|GE, SHAOKUI - UC BERKELEY, CA
|GONG, PENG - UC BERKELEY, CA
Submitted to: International Journal of Remote Sensing
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/24/2005
Publication Date: 3/1/2006
Citation: Ge, S., Carruthers, R.I., Reddy, A.M., Gong, P. 2006. Texture analysis for mapping Tamarix parviflora using aerial photographs along the Cache Creek, California. International Journal of Remote Sensing. 114:65-83.
Interpretive Summary: Aerial photography was used to assess the invasion of the exotic shrub Tamarix parviflora (saltcedar) along an environmentally sensitive riparian zone in Northern California. This is one of the worst exotic and invasive species infesting riparian areas in several western states. Basic color photographs allowed saltcedar to be visually identified during periods of bloom but were not easily able to be classified just based on color characteristics alone. Texture analyses were then conducted over a range of different levels of resolution to determine if classification could be improved and the process automated. The final method allowed highly accurate classification of saltcedar along a 40 km section of Cache Creek. These results have been useful by local landowners and conservancy managers in implementing control procedures for this invasive species, and can be used at other locations to identify saltcedar infestations.
Technical Abstract: Natural color photographs were used to detect the coverage of tamarix parviflora and other vegetation along a 40 km portion of Cache Creek in Northern California in 2001. Color aerial photos were digitized and georeferenced. Eight types of ground cover (tamarix patches, crops, roads, rocks, water bodies, evergreen trees, non-evergreen trees and shrubs (excluding tamarix)) were chosen from the digitized photos for separability analysis and subsequent supervised classification. Due to color similarities among the eight cover types, the average separability of the original data was very low. The separability was improved significantly through the inclusion of textures. Six types of texture measures with various window sizes were evaluated. The best textures were used as additional features for identifying tamarix. A total of 29 color photographs were processed to detect tamarix distribution using a combination of the original digital images and texture features. It was found that tamarix covered a total of 1.96 km2 in the study area. Through validation in downstream, middle-stream, and upstream areas respectively, the detection accuracy of this invasive shrub was over 80%. This study demonstrated that the area of tamarix invasion is strongly correlated to areas of both bare fields and water bodies.