Utilizing green normalized difference vegetation indices (GNDVI) for production level management zone delineation in irrigated corn.
T.M. Shaver, R. Khosla, and D.G. Westfall. Colorado State University, 1170 Campus Delivery, Fort Collins, CO 80523
Precision farming has been a major research focus of agronomists for over a decade. Much of this research has been directed towards enhancing the efficiency of overall farm inputs (e.g., fertilizers, herbicides, insecticides, water) without negatively impacting farm profitability and the environment. Of the many technological advancements made in precision farming, variable rate technology (VRT), arguably has had the greatest impact. In order to implement VRT a prescription map that accurately quantifies in-field variability must be generated. There are numerous methods available to create VRT prescription maps; these methods often include grid soil sampling and collecting in-season plant samples. These methods are often cost prohibitive, labor intensive, and destructive to the crop. Therefore, a cost effective alternative to prescription VRT maps that describes in-field variability is needed. One such alternative is production level management zones (MZs), which quantifies in-field variability and is based on soil color from aerial photographs, topography, and the farmers' management experiences. Production level management zones generally divide a field into areas of high, medium, and low yield potential. Studies have shown that MZs are effective in delineating in-field soil variability. However, an important component of the delineation process is aerial imagery. There may be instances where this imagery is unavailable or too costly to obtain. With this in mind, we designed a study to determine if remotely sensed plant canopy data acquired using Green Seeker GNDVI (Green Normalized Difference Vegetative Index) active remote sensors can be used to assist in MZ delineation. These indices are much easier and less expensive to obtain than aerial imagery. Studies have shown that vegetative indices relate well to grain yield, therefore, they should relate well to production potential. This study was conducted at two research sites in 2004 (Sites 1 and 2) and two research sites in 2005 (Sites 3 and 4) totaling 4 site years in eastern Colorado, USA. Each research site was previously classified into three site-specific MZs by identifying areas of high, medium, and low-productivity potential. Assignment of each zone was accomplished using the aforementioned MZ delineation technique. These MZs have been shown to correlate with grain yield in previous studies. In order to determine if GNDVI can be used to aid in the delineation of MZs, readings were collected across all three MZs at each site at three corn growth stages (V8, V12, and V16) and were then analyzed to determine if they were related to the actual zones. Results show that GNDVI readings collected at the V12 corn growth stage ranged from 0.398 in the low MZ to 0.696 in the high MZ. At this growth stage GNDVI differences between the high and low MZs correlated significantly with the actual MZs in 3 out of the 4 site years. Significant correlations between the high and medium MZs were only present in 1 out of the 4 site years. This was also the case when examining differences between the medium and low MZs with only 1 out of 4 site years showing significant correlation to the actual MZs. Readings collected at the V8 and V16 growth stages showed no correlation with management zone at any site in 2004 or 2005. The results of this study suggest that GNDVI readings collected in corn at the V12 growth stage relate well to previously delineated high and low production level MZs. Whether or not these readings alone would be enough to accurately delineate MZs has yet to be determined, however this study has demonstrated that there is potential for GNDVI readings to aid in the delineation of production level MZs.