Change in spatial variability structure of NDVI readings related to observation scale
In: Precision Agriculture ‘05epena0@uky.edu
Search for other papers by E.M. Pena-Yewtukhiw in
Current site
Google Scholar
PubMed
Search for other papers by G.J. Schwab in
Current site
Google Scholar
PubMed
Search for other papers by O. Wendroth in
Current site
Google Scholar
PubMed
Search for other papers by L.W. Murdock in
Current site
Google Scholar
PubMed
Search for other papers by T. Stombaugh in
Current site
Google Scholar
PubMed
Purchase instant access (PDF download and unlimited online access):
The efficient use of real-time canopy sensors for the variable rate application of nitrogen (N) requires knowledge of the scale (resolution) of the variation in the measured property. Knowing the amount of optical data needed per unit area for efficient fertilizer application requires that we know or estimate the most efficient combination of physical sensor density (number of sensors along the applicator boom) and sensor output density (sensor readings per unit distance/area). However, reducing the density of sensor and their output would reduce the capital cost of N applicators. The objective of this study was to test the number of sensors and sampling grid size that will adequately describe field variation in canopy NDVI (normalized difference vegetative index). The NDVI data were collected in February, 2004, on wheat at the Zadoks Stage 26 growth stage. The spatial structure of NDVI was characterized by variogram analysis. Tested grid sizes ranged from 0.56 to 5.06 m2. Variograms for high density data sets were compared with variograms obtained with decreasing numbers of sampling points (greater grid size). It was possible to decrease data density while increasing grid size from 0.56 m2 to 5.06 m2, without affecting the field’s NDVI spatial structure. The nugget, range and sill values were maintained across the evaluated grid sizes. We conclude that it is possible to increase the effective grid size to 5.06 m2, both by decreasing the number of sensors along the toolbar and by increasing the time interval in sensor data acquisition, implying that farmers could achieve optimal N applications using less capital intense machinery, and with less resolution constraints.