Precision agriculture using hyperspectral remote sensing and GIS

The objectives of this study were to utilize hyperspectral Real-time Data Acquisition Camera System (RDACS-3; 120 bands and 2×2m pixel resolution) imagery to examine spectrally sensitive regions for the detection of Nitrogen (N) deficiency in corn and to determine whether hyperspectral and/or multispectral remote sensing, and Geographic Information Systems (GIS) could be used to improve N management through early detection of vegetation stress. Several N studies with varying rates of N fertilizer were conducted in Calloway County, Kentucky, USA to examine the relationships between crop bio physical variables and crop stress. Multi-temporal Hyperspectral RDACS data were collected for the study area. Logistic Regression and Multiple Linear Regression techniques identified spectrally sensitive regions in blue, red and near-infrared wavelength regions of the Electromagnetic spectrum. These regions were modeled and compared with traditional hyperspectral and multispectral techniques. The results of these comparisons revealed the greater effectiveness of hyperspectral imagery feature selection in the shorter red region over the typical Normalized Difference Vegetation Index approach. Hyperspectral imagery with high spectral and spatial resolutions offers distinct advantages over multispectral data for early detection of stress in vegetation. The application of high resolution remote sensing in agriculture should improve fertilizer N use efficiency and reduce N losses to the environment.