Wednesday, November 15, 2006

Optimum Spatial Scale and Management Resolution for Corn (Zea mays L.).

Byungkyun Chung1, Brenda S. Tubana1, Brian Arnall1, Starr Holtz1, Olga Walsh1, Kent Martin2, Kefyalew Desta1, and William Raun3. (1) Oklahoma State Univ, Ag Hall, Stillwater, OK 74078, (2) Kansas State Univ, 112 Parkview Dr, Junction City, KS 66441, (3) 044 N Ag Hall, OK State Univ, Dept of Plant & Soil Sciences, Stillwater, OK 74078-0507

The optimum spatial scale at which corn (Zea mays L.) should be managed has not been evaluated. Identifying the correct resolution to manage sidedress nitrogen (N) application in corn can help improve N use and increase economic and environmental benefits. The objective of this study was to determine the optimum resolution for predicting corn grain yield using optical sensors.  Corn rows, 30 m in length were randomly selected at three locations where the exact location of each plant was determined.  In 2005 and 2006, 15 m rows were selected at two locations. At each location, 4 rows received no additional N and 4 rows were fertilized with 150 kg N/ ha as UAN (28-0-0).  A GreenSeeker™ optical sensor attached to a bicycle with a shaft driven encoder was used to measure NDVI from V6 to VT growth stages and record distance, thus matching exact NDVI readings with each plant.  The average NDVI for 1 to 15, 20, 25, and 30 plants and the average NDVI over fixed distances (0.2, 0.4, 0.6, 0.8, 1.0, 0.46, 0.91, and 2.7 m) was determined.  Individual corn plants were harvested by hand and grain yield was determined. Correlation of corn grain yield versus NDVI was evaluated over increasing distances and increasing number of plants.  No significant differences in resolution between fertilized and non-fertilized plots were found in 2005.  Coefficient of determination (R2) of NDVI versus grain yield were maximized at the V8 growth stage when NDVI and yield was averaged over every 5 plants and over fixed distances of 1 m in 2004 and every 7 plants and a fixed distance of 1.4 m in 2005.  Management decisions at resolutions greater than 1.4 m will likely ignore detectable differences in grain yield that exist at V8.