Integrating Field Spectra and Worldview-2 Data for Grapevine Productivity in Different Irrigation Treatments

Friday, 18 December 2015
Poster Hall (Moscone South)
Maimaiti Maimaitiyiming, Saint Louis University Main Campus, St. Louis, MO, United States, Arianna Bozzolo, University of Missouri, The Grape and Wine Institute (ICCVE), Columbia, MO, United States, Abuduwasiti Wulamu, Saint Louis University Main Campus, Saint Louis, MO, United States and Joseph L Wilkins, Saint Louis University, Saint Louis, MO, United States
Precision farming requires high spectral, spatial and temporal resolution remote sensing data to detect plant physiological changes. The higher spatial resolution is particularly as important as the spectral resolution for crop monitoring. It is important to develop data integration techniques between field or airborne hyperspectral data with spaceborne broad band multispectral images for plant productivity monitoring. To investigate varying rootstock and irrigation interactions, different irrigation treatments are implemented in a vineyard experimental site either i) unirrigated ii) full replacement of evapotranspiration (ET) iii) irrigated at 50 % of the potential ET. In summer 2014, we collected leaf and canopy spectra of the vineyard using field spectroscopy along with other plant physiological and nutritional variables. In this contribution, we integrate the field spectra and the spectral wavelengths of WorldView-2 to develop a predictive model for plant productivity,i.e., fruit quality and yield. First, we upscale field and canopy spectra to WorldView-2 spectral bands using radiative transfer simulations (e.g., MODTRAN). Then we develop remote sensing techniques to quantify plant productivity in different scenarios water stress by identifying the most effective and sensitive wavelengths, and indices that are capable of early detection of plant health and estimation of crop nutrient status. Finally we present predictive models developed from partial least square regression (PLSR) for plant productivity using spectral wavelengths and indices derived from integrated field and satellite remote sensing data.