We report the preliminary results using airborne remote sensing images for citrus pest stress monitoring in Lower Rio Grande Valley of South Texas. In order to accommodate the in-field spectral variability, unsupervised classification is applied. In addition, fully constrained linear unmixing is performed at the sub-pixel level to quantify the stress severity. The results using multispectral and hyperspectral images are compared, which demonstrates the potential improvement that hyperspectral imaging can provide. In conjunction with variable rate technology in pesticide adoption, tree-specific stress information derived from remote sensing imagery will support a well-targeted pest management plan for cost effectiveness and environmental friendliness.