20 Jun 2024
 | 20 Jun 2024
Status: this preprint is open for discussion and under review for Atmospheric Measurement Techniques (AMT).

A Machine-learning Based Marine Planetary Boundary Layer (MPBL) Moisture Profile Retrieval Product from GNSS-RO Deep Refraction Signals

Jie Gong, Dong Liang Wu, Michelle Badalov, Manisha Ganeshan, and Minghua Zheng

Abstract. Marine planetary boundary layer (MPBL) water vapor amount and gradient impact the global energy transport through directly affecting the sensible and latent heat exchange between the ocean and atmosphere. Yet, it is a well-known challenge for satellite remote sensing to profile MABL water vapor, especially when cloud or sharp gradient of water vapor are present. Wu et al. (2022) identified good correlations between Global Navigation Satellite System (GNSS) deep refraction signals (SNR) and the global MPBL water vapor specific humidity when the radio occultation (RO) signal is ducted by the moist PBL layer, and they laid out the underlying physical mechanisms to explain such a correlation. In this work, we apply a machine-learning/artificial intelligence (ML/AI) technique to realize pixel-level MPBL water vapor profiling. A convolutional neural network (CNN) model is trained using 20 months of global collocated hourly ERA-5 reanalysis and COSMIC1 1 Hz SNR observations between 975 – 850 hPa with 25 hPa vertical resolution, and then the model is applied to both COSMIC1 and COSMIC2 in other time ranges for independent retrieval and validation. Monte Carlo Dropout method was employed for the uncertainty estimation. Comparison against multiple field campaign radiosonde/dropsonde observations globally suggests SNR-retrieved water vapor consistently outperforms ERA-5 reanalysis and the Level-2 standard retrieval product at all six pressure levels between 975 hPa and 850 hPa, indicating real and useful information is gained from the SNR signal albeit training was performed against the reanalysis. The only exception is in the deep tropics where the fundamental assumption for SNR-retrieval to work is invalidated frequently by interactions among ocean surface, MPBL and shallow convections. Climatology and diurnal cycle of MPBL structure constructed from the ML-SNR technique is studied and compared to the reanalysis. Disparities of climatology suggest ERA-5 may systematically produces dry biases at high-latitudes, and wet biases in marine stratocumulus regions. The diurnal cycle amplitudes are too weak and off-phase in ERA-5, especially in Arctic and stratocumulus regions. These areas are particularly prone to PBL processes where this GNSS-SNR water vapor product may contribute the most.

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.
Jie Gong, Dong Liang Wu, Michelle Badalov, Manisha Ganeshan, and Minghua Zheng

Status: open (until 26 Jul 2024)

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
  • RC1: 'Comment on egusphere-2024-973', Anonymous Referee #1, 28 Jun 2024 reply
  • RC2: 'Comment on egusphere-2024-973', Anonymous Referee #2, 04 Jul 2024 reply
Jie Gong, Dong Liang Wu, Michelle Badalov, Manisha Ganeshan, and Minghua Zheng
Jie Gong, Dong Liang Wu, Michelle Badalov, Manisha Ganeshan, and Minghua Zheng


Total article views: 175 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
136 29 10 175 5 5
  • HTML: 136
  • PDF: 29
  • XML: 10
  • Total: 175
  • BibTeX: 5
  • EndNote: 5
Views and downloads (calculated since 20 Jun 2024)
Cumulative views and downloads (calculated since 20 Jun 2024)

Viewed (geographical distribution)

Total article views: 169 (including HTML, PDF, and XML) Thereof 169 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
Latest update: 24 Jul 2024
Short summary
Marine boundary layer water vapor is among the key factors to couple the ocean and atmosphere, but it is also among the hardest to retrieve from satellite remote sensing perspective. Here we propose a novel way to retrieve MPBL specific humidity profiles using the GNSS Level-1 signal-to-noise ratio. Using a machine learning approach, we successfully obtained a retrieval product that outperforms the ERA-5 reanalysis and operational Level-2 retrievals globally except in the deep tropics.