Instrument uncertainties of network-suitable ground-based microwave radiometers: overview, quantification, and mitigation strategies

Abstract. To enhance observations of the Atmospheric Boundary Layer (ABL), the European Meteorological Network, EUMETNET, and the Aerosol, Clouds, and Trace Gases Research Infrastructure, ACTRIS, are currently collaborating to establish networks of MicroWave Radiometers (MWRs). MWRs can be used to derive thermodynamic profiles within the ABL. Understanding and assessing instrument uncertainties of state-of-the-art MWRs is therefore crucial for accurate observations and also data assimilation purposes. Some national weather services are currently exploring the potential of MWR networks to improve short-term weather forecasts. In this paper, we discuss uncertainties inherent to the MWR instrument itself, namely (1) radiometric noise, (2) long-term drifts and jumps, (3) calibration repeatability, (4) biases/systematic differences between instruments, and (5) radome degradation. These uncertainties are expressed in brightness temperatures. For state-of-the-art MWRs (here, Generation 5 Humidity and Temperature PROfiler HATPRO-Gen5), radiometric noise at ambient temperatures is a maximum of 0.32 K in the V-band but usually lower, especially near cold load temperature ranges in the K-band (≤ 0.11 K). Long-term drifts or jumps between calibrations, which are at least two months apart, are usually below 0.4 K in the K-band and 1.0 K in the V-band but can also be higher. Drifts do not follow a discernable timely pattern and are therefore not easily quantifiable in a meaningful way. Calibrating at least every six months is thus advised. Calibration repeatability is shown to be well under 0.16 K. Mean brightness temperature differences between two HATPRO-Gen5 instruments are shown to be as high as 0.15 K in the K-band and 0.58 K in the V-band at zenith viewing direction. The radome has significantly degraded due to weathering and needs to be replaced if, 10 min after a rain event, residual water on its surface still causes a temperature deviation of more than 2 K compared to a dry radome.