Updates on Radar Refractivity Retrieval – Quality Control Improvements and the 2009 Field Experiment to Determine Causes of Bias
Unfortunately using current quality control (QC) methods, the domain used for refractivity can contain clutter points with poor phase coherency, which should be removed before being used in the algorithm. The algorithm interpolates and spatially filters the data due to the inherent statistically uncertainty and the general sparseness of the phase measurements, resulting in a spatial resolution of approximately 4 km. If clutter points with questionable phase data are allowed to pass through the algorithm, surrounding data points (up to a distance of 4 km) may be impacted, especially in regions with a low number of clutter points. Consequently, estimates of phase and refractivity will have degraded quality. This occurs most frequently near the edge of the clutter domain or where clutter signals may be dominated by tress. It can be shown that the existing algorithm had deteriorated refractivity quality when used on windy days, when these clutter targets were moving irregularly and introducing error-prone phase data into the algorithm. By determining which targets move in these situations using their spectral characteristics and phase coherency, we can censor these points and improve estimates of the refractivity field. An extensive statistical study will be presented using the Oklahoma Mesonet data as ground truth. In addition, an improved phase QC methodology will be proposed, which significantly improves the ultimate quality of the refractivity data.
During the course of this project a diurnal refractivity bias between Mesonet surface observations and the radar has been observed. This bias tends to peak around 00 UTC, and is most prevalent in the summertime. The observations of this bias has led to a field experiment aimed at answering the underlying cause. The experiment includes a modified Mesonet station, which can monitor refractivity at 1.5 and 9 meters at one-minute intervals, as well as vertical profiles provided by an unmanned aerial vehicle (UAV) outfitted with a full array of meteorological sensors. The Mesonet tower provides insight as to the near-surface refractivity gradient during periods of bias, and whether or not it may be due to changes in sampling height as a side effect of beam propagation changes. The UAV provides a refractivity profile over a deep layer, and can sample the evolution of the changing profile in accompaniment to Mesonet data during periods of bias. Profiles of refractivity for periods of bias using both data sources, as well as an analysis of correlation with the intensity and duration of observed bias, will be presented.