Image use in the characterization of field parameters: incorporation of remote sensing with hydrologic simulation modeling

Abstract

The incorporation of hydrologic/crop growth models and remote sensing could lead to the development of improved precision farming systems. This research focused on strategies for incorporating these two individual devices. The first objective included obtaining sets of digital aerial images throughout the growing season of a corn crop and radiometrically correcting the images for comparison under a common illumination scene. Influential model parameters attainable through remote sensing were identified. The research evaluated the use of remote sensing in deriving vegetation parameters for the purpose of calibrating a hydrologic/crop growth model, and the use of remotely sensed soil estimates in improving the representation of the spatial heterogeneity within the field. Ultimately, the purpose of this research was to serve as a beginning step in incorporating remote sensing and hydrologic/crop growth models to improve simulation capability in accessing production, hydrologic response, and environmental sensitivities of watersheds. Results led to the development of a new radiometric correction procedure that made use of image soil lines in correcting the images to a reference scene. The procedure outperformed a standard radiometric correction procedure (histogram matching) in terms of matching the theoretically expected growth of vegetation away from image soil lines. Estimates of organic matter (OM) and cation exchange capacity (CEC) were derived from soil lines of bare soil images. Leaf area index (LAI) estimates were derived from the radiometrically corrected images to characterize growth of the corn throughout the growing season. The LAI measurements were then used to calibrate a model's (APEX) vegetation database for parameters simulating the development and shape of the sigmoid, S-shaped LAI development curve. Remotely sensed soil measurements were deemed important in model simulations. Predicted annual runoff and yield were significantly different for model simulations with and without the remotely sensed soil estimates, especially in fields with large variations in soil parameters. Overall, this research provided a template for the development of an eventual automated system for incorporating remote sensing and hydrologic simulation modeling.