Improving Precipitation Forecasts with Convolutional Neural Networks

被引:9
作者
Badrinath, Anirudhan [1 ]
delle Monache, Luca [2 ]
Hayatbini, Negin [2 ]
Chapman, Will [2 ]
Cannon, Forest [2 ]
Ralph, Marty [2 ]
机构
[1] Univ Calif Berkeley, Dept Comp Sci, Berkeley, CA 94720 USA
[2] Univ Calif San Diego, Scripps Inst Oceanog, Ctr Western Weather & Water Extremes, La Jolla, CA USA
关键词
Numerical weather prediction; forecasting; Postprocessing; Machine learning; WIND-SPEED FORECASTS; ANALOG ENSEMBLE; PREDICTABILITY; TEMPERATURE; UNCERTAINTY;
D O I
10.1175/WAF-D-22-0002.1
中图分类号
P4 [大气科学(气象学)];
学科分类号
0706 ; 070601 ;
摘要
A machine learning method based on spatial convolution to capture complex spatial precipitation patterns is proposed to identify and reduce biases affecting predictions of a dynamical model. The method is based on a combina-tion of a classification and dual-regression model approach using modified U-Net convolutional neural networks (CNN) to postprocess daily accumulated precipitation over the U.S. West Coast. In this study, we leverage 34 years of high-resolution deterministic Western Weather Research and Forecasting (West-WRF) precipitation reforecasts as training data for the U-Net CNN. The data are split such that the test set contains 4 water years of data that encompass characteristic West Coast precipitation regimes: El Nin similar to o, La Nin similar to a, and dry and wet El Nin similar to o-Southern Oscillation (ENSO neutral) water years. On the unseen 4-yr dataset, the trained CNN yields a 12.9%-15.9% reduction in root-mean-square error (RMSE) and 2.7%-3.4% improvement in Pearson correlation (PC) over West-WRF for lead times of 1-4 days. Compared to an adapted model output statistics correction, the CNN reduces RMSE by 7.4%-8.9% and improves PC by 3.3%-4.2% across all events. Effectively, the CNN adds more than a day of predictive skill when compared to West-WRF. The CNN outper-forms the other methods also for the prediction of extreme events, which we define as the top 10% of events with the great-est average daily accumulated precipitation. The improvement over West-WRF's RMSE (PC) for these events is 19.8%-21.0% (4.9%-5.5%) and MOS's RMSE (PC) is 8.8%-9.7% (4.2%-4.7%). Hence, the proposed U-Net CNN shows significantly im-proved forecast skill over existing methods, highlighting a promising path forward for improving precipitation forecasts.
引用
收藏
页码:291 / 306
页数:16
相关论文
共 50 条
  • [1] A novel application of an analog ensemble for short-term wind power forecasting
    Alessandrini, S.
    Delle Monache, L.
    Sperati, S.
    Nissen, J. N.
    [J]. RENEWABLE ENERGY, 2015, 76 : 768 - 781
  • [2] Improving the Analog Ensemble Wind Speed Forecasts for Rare Events
    Alessandrini, Stefano
    Sperati, Simone
    Delle Monache, Luca
    [J]. MONTHLY WEATHER REVIEW, 2019, 147 (07) : 2677 - 2692
  • [3] Alsallakh B, 2020, Arxiv, DOI [arXiv:2010.02178, 10.48550/arXiv.2010.02178]
  • [4] On the discriminative power of Hyper-parameters in Cross-Validation and how to choose them
    Anelli, Vito Walter
    Di Noia, Tommaso
    Di Sciascio, Eugenio
    Pomo, Claudio
    Ragone, Azzurra
    [J]. RECSYS 2019: 13TH ACM CONFERENCE ON RECOMMENDER SYSTEMS, 2019, : 447 - 451
  • [5] [Anonymous], 2019, Glossary of Meteorology
  • [6] Badrinath A., 2023, IMPROVING PRECIPITAT, DOI [10.6075/J0J103BG, DOI 10.6075/J0J103BG]
  • [7] Observations and Predictability of a High-Impact Narrow Cold-Frontal Rainband over Southern California on 2 February 2019
    Cannon, Forest
    Oakley, Nina S.
    Hecht, Chad W.
    Michaelis, Allison
    Cordeira, Jason M.
    Kawzenuk, Brian
    Demirdjian, Reuben
    Weihs, Rachel
    Fish, Meredith A.
    Wilson, Anna M.
    Ralph, F. Martin
    [J]. WEATHER AND FORECASTING, 2020, 35 (05) : 2083 - 2097
  • [8] Short-term photovoltaic power forecasting using Artificial Neural Networks and an Analog Ensemble
    Cervone, Guido
    Clemente-Harding, Laura
    Alessandrini, Stefano
    Delle Monache, Luca
    [J]. RENEWABLE ENERGY, 2017, 108 : 274 - 286
  • [9] Improving Atmospheric River Forecasts With Machine Learning
    Chapman, W. E.
    Subramanian, A. C.
    Delle Monache, L.
    Xie, S. P.
    Ralph, F. M.
    [J]. GEOPHYSICAL RESEARCH LETTERS, 2019, 46 (17-18) : 10627 - 10635
  • [10] Probabilistic Predictions from Deterministic Atmospheric River Forecasts with Deep Learning
    Chapman, William E.
    Delle Monache, Luca
    Alessandrini, Stefano
    Subramanian, Aneesh C.
    Ralph, F. Martin
    Xie, Shang-Ping
    Lerch, Sebastian
    Hayatbini, Negin
    [J]. MONTHLY WEATHER REVIEW, 2022, 150 (01) : 215 - 234