Submitted as: model evaluation paper
11 Jan 2023
Submitted as: model evaluation paper | 11 Jan 2023
Status: this preprint is currently under review for the journal GMD.

Key factors for quantitative precipitation nowcasting using ground weather radar data based on deep learning

Daehyeon Han1,, Yeji Shin2,, Jungho Im1, and Juhyun Lee1 Daehyeon Han et al.
  • 11Department of Urban and Environmental Engineering, Ulsan National Institute of Science and Technology, Ulsan, 44919, South Korea
  • 2National Institute of Meteorological Sciences, Korea Meteorological Administration, Jeju-do, South Korea
  • These authors contributed equally to this work.

Abstract. Quantitative precipitation nowcasting (QPN) may assist to mitigate the tremendous socioeconomic harm caused by severe weather. Because of the rapid atmospheric variability, the QPN has been a challenging problem to solve. Recent QPN research has presented data-driven models that make use of deep learning (DL) and ground weather radar. Previous research has mostly concentrated on constructing DL models, while other elements for DL-QPN have received less attention. This research looks at four crucial aspects in the DL-QPN and their impact on predicting performance. The prediction strategy (single, recursive, and multiple predictions), deep learning model (U-Net and convolutional long short term memory; ConvLSTM), input past sequence length (60 and 120 min), and output future sequence length were the four key factors (60 and 120 min). Using weather radar data from South Korea, twelve schemes have been developed to assess the influence of each factor. A long-term operational study for 2018–2020 was conducted, and a summer high rainfall event was examined to investigate the extreme case. In both situations, U-Net outperformed the critical score index (CSI) using a multiple prediction design. While ConvLSTM did not show a definite CSI difference across input sequence length, U-Net performed better with shorter input sequences (i.e., 60 min) than with longer input sequences (i.e., 120 min). The length of future sequences has little influence on model performance. As the lead time increased, all of the DL-QPN schemes showed underestimation and blurry outcomes. U-Net was shown to be significantly reliant on the most recent input time (i.e., 0 previous minutes) in sensitivity analysis, while ConvLSTM was more responsive to multiple time steps. This work may give a modeling technique and contingency plan for future DL-QPN employing weather radar data by explicitly comparing critical elements.

Daehyeon Han et al.

Status: open (until 08 Mar 2023)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse

Daehyeon Han et al.

Daehyeon Han et al.


Total article views: 224 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
187 36 1 224 1 0
  • HTML: 187
  • PDF: 36
  • XML: 1
  • Total: 224
  • BibTeX: 1
  • EndNote: 0
Views and downloads (calculated since 11 Jan 2023)
Cumulative views and downloads (calculated since 11 Jan 2023)

Viewed (geographical distribution)

Total article views: 206 (including HTML, PDF, and XML) Thereof 206 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
Latest update: 28 Jan 2023
Short summary
The four key factors in the DL-QPN are summarized and compared in this work. A total of 12 schemes were compared, with the prediction design, deep learning model, and previous and future sequence lengths. Multiple U-Net predictions seem to be the best combination. The impact of the input-output ratio was smaller than expected. The DL-QPN has limitations, including underestimation and smoothed spatial patterns.