Content area
Full Text
ABSTRACT
A three-dimensional radar simulator capable of generating simulated raw time series data for a weather radar has been designed and implemented. The characteristics of the radar signals (amplitude, phase) are derived from the atmospheric fields from a high-resolution numerical weather model, although actual measured fields could be used. A field of thousands of scatterers is populated within the field of view of the virtual radar. Reflectivity characteristics of the targets are determined from well-known parameterization schemes. Doppler characteristics are derived by forcing the discrete scatterers to move with the three-dimensional wind field. Conventional moment-generating radar simulators use atmospheric conditions and a set of weighting functions to produce theoretical moment maps, which allow for the study of radar characteristics and limitations given particular configurations. In contrast to these radar simulators, the algorithm presented here is capable of producing sample-to-sample time series data that are collected by a radar system of virtually any design. Thus, this new radar simulator allows for the test and analysis of advanced topics, such as phased array antennas, clutter mitigation schemes, waveform design studies, and spectral-based methods. Limited examples exemplifying the usefulness and flexibility of the simulator will be provided.
(ProQuest: ... denotes formulae omitted.)
1. Introduction
A realistic numerical simulation can provide a deterministic and controlled environment for a wide variety of engineering and scientific studies. Extreme scenarios can be simulated to test the robustness and limitations of signal-processing techniques, which help identify and scrutinize factors that may have been overlooked during the development process. Weather radar-like signals have been simulated since at least the 1970s. For example, work by Zrnié (1975) was based on an assumed, but arbitrary, Doppler spectral form. The inverse Fourier transform of this spectrum was performed to produce time series data corresponding to that spectral shape. Numerous statistical studies were made possible using this simulation. Based on the algorithm by Zrnic (1975), Chandrasekar and Bringi (1987) developed a simulation scheme to generate radar reflectivity for a simulated raindrop size distribution that had a gamma distribution (Ulbrich 1983). In that work, the simulation was used to investigate the correlation of radar estimates and rainfall rate.
To devise a more realistic time series simulator, Capsoni and D'Amico (1998) formulated a physically based procedure to simulate pulse-to-pulse...