|WU, JIXIANG - Mississippi State University|
Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: 4/3/2009
Publication Date: 1/15/2010
Citation: Willers, J.L., Wu, J., Jenkins, J.N. 2010. Categorical likelihood method for combining NDVI and elevation information for cotton precision agricultural applications. Proceedings Fifth International Workshop on the Analysis of Multi-temporal Remote Sensing Images, July 28-30, 2009, Mystic, CT. p. l.
Technical Abstract: This presentation investigates an algorithm to fuse the Normalized Difference Vegetation Index (NDVI) with LiDAR elevation data to produce a map useful for the site-specific scouting and pest management (Willers et al. 1999; 2005; 2009) of the cotton insect pests, the tarnished plant bug (Lygus lineolaris (Palisot de Beauvois) Heteroptera: Miridae) and the stinkbug complex (Heteroptera: Pentatomidae). This fused map is used to facilitate geographic sampling techniques in assessing the spatial abundance of these pests in Mid-South cotton fields. The algorithm is based upon a bi-variate Gaussian density distribution, followed by the application of nominal and ordinal categorical attributes to fuse on a pixel by pixel basis the raster representations of the NDVI and the elevation (m). During 2008, the fused map was used to sample a cotton field for tarnished plant bugs and stink bugs in each of four nominal habitat categories derived from the algorithm. Since crop phenology (captured by the NDVI component) is strongly affected by water availability, which is correlated with the elevation component, this fusion procedure is useful because only one map is necessary, instead of two separate maps for each component. Preliminary results after only one production season indicate that the fused raster data product results is a spatial and temporal way to better sample and economically control these cotton insect pests, while preventing loss in cotton yield, preserving fiber quality, and may be helpful in conserving beneficial insects.