Compressed Sensing Applied to Weather Radar

TL;DR

This paper introduces a novel meteorological radar that employs compressed sensing with matrix completion algorithms to efficiently sample weather data without losing accuracy, addressing the challenge of non-sparse precipitation echoes.

Contribution

It extends compressed sensing techniques to volumetric weather radar data by utilizing matrix completion, overcoming the limitations of previous CS methods for non-sparse weather signals.

Findings

Successful application of matrix completion algorithms to weather radar data

Efficient sampling of precipitation echoes without accuracy loss

Demonstrated feasibility using Iowa X-band Polarimetric radar data

Abstract

We propose an innovative meteorological radar, which uses reduced number of spatiotemporal samples without compromising the accuracy of target information. Our approach extends recent research on compressed sensing (CS) for radar remote sensing of hard point scatterers to volumetric targets. The previously published CS-based radar techniques are not applicable for sampling weather since the precipitation echoes lack sparsity in both range-time and Doppler domains. We propose an alternative approach by adopting the latest advances in matrix completion algorithms to demonstrate the sparse sensing of weather echoes. We use Iowa X-band Polarimetric (XPOL) radar data to test and illustrate our algorithms.