Retrieval algorithms for atmospheric attenuation in the frequency band 15–52 GHz from two-channel microwave radiometer observations
Conference poster, 2019
Retrieval algorithms have been produced within the context of the prototype development of a dual-channel microwave radiometer by Omnisys Instruments, Gothenburg, Sweden.
Input data to describe the atmospheric properties are taken from ERA-Interim via its web interface. The quantities used are: surface pressure, temperature profile, geopotential altitude, humidity profile, liquid water content profile, and low cloud fraction. Data were downloaded for the years 2001–2014, for UT 00, 06, 12 and 18 at the highest available resolution. The position was selected to match the location of the Onsala Space Observatory. This atmospheric database contains about 20 000 cases.
The following assumptions were made about the radiometer: The antenna pattern was assumed to be sufficiently narrow that a pencil beam calculation represents the complete antenna temperature. The instrument channels were assumed to be sufficiently narrow that a monochromatic calculation represents the complete channel brightness temperature. The instrument was assumed to have two channels, with centre frequencies on and off the water vapour emission line at 22 GHz. The magnitude of uncorrelated (thermal) noise was assumed to be 0.6 K, for both channels (including noise added by the atmosphere and the calibration process). Errors fully correlated between the two channels were assumed to be between 0.37 K and 0.47 K. All instrument errors are assumed to be independent of the observed air mass.
The observation database produced by the ARTS forward model covers 15 GHz to 52 GHz, in steps of 200 MHz, and holds data for different viewing angles matching air mass factors between 1 and 6. The following quantities were calculated: brightness temperature, transmission, zenith hydrostatic delay, slant total delay, slant wet delay, geometric delay due to bending, water vapour path, and liquid water path.
Polynomial regression models are presented. The Onsala site is used for testing and demonstration of the retrieval performance, but the retrievals can easily be adopted to the conditions at other sites as the atmospheric data used to determine the regression coefficients are taken from a global atmospheric model (e.g. the ERA-Interim).
The errors found should be acceptable and match what is achieved by existing radiometers. The requirement for the transmission retrieval is met. In fact, this requirement is met with some marginal and the instrument performance could be poorer than estimated without violating the requirement.