Day 4

Detailed paper information

Back to list

Paper title Identifying relevant large-scale predictors for sub-seasonal precipitation forecast using explainable neural networks
Authors
  1. Niclas Rieger Centre de Recerca Matemàtica
  2. Linus Magnusson European Weather Centre (ECMWF)
  3. Laura Ferranti European Weather Centre (ECMWF)
  4. Estrella Olmedo Institut of Marine Sciences Speaker
  5. Álvaro Corral Centre de Recerca Matemàtica (CRM)
  6. Antonio Turiel Institute of Marine Sciences - CSIC
Form of presentation Poster
Topics
  • C1. AI and Data Analytics
    • C1.07 ML4Earth: Machine Learning for Earth Sciences
Abstract text The last few years have seen an ever growing interest in weather predictions on sub-seasonal time scales ranging from 2 weeks to about 2 months. By forecasting aggregated weather statistics, such as weekly precipitation, it has indeed become possible to overcome the theoretical predictability limit of 2 weeks (Lorenz 1963; F. Zhang et al. 2019), bringing life to time scales which historically have been known as the “predictability desert”. The growing success at these time scales is largely due to the identification of weather and climate processes providing sub-seasonal predictability, such as the Madden-Julian Oscillation (MJO) (C. Zhang 2013) and anomaly patterns of global sea surface temperature (SST) [Woolnough 2007, Saravanan & Chang 2019], sea surface salinity (Li et al. 2016; Chen et al. 2019; Rathore et al. 2021), soil moisture (Koster et al. 2010) and snow cover (Lin and Wu 2011). Although much has been gained by these studies, a comprehensive analysis of all potential predictors and their relative relevance to forecast sub-seasonal rainfall is still missing.

At the same time, data-driven machine learning (ML) models have proved to be excellent candidates to tackle two common challenges in weather forecasting: (i) resolving the non-linear relationships inherent to the chaotic climate system and (ii) handling the steadily growing amounts of Earth observational data. Not surprisingly, a variety of studies have already displayed the potential of ML models to improve the state-of-the-art dynamical weather prediction models currently in use for sub-seasonal predictions, in particular for temperatures (Peng et al. 2020; Buchmann and DelSole 2021) , precipitation (Scheuerer et al. 2020) and the MJO (Kim et al. 2021; Silini, Barreiro, and Masoller 2021). It seems therefore inevitable that the future of sub-seasonal prediction lies in the combination of both the dynamical, process-based and the statistical, data-driven approach (Cohen et al. 2019).

In the advent of this new age of combined Neural Earth System Modeling (Irrgang et al. 2021), we want to provide insight and guidance for future studies (i) to what extent large-scale teleconnections on the sub-seasonal scale can be resolved by purely data-driven models and (ii) what the relative contributions of the individual large-scale predictors are to make a skillful forecast. To this end, we build neural networks to predict sub-seasonal precipitation based on a variety of large-scale predictors derived from oceanic, atmospheric and terrestrial sources. As a second step, we apply layer-wise relevance propagation (Bach et al. 2015) to examine the relative importance of different climate modes and processes in skillful forecasts.

Preliminary results show that the skill of our data-driven ML approach is comparable to state-of-the-art dynamical models suggesting that current operational models are able to correctly model large-scale teleconnections within the climate system. The ML model achieves highest skills over the tropical Pacific, the Maritime Continent and the Caribbean Sea (Fig. 1), in agreement with dynamical models. By investigating the relative importance of those large-scale predictors for skillful predictions, we find that the MJO and processes associated to SST anomalies like the El Niño-Southern Oscillation, the Pacific decadal oscillation and the Atlantic meridional mode all play an important role for individual regions along the tropics.

Additional material

Figure 1 | Forecast skill of the ML model as represented by the Brier skill score (BSS) calculated with respect to climatology. Red color shadings show regions where the ML model performs better, while blue color shadings indicate worse skill than climatology. The BSS was calculated as an average over the period from 2015 to 2020 using weekly forecasts totalling 310 individual samples, which were set aside before the training process as a test set.

References

Bach, Sebastian, Alexander Binder, Grégoire Montavon, Frederick Klauschen, Klaus-Robert Müller, and Wojciech Samek. 2015. “On Pixel-Wise Explanations for Non-Linear Classifier Decisions by Layer-Wise Relevance Propagation.” PloS One 10 (7): e0130140.
Buchmann, Paul, and Timothy DelSole. 2021. “Week 3-4 Prediction of Wintertime CONUS Temperature Using Machine Learning Techniques.” Frontiers in Climate 3: 81.
Chen, B, H Qin, G Chen, and H Xue. 2019. “Ocean Salinity as a Precursor of Summer Rainfall over the East Asian Monsoon Region.” Journal of Climate 32 (17): 5659–76. https://doi.org/10.1175/JCLI-D-18-0756.1.
Cohen, Judah, Dim Coumou, Jessica Hwang, Lester Mackey, Paulo Orenstein, Sonja Totz, and Eli Tziperman. 2019. “S2S Reboot: An Argument for Greater Inclusion of Machine Learning in Subseasonal to Seasonal Forecasts.” WIREs Climate Change 10 (2): e00567. https://doi.org/10.1002/wcc.567.
Irrgang, Christopher, Niklas Boers, Maike Sonnewald, Elizabeth A. Barnes, Christopher Kadow, Joanna Staneva, and Jan Saynisch-Wagner. 2021. “Will Artificial Intelligence Supersede Earth System and Climate Models?,” January. https://arxiv.org/abs/2101.09126v1.
Kim, H., Y. G. Ham, Y. S. Joo, and S. W. Son. 2021. “Deep Learning for Bias Correction of MJO Prediction.” Nature Communications 12 (1): 3087. https://doi.org/10.1038/s41467-021-23406-3.
Koster, R. D., S. P. P. Mahanama, T. J. Yamada, Gianpaolo Balsamo, A. A. Berg, M. Boisserie, P. A. Dirmeyer, et al. 2010. “Contribution of Land Surface Initialization to Subseasonal Forecast Skill: First Results from a Multi-Model Experiment.” Geophysical Research Letters 37 (2). https://doi.org/10.1029/2009GL041677.
Li, L, R Schmitt, CC Ummenhofer, and KB Karnauskas. 2016. “North Atlantic Salinity as a Predictor of Sahel Rainfall.” Science Advances 2 (5): e1501588. https://doi.org/10.1126/sciadv.1501588.
Lin, Hai, and Zhiwei Wu. 2011. “Contribution of the Autumn Tibetan Plateau Snow Cover to Seasonal Prediction of North American Winter Temperature.” Journal of Climate 24 (11): 2801–13.
Lorenz, Edward N. 1963. “Deterministic Nonperiodic Flow.” Journal of Atmospheric Sciences 20 (2): 130–41.
Peng, Ting, Xiefei Zhi, Yan Ji, Luying Ji, and Ye Tian. 2020. “Prediction Skill of Extended Range 2-m Maximum Air Temperature Probabilistic Forecasts Using Machine Learning Post-Processing Methods.” Atmosphere 11 (8): 823.
Rathore, Saurabh, Nathaniel L. Bindoff, Caroline C. Ummenhofer, Helen E. Phillips, Ming Feng, and Mayank Mishra. 2021. “Improving Australian Rainfall Prediction Using Sea Surface Salinity.” Journal of Climate 1 (aop): 1–56. https://doi.org/10.1175/JCLI-D-20-0625.1.
Scheuerer, Michael, Matthew B. Switanek, Rochelle P. Worsnop, and Thomas M. Hamill. 2020. “Using Artificial Neural Networks for Generating Probabilistic Subseasonal Precipitation Forecasts over California.” Monthly Weather Review 148 (8): 3489–3506. https://doi.org/10.1175/MWR-D-20-0096.1.
Silini, Riccardo, Marcelo Barreiro, and Cristina Masoller. 2021. “Machine Learning Prediction of the Madden-Julian Oscillation.” Earth and Space Science Open Archive ESSOAr.
Zhang, Chidong. 2013. “Madden–Julian Oscillation: Bridging Weather and Climate.” Bulletin of the American Meteorological Society 94 (12): 1849–70.
Zhang, Fuqing, Y. Qiang Sun, Linus Magnusson, Roberto Buizza, Shian-Jiann Lin, Jan-Huey Chen, and Kerry Emanuel. 2019. “What Is the Predictability Limit of Midlatitude Weather?” Journal of the Atmospheric Sciences 76 (4): 1077–91. https://doi.org/10.1175/JAS-D-18-0269.1.