Submitted to: International Geoscience and Remote Sensing Symposium Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 6/8/2006
Publication Date: 7/31/2006
Citation: Walthall, C.L., Daughtry, C.S., Vanderbilt, V., Pachepsky, L., Lydon, J., Erbe, E., Higgins, M., Bobbe, T. 2006. Detection of illegal cannabis cultivation using remote sensing. Proceedings of the International Geoscience and Remote Sensing Symposium, July 31-August 4, Denver, Colorado. 2006 CDROM. Interpretive Summary:
Technical Abstract: Detection of illegal Cannabis cultivation by law enforcement agencies currently relies on low flying aircraft manned by trained aerial spotters. This is physically tiring for the aircrew, inefficient for large or complex landscapes, and is often foiled by camouflaged grow sites. A solution for detecting illegal Cannabis cultivation using remote sensing would be of considerable assistance and may increase the probability of detection. If multiple plants are grown in large homogeneous plots and the location of at least one cultivation site is known, then detection can be considered a traditional spectral land cover/land use classification problem. Probability of success is principally dependent upon the degree to which the known grow site is representative of unknown sites, and the adequacy of the sensor system spatial resolution. However, most illegal cultivation sites in the U.S. are “distributed” with a few isolated plants interspersed with other plant species. For distributed cultivations, requirements for high spatial resolution imagery become paramount and an alternative signature approach may be warranted. Analysis of Cannabis diffuse, nadir spectral signatures using laboratory, field and airborne data show a lack of stable, unique absorption features to use as a universal reference signature. However, Cannabis leaves appear to have a specular reflectance feature that contributes to the “emerald green” appearance reported by aerial spotters. This specular signature may be worthy of exploitation for detection. The leaf shape and plant architecture of Cannabis also offer potential spatial signatures that can be quantified using image texture analysis. These spatial signatures also warrant further investigation.