| description abstract | AbstractRadiosonde observations are used the world over to provide critical upper-air observations of the lower atmosphere. These observations are susceptible to errors that must be mitigated or avoided when identified. One source of error not previously addressed is radiosonde icing in winter storms, which can affect forecasts, warning operations, and model initialization. Under certain conditions, ice can form on the radiosonde, leading to decreased response times and incorrect readings. Evidence of radiosonde icing is presented for a winter storm event in Norman, Oklahoma, on 24 November 2013. A special sounding that included a particle imager probe and a GoPro camera was flown into the system producing ice pellets. While the iced-over temperature sensor showed no evidence of an elevated melting layer (ML), complementary Particle Size, Image, and Velocity (PASIV) probe and polarimetric radar observations provide clear evidence that an ML was indeed present. Radiosonde icing can occur while passing through a layer of supercooled drops, such as frequently found in a subfreezing layer that often lies below the ML in winter storms. Events that have warmer/deeper MLs would likely melt any ice present off the radiosonde, minimizing radiosonde icing and allowing the ML to be detected. This paper discusses the hypothesis that the absence of an ML in the radiosonde data presented here is more likely to occur in winter storms that produce ice pellets, which tend to have cooler/shallower MLs. Where sounding data do appear to be compromised by icing, polarimetric radar data might be used to identify MLs for nowcasting purposes and numerical model initialization. | |
