A predictive methodology for the detection of vegetation stress using atmospherically corrected multispectral scanner imagery

Predictive modeling can optimize data collection for a specific task. Further, for existing imagery, modeling can optimize the selection and analysis of data from an existing database.;A priori modeling can predict optimal environmental conditions needed for data acquisition. Image data should be collected in conditions known to minimize the effects of the atmosphere and other environmental variables, and any quantitative image analysis must include corrections for atmospheric effects. Modeling methods can also help correct for atmospheric effects.;A predictive methodology is evaluated which optimizes the environmental parameters for image data collection and subsequent analysis and is based on the personal computer Sensor-Atmosphere-Target (SENSAT) (Richter 1991) model modified for use in the Windows operating environment (SENSWIN). This methodology is reviewed with reference to existing non-optimal image data.;Additionally, image processing of existing imagery is conducted to identify bands or linear combination of bands from those in the available dataset that best identify stressed vegetation embedded in an unstressed background.