Neighboring Ray Estimation Based Weather Radar Data Compression
Bong-Joo Jang1, Sanghun Lim1 and Keon-Haeng Lee1
1Korea Institute of Civil Engineering and Building Technology, 283 Goyangdae-Ro Ilsanseo-Gu Goyang-Si Gyeonggi-Do 411-712 Korea
In the past decades, with the introduction of Doppler and dual-polarization technology, weather radars have evolved significantly. The accuracy of interpretation of weather phenomena has also been improved with high spatiotemporal radar observation. Along with these developments in weather radars, amounts of data to be processed increase exponentially. In addition, for data fusion from networked radar system it is required to store and analyze huge volume data consistently.
Therefore, in terms of radar data gathering or its management, efforts to reduce the huge radar data volume are required to store and transmit the data efficiently. Currently most radar centers are using general compression method such as gz (Gzip library, 2015) to reduce the radar data volume size. Most compression algorithms for radar data had been proposed in two different perspectives. One is the lossless compression method that handles by treating the radar data as numerical data, and the other is the lossy compression using symbolization or quantization that handles by treating the radar data as a symbol or an image data. Näppi (1994) proposed a lossless compression for radar data using run-length encoding and VLC (variable length coding). This method has a significant meaning in calculating the entropy of radar data. Lakshmanan (2007) applied several lossless compression methods to store radar data and compare and analyze the methods about their performances and applicability for weather radar. Kruger and Krajewski (1997) proposed lossy compression using VLC for radar data after signal-to-noise thresholding in QC (quality control) processing. This method has high compression ratio as a lossy compression but compressed data cannot be used for radar maintenances such as radar calibration or modification of quality control because some noises and non-meteorological echoes necessary to radar system maintenance are removed from QC processing. Huang and Ai (2009) perform the lossy radar data compression by applying EZW (embedded zerotree wavelet) algorithm from JPEG2000 technique for image compression. This method also same disadvantages with Kruger's method because it regards the radar data as a quantized image data.
In this paper, we proposed a hierarchical compression method based on estimation between neighboring rays for weather radar data having high spatio-temporal resolution. The method is applied to radar reflectivity and evaluated in aspects of accuracy of quantitative rainfall intensity. The technique provides three compression levels from only one compressed stream for three radar user group-signal processor, quality controller, and weather analyst. Experimental results show that the method has maximum 13% and minimum 33% of compression rates, and outperforms 25% higher than general compression technique such as gz.
Fig. 1. Concept of the proposed radar data compression algorithm.
Keywords: Radar data compression; Data management; Weather radar; Ray estimation
Gzip library, accessed Jan., 21, 2015: http://www.math.utah.edu/docs/info/gzip_toc.html.
Huang, Y. and W. Ai, 2009: Weather radar data compression based on zerotree wavelet algorithm, Image and Signal Processing, CISP'09. 2nd International Congress on. IEEE, 1-4.
Kruger, A. and W. F. Krajewski, 1997: Efficient Storage of Weather Radar Data, Softw: Pract. Exper., 27, 623-635.
Lakshmanan, V., 2007: Overview of radar data compression, Proc. SPIE 6683, Satellite Data Compression, Communications, and Archiving III, 668308.
Näppi, J. J., 1994: Lossless Compression of Interpolated and Raw MRL-5 Weather Radar Data, Geophysica, 30, 93–105.