. https://doi.org/10.21729/KSDS.2019.12.2.1
  4. Choi, M. H. and Yun, J. I. (2009). On Recent Variations in Solar Radiation and Daily Maximum Temperature in Summer. Korean Journal of Agricultural and Forest Meteorology. 11(4): 185-191. https://doi.org/10.5532/KJAFM.2009.11.4.185
  5. Joints of Related Ministries (2018) Abnormal Weather Report. Seoul: Korea Meteorological Administration.
  6. Kim, J., Lee, D. G., Park, I. S., Choi, B. C., and Kim, J. S. (2006). Influences of Heat Waves on Daily Mortality in South Korea. Atmosphere. 16(4): 269-278.
  7. Korea Meteorological Administration (2018). Evaluation of Contribution of Meteorological Observation Data to Weather Forecasts. Seoul: Korea Meteorological Administration.
  8. Korea Meteorological Administration (2019). Forecast Services Regulations APPENDUM Article17. Seoul: Korea Meteorological Administration.
  9. Korea Meteorological Administration (2019). A Study on the Diagnosis and Development Direction of the Forecasting System. Seoul: Korea Meteorological Administration.
  10. Lee, S. G., Jung, S. G., Lee, W. S., and Park, G. H. (2011). A Predictive Model for Urban Temperature Using the Artificial Neural Network. Korea Planners Association. 46(1): 129-142.
  11. Park, J.E., Heo, B.Y., and Sunwoo, Y. (2016). A Study on Human Damage due to Heat Wave by Region. Journal of the Korean Society of Hazard Mitigation. 16(1): 103-109. https://doi.org/10.9798/KOSHAM.2016.16.1.103
  12. Qing, X. and Niu, Y. (2018). Hourly Day-ahead Solar Irradiance Prediction Using Weather Forecasts by LSTM. Energy. 148: 461-468. https://doi.org/10.1016/j.energy.2018.01.177
  13. Sharma, N., Sharma, P., Irwin, D., and Shenoy, P. (2011). Predicting Solar Generation from Weather Forecasts Using Machine Learning. In 2011 IEEE International Conference on Smart Grid Communications (SmartGridComm). 528-533.
  14. Won, Y. J., Yeh, S. W., Yim, B. Y., and Kim, H. K. (2017). Relationship between Korean Monthly Temperature during Summer and Eurasian Snow Cover during Spring. Atmosphere. 27(1): 55-65. https://doi.org/10.14191/Atmos.2017.27.1.055
  15. Yoo, H., Lee, S. O., Choi, S., and Park, M. (2019). A Study on the Data Driven Neural Network Model for the Prediction of Time Series Data: Application of Water Surface Elevation Forecasting in Hangang River Bridge. Journal of Korean Society of Disaster and Security. 12(2): 73-82. https://doi.org/10.21729/KSDS.2019.12.2.73
  16. Yoon, J.W. and Jeon, M.G. Temperature Forecasting Model by Using Deep Learning Technology based on LSTM. Proceeding of Institute of Electronics and Information Engineers Conference. 912-915.
  17. Zhang, Q., Wang, H., Dong, J., Zhong, G., and Sun, X. (2017). Prediction of Sea Surface Temperature Using Long Short-term Memory. IEEE Geoscience and Remote Sensing Letters. 14(10): 1745-1749. https://doi.org/10.1109/LGRS.2017.2733548