Quality Control of Ground-Based Radiometric Observations of Integrated Moisture Using Surface Meteorological Observations

View More View Less
  • 1 CIRES/NOAA, Boulder, Colorado
  • | 2 NOAA/ERL/Wave Propagation Laboratory, Boulder, Colorado
© Get Permissions
Full access

Abstract

A method of quality control is presented for ground-based, dual-channel microwave measurements of integrated moisture. Such a method is necessary to eliminate spurious data arising from calibration uncertainties, electronic fluctuations, and strong rain and melting snow on the radiometer antenna. The method is based on the prediction of integrated moisture content from surface measurements of temperature and dewpoint temperature. The statistical prediction was based on regression using a carefully screened multiyear training set of surface meteorological observations, and radiosonde and dual-channel radiometric measurements of moisture. Five years of twice-daily data (six years for the summer months) from Denver, Colorado, as well as data obtained from special experiments at Elbert and Platteville, Colorado, formed the training set. Both linear and nonlinear predictions were compared. The method was applied to independent data obtained during 1991–92 experiments at the three locations. The method predicted data quality with a 91% accuracy rate.

Abstract

A method of quality control is presented for ground-based, dual-channel microwave measurements of integrated moisture. Such a method is necessary to eliminate spurious data arising from calibration uncertainties, electronic fluctuations, and strong rain and melting snow on the radiometer antenna. The method is based on the prediction of integrated moisture content from surface measurements of temperature and dewpoint temperature. The statistical prediction was based on regression using a carefully screened multiyear training set of surface meteorological observations, and radiosonde and dual-channel radiometric measurements of moisture. Five years of twice-daily data (six years for the summer months) from Denver, Colorado, as well as data obtained from special experiments at Elbert and Platteville, Colorado, formed the training set. Both linear and nonlinear predictions were compared. The method was applied to independent data obtained during 1991–92 experiments at the three locations. The method predicted data quality with a 91% accuracy rate.

Save