The first decadal-scale ground-based microwave radiometer dataset in China: Brightness temperature and thermodynamic profiles from Xianghe (2013–2022)

This Zenodo record provides two complementary atmospheric data products from Xianghe (2013-2022), distributed in a flat file layout (no subfolders).

Product group A: brightness temperature and QC/weather flags (1-minute resolution)
These files contain continuous ground-based MWR brightness temperature observations with quality-control and weather-condition identifiers. They include branch QC flags (n1-n4) and final QC labels (nflag) for reliability screening and condition-dependent analysis.
File pattern: annual files named as YYYY_final_data_with_flag_units_v3.csv (2013-2022).
Documentation: README_BT_flag_units_v3.md

Product group B: retrieved temperature and humidity profiles (10-minute resolution)
These files contain atmospheric profile retrievals derived from 10-minute averaged MWR observations. The profiles extend from 0 m to 10 km (m above ground level) and include retrieval-method labels (OE/DNN). Surface meteorological variables from AWS are included in the surface0m versions. "The Temp_0m and RH_0m" (0 m surface air temperature and ) were affected by AWS instrument operational issues, resulting in a high proportion of missing data (NaN) during the 2013–2022 period.
Main files:
2013-2022final_merged_temperature_with_method_surface0m.csv
2013-2022final_merged_RH_with_method_surface0m.csv
Documentation: README_profiles_with_method.md

Time and usage notes

Brightness temperature + flags: 1-minute temporal resolution.
Retrieved profiles: 10-minute temporal resolution.
Timestamps are in UTC (see the two README files for variable definitions, units, and QC logic).

For file interpretation, please read README_profiles_with_method.md for profile products and README_BT_flag_units_v3.md for brightness temperature/QC products first.

Regarding the Python scripts, 02_optimal_alllayer.py is designated for OE retrieval, while 11_DNN_rh_ERA5_obs_true(1).py and 11_DNN_t_ERA5obs_true.py correspond to the DNN implementation