Laser & Optoelectronics Progress, Volume. 60, Issue 17, 1701002(2023)

Retrieving Atmospheric Motion Vectors from Geostationary Satellite Images Using Generative Adversarial Networks

Xiaoyong Li1,2 and Keyi Chen1、*
Author Affiliations
  • 1School of Atmospheric Sciences, Chengdu University of Information Technology, Chengdu 610225, Sichuan , China
  • 2Network and Equipment Support Center, Taizhou Meteorological Bureau, Taizhou 318000, Zhejiang , China
  • show less
    Figures & Tables(9)
    Weighting functions of water vapor channel and infrared channel of GOES satellite, quoted from http://cimss.ssec.wisc.edu
    U-net architecture used by generator of pix2pix. Dimensions of data are shown as (channels, width, height)
    Test error for experiments using 1 h satellite image intervals, pix2pix architecture, high resolution data, and without visible channels. (a) 40000 iterations; (b) 500000 iterations
    Comparison of wind speed retrieved by neural network and wind speed of NCEP/NCAR reanalysis. (a) 850 hPa; (b) 200 hPa
    Spatial distribution of MVD of wind retrieved by neural network. (a) 850 hPa; (b) 200 hPa
    Comparison of wind field retrieved by neural network and wind field of NCEP/NCAR reanalysis. (a) 850 hPa wind direction (streamline) and speed (shaded) retrieved by neural network at October 9, 2019 12:00 UTC; (b) same as (a), but with NCEP/NCAR reanalysis data; (c) 200 hPa wind direction (streamline) and speed (shaded) retrieved by neural network at October 9, 2019 12:00 UTC; (d) same as (c), but with NCEP/NCAR reanalysis data; (e)-(h) same as (a)-(d), but at January 26, 2018 18:00 UTC
    • Table 1. Results of wind field retrieve with different input satellite image intervals, channels, and neural network types

      View table

      Table 1. Results of wind field retrieve with different input satellite image intervals, channels, and neural network types

      Image intervalVisible channelNetwork type

      850 hPa

      RMSE

      850 hPa

      MAE

      850 hPa

      Corr.

      200 hPa

      RMSE

      200 hPa

      MAE

      200 hPa

      Corr.

      6 hWithout visible channelpix2pix3.28112.51790.83916.00104.56300.9311
      6 hWithout visible channelU-net3.33762.54180.83155.96324.55570.9323
      1 hWithout visible channelpix2pix3.20872.42670.83805.87234.43690.9343
      1 hWith visible channelpix2pix3.29332.50440.83116.11424.65130.9272
    • Table 2. Results of wind field retrieve with different resolutions of data

      View table

      Table 2. Results of wind field retrieve with different resolutions of data

      Data (resolution)

      850 hPa

      RMSE

      850 hPa

      MAE

      850 hPa

      Corr.

      200 hPa

      RMSE

      200 hPa

      MAE

      200 hPa

      Corr.

      NCEP/NCAR reanalysis (64×64)3.20872.42670.83805.87234.43690.9343
      ERA5 (512×512)10.70338.56500.441627.145822.08820.7435
      ERA5 (64×64)10.66008.53440.442627.192622.14280.7433
    • Table 3. Comparison of results between neural network and H8 AMV product

      View table

      Table 3. Comparison of results between neural network and H8 AMV product

      MVD /(m·s-1RMSVD /(m·s-1Bias /(m·s-1
      H8 AMV (100-400 hPa)4.865.77-0.20
      Neural network (200 hPa)4.865.77-1.11
      H8 AMV (700-1100 hPa)3.143.860.68
      Neural network (850 hPa)3.203.81-0.42
    Tools

    Get Citation

    Copy Citation Text

    Xiaoyong Li, Keyi Chen. Retrieving Atmospheric Motion Vectors from Geostationary Satellite Images Using Generative Adversarial Networks[J]. Laser & Optoelectronics Progress, 2023, 60(17): 1701002

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Atmospheric Optics and Oceanic Optics

    Received: Mar. 17, 2022

    Accepted: Jun. 13, 2022

    Published Online: Aug. 29, 2023

    The Author Email: Keyi Chen (ckydlt@aliyun.com)

    DOI:10.3788/LOP221036

    Topics