Advancing Satellite Precipitation Retrievals With Data Driven Approaches: Is Black Box Model Explainable?. Issue 2 (23rd February 2021)

• Record Type: Journal Article
• Authors: Li, Zhi; Wen, Yixin; Schreier, Mathias; Behrangi, Ali; Hong, Yang; Lambrigtsen, Bjorn
• Is Part Of: Earth and space science. Volume 8:Issue 2(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-02-23
• Subjects: deep learning -- interpretable machine learning -- precipitation estimation; Space sciences -- Periodicals; Geophysics -- Periodicals; 500.5
• DOI: 10.1029/2020EA001423
• Languages: English
• ISSNs: 2333-5084

Abstract

Abstract: Satellite‐based precipitation retrieval is an essential and long‐standing scientific problem. With an increase of observational satellite data, the advances of data‐driven approaches such as machine learning (ML)/deep learning (DL) are favored to deal with large data sets and potentially improve the accuracy of precipitation estimates. In this study, we took advantage of new technologies by wrapping up a ML/DL‐based model pipeline (LinkNet segmentation + tree ensemble). This approach is applied to the Advanced Microwave Sounding Unit (AMSU) on National Oceanic and Atmospheric Administration 18 and 19 flight, and compared with the MultiRadar MultiSensor. Four simulations were configured to examine the performance gain by incorporating three components: (1) precipitation identification, (2) nonlocal features, and (3) precipitation classification. More importantly, we examined the interpretability of the "black box" model to get a better understanding of the underlying physical connections. First, the results by this model pipeline suggest the advantages of the ML model by reducing the systematic error and instantaneous error to a factor of two. Second, identifying precipitation pixels helps to reduce the systematic error by 130%, and predicting precipitation classification benefits improved correlations by 32%. Last, channels at higher frequencies (beyond 150 GHz) are favored to identify precipitation regions, and also channels at 89 and 150 GHz are ranked as the two most important features to precipitation retrieval. This study explores the potentials of AMSU precipitation estimates with ML algorithms and provides means of interpreting the models to facilitate the better prediction of precipitation. Plain Language Summary: Accurate precipitation estimations could undoubtedly benefit water resources management and decision makings. As the precipitation data explode, especially for remote sensing, there is a growing opportunity to leverage these myriads of data to advance the satellite retrieval algorithms. In this study, we apply this novel machine learning/deep learning pipeline with National Oceanic and Atmospheric Administration (NOAA)‐18 and NOAA‐19 satellite to compare with an operational retrieval algorithm. Beyond that, the black box model interpretability is explored to discover how the model predicts the outcome. The results prove that this model not only outperforms than the operational product, it but also makes physical sense with respect to its functionality. Key Points: A proposed machine learning (ML) pipeline to improve precipitation quantification The proposed framework applies sucessfully to precipitation classification and quantification Interpretable ML bridges the precipitation physics and the black box mode