Skip to main content
ARS Home » Pacific West Area » Maricopa, Arizona » U.S. Arid Land Agricultural Research Center » Water Management and Conservation Research » Research » Publications at this Location » Publication #185960
Title: SPIDER MITE DAMAGE MAPPING AND COTTON CANOPY BIOPHYSICAL CHARACTERIZATION USING SPECTRAL MIXTURE ANALYSIS

Author
item Fitzgerald, Glenn
item MAAS, STEPHEN - TEXAS TECH UNIV
item Pinter Jr, Paul
item Hunsaker, Douglas - Doug
item Clarke, Thomas

Submitted to: Meeting Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 9/4/2005
Publication Date: 11/15/2005
Citation: Fitzgerald, G.J., Maas, S.J., Pinter Jr, P.J., Hunsaker, D.J., Clarke, T.R. 2006. Spider mite damage mapping and cotton canopy biophysical characterization using spectral mixture analysis. Proceedings of the Sumposium on American-Taiwan Agricultural Cooperative Projects, Taipei, Taiwan, Nov. 15, 2005. CD-Rom unpagenated.

Interpretive Summary: Detecting pest infestation and measuring canopy characteristics are important for managing crops effectively. Remotely-sensed imagery of agricultural fields can provide useful information about crop cover, crop height, soil, and stressed or diseased plants. Detection and mapping of these characteristics can improve crop management and be used as inputs for precision application of pesticides for control of pests as well as inputs to crop and irrigation models and water and nitrogen stress detection. Constructing such maps is now feasible through the use of advanced image processing, which can estimate the type and amount of these characteristics within an image. An advanced processing technique, called Spectral Mixture Analysis (SMA) was applied to two different sets of images. In one experiment, measurement of strawberry spider mite (Tetranychus turkestani U.N.) damage in cotton was investigated. This pest causes leaf puckering and reddish discoloration, reducing boll yield. Analysis showed that damage from this pest could be successfully mapped across large research plots. In a second study, SMA was shown to successfully quantify cotton cover, height, and chlorophyll a concentrations, which are important inputs for irrigation scheduling models and quantifying nitrogen stress. These results show that advanced image processing techniques can improve our ability to interpret remotely-sensed imagery and have potential value to provide farm and land managers with useful management information.

Technical Abstract: Spectral mixture analysis (SMA) is an analytical tool that can identify and measure the components that make up mixed pixels. Used with hyperspectral remotely-sensed (HRS) imagery, SMA permits mapping of surface components from research plot to landscape scales. These technologies are being used increasingly in agronomic research and have potential for precision agriculture to map spatially and temporally variable crop conditions and stresses. SMA was applied to two data sets collected from different hyperspectral imagers. For both studies, radiometrically-calibrated crop and soil spectra were collected using ground-based spectroradiometers and imagers to provide reference spectra. In one experiment, strawberry spider mite (Tetranychus turkestani U.N.) damage in cotton was mapped. This pest causes leaf puckering and reddish discoloration and can reduce boll yield. Analysis showed that pest damage could be successfully identified, the relative abundance mapped in large research plots, and temporal changes in damage were consistent with known ground conditions. One confounding factor was shadows which interfered with proper abundance estimation of mite damage. To understand the effects of shadows on estimating canopy characteristics, a second study was conducted where spectra of sunlit and shadowed leaves and sunlit and shadowed soil were measured in cotton research plots. The intent was to determine whether SMA and HRS could simultaneously measure percent cotton cover, height, and chlorophyll concentration and whether the influence of shadows could be isolated. Results showed that including shadows as reference spectra improved relationships between imagery and ground measurements. Mapping sub-pixel landscape components based on spectral signatures could be useful for crop and ecosystem models, irrigation scheduling, locating water and nitrogen stress, and simultaneous measurement of key plant and crop parameters.