Multiple-Doppler Radar Network Design

View More View Less
  • 1 National Severe Storms Laboratory, Norman, OK 73069
© Get Permissions Rent on DeepDyve
Restricted access

Abstract

Observing programs utilizing Doppler radar must have them deployed in optimum locations to best satisfy experimental objectives and maximize economies. One wishes to determine the coordinate triples (xi, yi, zi), where i equals the number of radars, which maximize the value of the data to be collected. The optimum location is governed by a value or objective function. Here, possible networks of two to nine radars are given for two different error specifications. The objective functions with both error distributions maximize the quantity (AREAL COVERAGE/ERROR). The procedure is to search the finite number of local maxima for the global maximum in the value of the objective function. This is done by employing a searching algorithm at each of a number of starting vectors which are close enough to the local maxima to converge to the desired local maxima. In all cases, the network obtained by considering all radars simultaneously is superior to that obtained by combining optimum smaller sub-networks. Our results suggest the expected benefits for networks with additional constraints, reflecting the more complex experimental objectives particular to some individual field program. For example, the number of radars needed and their optimal configuration can be determined for a field program requiring a specified areal coverage (probability that a desired event will occur) and resolution (to retrieve a specified scale of motion).

Abstract

Observing programs utilizing Doppler radar must have them deployed in optimum locations to best satisfy experimental objectives and maximize economies. One wishes to determine the coordinate triples (xi, yi, zi), where i equals the number of radars, which maximize the value of the data to be collected. The optimum location is governed by a value or objective function. Here, possible networks of two to nine radars are given for two different error specifications. The objective functions with both error distributions maximize the quantity (AREAL COVERAGE/ERROR). The procedure is to search the finite number of local maxima for the global maximum in the value of the objective function. This is done by employing a searching algorithm at each of a number of starting vectors which are close enough to the local maxima to converge to the desired local maxima. In all cases, the network obtained by considering all radars simultaneously is superior to that obtained by combining optimum smaller sub-networks. Our results suggest the expected benefits for networks with additional constraints, reflecting the more complex experimental objectives particular to some individual field program. For example, the number of radars needed and their optimal configuration can be determined for a field program requiring a specified areal coverage (probability that a desired event will occur) and resolution (to retrieve a specified scale of motion).

Save