Submitted to: Airborne Science Workshop
Publication Type: Proceedings
Publication Acceptance Date: March 5, 2002
Publication Date: N/A
Interpretive Summary: The strawberry spider mite, Tetranychus turkestani, is the most serious of the mite pests on cotton in the San Joaquin Valley in California. When mites feed on plants, they cause wounds that result in leaf puckering and discoloration in early stages of infestation and leaf drop later. This leads to yield reduction. It was hypothesized that the reddish discoloration that appears on the upper surface of the leaves and the patterns mites cause in the field could be detectable with remote sensing. Using NASA's AVIRIS sensor, ground-based images, and an image analysis technique called spectral mixture analysis, mite-damaged cotton was successfully differentiated from healthy cotton and soil in USDA-ARS research plots at Shafter, California. Additionally, the relative percentage of mite, soil, and healthy plants was estimated in the resulting images. The spatial nature of the images would permit a farm manager to locate the identified infestations in a field and apply the correct amount of pesticides or biological control agents. This could lower input costs and benefit the environment by reducing the amount of pesticide applied. Thus, growers, consumers, and the environment will benefit from this research.
Technical Abstract: Spectral mixture analysis and hyperspectral remote sensing are tools new to precision agriculture that have the potential to detect and identify various crop stresses and other plant and canopy characteristics. One such stressor, the strawberry spider mite (Tetranychus turkestani), is a serious pest in California cotton causing reddish discoloration on leaf surfaces. To determine the feasibility of detecting the damage caused by this pest, AVIRIS imagery was collected from USDA-ARS cotton research fields at Shafter, CA in 1999. Additionally, cotton plants and soil were imaged in situ with a liquid crystal tunable filter camera system. Spectra were collected in 10 nm increments from 450-1050 nm for four scene components: mite-damaged areas on leaves, healthy leaves, tilled shaded soil, and tilled sunlit soil. These spectral signatures (endmembers) were used in a linear spectral mixture analysis procedure to unmix the AVIRIS data to produce abundance maps for each endmember. The procedure successfully distinguished between a field that had been kept mite-free and an adjacent field where mites were allowed to infest the cotton crop. Areas of relatively greater mite damage (abundance) within the mite-infested field were also identified. Therefore, this technology has potential for identifying spatial extent and severity of within-field anomalies and stresses for use in precision agriculture.