Assessment of EOS-07 MHS satellite observations and retrieval of specific humidity profiles using a random forest-based algorithm

Abstract

An in-house-developed millimeter-wave humidity sounder onboard EOS-07 (EOS-07 MHS), launched in February 2023, operates at six frequencies around the 183.3 GHz water vapor absorption band. This study presents a preliminary performance assessment of EOS-07 MHS, including brightness temperature validation, humidity profile retrieval methodology and its validation.

Under clear-sky conditions, the biases in brightness temperature measured by EOS-07 MHS, relative to RTTOV simulations were within ±1 K, except for channels 1 and 6. Similarly, intercomparisons with ATMS observations showed biases within ±1 K and a standard deviation of 2–3 K.

A random forest-based method was employed to retrieve specific humidity profiles from EOS-07 MHS observations demonstrated agreement with ERA5 reanalysis and radiosonde observations. Compared with radiosonde data, the mean bias and standard deviation of retrieved specific humidity were approximately 0.78 g/kg and 2.3 g/kg, respectively. The mean percentage bias was within ±20 % below the 800 hPa pressure level, and ranged between ±20 % and ± 40 % above the 800 hPa pressure level. Relative to ERA5, the mean bias and root-mean-square deviation (RMSD) were under 30 % and 50 %, respectively. The estimated total precipitable water vapor showed a mean bias of 1.7–3.1 mm and a standard deviation of 5.2–5.7 mm compared to ERA5. Additionally, the EOS-07 MHS data were assimilated into the WRF model, resulting in improved atmospheric analyses and forecasts. A month-long cyclic assimilation experiment demonstrated consistent enhancements in moisture representation across the lower and middle atmosphere.