To assess the feasibility and reliability of high spatial resolution QuickBird satellite imagery data in field scale applications, a three-year field campaign was conducted to provide ground-based measurements of canopy biophysical and biochemical characteristics. Image-based algorithms for atmospheric correction were evaluated and improved to retrieve surface reflectance. The potential of using QuickBird reflectance for determining leaf area index (LAI) and variations in canopy nitrogen (N) conditions was investigated. Broadband spectral vegetation indices (VIs) were evaluated to identify the best VI for retrieving LAI. High resolution LAI was retrieved with identified VI from QuickBird imagery and validated with ground measurements. The effectiveness of QuickBird VIs in detecting variability in plant N status was compared with the petiole sampling method and the leaf chlorophyll meter. The results indicated that the imagebased model was effective for the visible bands, but not for the near infrared (NIR) band. A contour map was developed to interpolate atmospheric transmittance for clear days under average atmospheric aerosol conditions. With the interpolated atmospheric transmittance, the accuracy of NIR band reflectance was significantly improved. All selected VIs were well correlated with LAI but with different efficiencies in estimating LAI. The modified soil-adjusted vegetation index (MSAVI) proved to be the best LAI estimator. QuickBird-derived LAI with MSAVI-LAI relationships agreed well with ground-measured LAI in both absolute values and spatial variability. QuickBird images acquired about one month after emergence were able to detect the same N treatment variations detected with petiole NO3-N concentrations and SPAD meter readings. However, treatment differences in VI value were insignificant when LAI reached large values. Based on high-frequency measurements of thermal infrared surface temperature in another field campaign, different methods were compared for estimating evaporation coefficient and quantifying soil evaporation. The results showed that the integration of remotely sensed surface temperature into physically based algorithms considerably improved the accuracy of the estimation. In summary, high resolution QuickBird imagery had great potential to be incorporated into image-based remote sensing approaches for site-specific crop management, and high-frequency thermal infrared data could be integrated into evaporation estimation for soil moisture assessment.
展开▼