1.5 The approach

This thesis attempts to thread the relative calibration approach together with the concept of data quality. Radar calibration ensures homogeneity in radar networks where comparable measurements of precipitation are essential in the overlapping regions of two or more weather radars. When combining two or more overlapping radar sweeps to produce a composite image, often the basis for selection is data quality.

Data quality is defined in Michelson et al. (2005) as the “attribute of the data which is inverse to uncertainties and errors, i.e. error-free data with few uncertainties are of high quality while data with errors or large uncertainties are of low quality”. A Quality Index metric classifies the data quality within an interval of 0 to 1, where 0 represents poor quality and 1 represents excellent quality. Quality indices are typically used in combining data from multiple radars to create a composite image over larger regions (e.g. radar composites for a specific catchment, or for an entire country). For bias calibration purposes, quality indices can be used as weights in a weighted-averaging approach for calibration.

In particular, this thesis looks at two factors affecting data quality—beam blockage and path-integrated attenuation:

  1. Beam Blockage: When the topography surrounding a radar interferes with the path of the radar beam, it may partially or completely hinder the radar’s ability to detect the precipitation further along the beam. Such topographic barriers may lead to a weaker backscattered signal. Flat regions within the radar coverage are assigned high data quality. Data quality quickly drops when the radar is blind due to the topographic barriers. This source of uncertainty is considered static, as the obstacles (such as mountains, buildings, or other permanent structures) do not change from scan to scan.

  2. Path-integrated attenuation: At wavelengths shorter than 10 cm (such as C-band radars), the radar signal becomes weaker as it passes through rainfall. The magnitude of attenuation is proportional to rain intensity, making it highly variable in space and time. The effects accumulate along the radar beam (hence the term path-integrated). This source of uncertainty is dynamic, as it depends on the rain intensity and therefore changes with every scan.

Determining calibration bias through comparison with spaceborne radars and integrating a quality-weighted approach brings together different threads of the field. Calibration estimation and correction attempts to address systematic errors that are homogeneous over the entire radar domain, whereas factoring in quality allows other sources of systematic errors that are heterogeneous in space to be addressed separately. It is always worthwhile to question the data quality and the reliability, when determining the calibration bias of the ground radar with respect to the spaceborne radar. Poor data quality used in such a comparison may lead to errors in bias estimation, resulting in inaccurate bias correction.

References

Michelson, Daniel, Thomas Einfalt, Iwan Holleman, Uta Gjertsen, Friedrich, Katja, Günther Haase, Magnus Lindskog, and Anna Jurczyk. 2005. Weather Radar Data Quality in Europe Quality Control and Characterisation. Luxembourg: Publications Office.