Abstract
Weather radars provide observations of precipitation events with high spatial and temporal resolutions. These observations are useful for generating quantitative precipitation estimates (QPEs), forecasting heavy rain events, or nowcasting flash-flood events. However, radar rainfall estimates are affected by several error sources such as radar miscalibrations, beam broadening and range degradation, ground clutter and anomalous propagation of the beam, beam blockage or attenuation of the radar signal. Moreover, the rain rates are not directly measured by the radars but derived from the electromagnetic properties of hydrometeors instead.The advent of dual polarisation in recent years has improved the radar data quality and consequently, the radar rain products. In 2018, the UK Met Office completed the process of upgrading its operational weather radar network with polarimetric capability. In this thesis, we use data collected by these radars to develop novel methodologies to improve the radar data quality and thereby, the rain estimates. In particular, we present (1) a novel, operational methodology to detect the boundaries of the melting layer (ML) using only polarimetric weather radar data. (2) a novel, robust methodology to compute the calibration offset of differential reflectivity ($Z_{DR}$) measurements using quasi-vertical profiles and (3) an open-source toolbox capable of processing C-band polarimetric weather radar data from the UK weather radar network. This toolbox contains libraries dedicated to process raw radar data and improve the quality of radar rainfall estimates (including those described in points (1) and (2)) for hydrological and meteorological applications.
Results showed that it is possible to detect the ML using vertical profiles (VPs) or quasi-vertical profiles (QVPs) with an accuracy of around 250 m compared to radiosonde data, which is similar to the accuracy achieved by numerical models. These results led to development of a new approach to detect the $Z_{DR}$ calibration offset using VPs and QVPs. The QVP-based $Z_{DR}$ calibration method showed differences of around 0.35 dB when compared to disdrometer data. This value is in good agreement with the required accuracy found in previous works. Finally, these and other algorithms to process the raw radar data are available for reproducibility in an open-source toolbox that has proven effective to read, process and display data generated by the UK Met Office polarimetric weather radar network.
| Date of Award | 12 May 2022 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Miguel A Rico-Ramirez (Supervisor) & Dawei Han (Supervisor) |