Validation of the new water vapour radiometer at Onsala for atmospheric corrections in space geodesy

The signal delay caused by atmospheric water vapour is an important error source in space geodetic techniques, such as Global Navigation Satellite Systems (GNSS) and Very Long Baseline Interferometry (VLBI). When GNSS and VLBI data are used to estimate signal delays, these are typically parameterized as zenith total delay (ZTD) and north/east linear horizontal gradients. A ground-based water vapour radiometer (WVR) can infer the delay due to water vapour, referred to as the wet delay, in specific directions in the sky. The WVR’s independent measurement of the wet delay can refine error modeling in VLBI/GNSS systems. Onsala Space Observatory has operated a WVR (named Konrad) alongside co-located GNSS/VLBI antennas for many years. In May 2023, a new WVR (named Greta) was installed, featuring key improvements. Greta is equipped with a heater-blower system to remove water from the feed during rainy conditions, calibrated using a liquid nitrogen load (unlike Konrad’s “tip-curve” method), and employs a neural network algorithm for retrieving atmospheric products using 14 frequency channels. In contrast, Konrad relies on a linear regression method with only two frequencies. More than a year of simultaneous observations with Konrad, Greta, GNSS, and VLBI have been used to validate Greta’s performance. The correlation between Greta and Konrad is very high (>0.99 for zenith wet delay (ZWD) and >0.85 for gradients), demonstrating strong consistency between the two instruments. Additionally, the correlation between Greta and GNSS gradients (≈0.8) slightly exceeds that of Konrad (≈0.74). A similar increase in correlation is observed when using VLBI as a reference, although with lower overall correlation levels (0.5-0.6). Both radiometers exhibit larger dynamic ranges for ZWD and gradients compared to GNSS (slope ≈1.05). However, radiometer-measured wet delays are sensitive to rain and other environmental conditions, making careful data quality control essential. Ongoing efforts are focused on testing a new retrieval algorithm tailored to local conditions to further enhance Greta’s performance.