Multiyear Crop Monitoring Using Polarimetric RADARSAT-2 Data

• Published in: IEEE transactions on geoscience and remote sensing Vol. 51; no. 4; pp. 2227 - 2240
• Main Authors: Chen Liu, Jiali Shang, Vachon, P. W., McNairn, H.
• Format: Journal Article
• Language: English
• Published: New York, NY IEEE 01.04.2013; Institute of Electrical and Electronics Engineers
• Subjects: Agriculture; Agronomy. Soil science and plant productions; Applied geophysics; Backscatter; Biological and medical sciences; Data acquisition; Earth sciences; Earth, ocean, space; Exact sciences and technology; Fundamental and applied biological sciences. Psychology; Generalities. Biometrics, experimentation. Remote sensing; Internal geophysics; Monitoring; multitemporal; polarimetric synthetic aperture radar (PolSAR); Remote sensing; Scattering; Springs; Synthetic aperture radar; Canada; Ontario; Eastern Canada; maps; Space remote sensing; fine-grained materials; accuracy; backscatter; decomposition; classification; North America; agriculture; multitemporal; maximum likelihood; growth; radar methods; Synthetic aperture radar; polarimetric synthetic aperture radar (PolSAR)
• ISSN: 0196-2892; 1558-0644
• DOI: 10.1109/TGRS.2012.2208649

This paper studies the feasibility of monitoring crop growth based on a trend analysis of three elementary radar scattering mechanisms using three consecutive years (2008-2010) of RADARSAT-2 (R-2) Fine Quad Mode data. The polarimetric synthetic aperture radar analysis is based on the Pauli decomposition. Multitemporal analysis is applied to RGB images constructed using surface scattering, double-bounce, and volume scattering, along with intensity analysis of these scattering mechanisms. The test site is located in Eastern Ontario, Canada where the cropping system is dominated by corn, spring wheat, and soybeans. Each crop has unique physical structural characteristics which provide different responses for these scattering mechanisms. Significant changes occur in these scattering mechanisms as the crops move from one phenological stage to the next. By monitoring these changes over the season, the crop growth cycle from emergence to harvest can be observed. When harvest occurs, the backscatter intensities change significantly, and these changes aid in identifying crops. The temporal evaluation of the intensity of the scattering mechanisms generally track the measured leaf area index and observed phenological plant development. Changes in growth stage are crop type specific. Thus, to monitor changes in crop phenology and the occurrence of harvest activities, knowledge of the crop grown in any particular field is required. To accommodate this requirement, a maximum likelihood classification was performed on the R-2 data to produce a crop map. An overall classification accuracy of 85% was achieved.