Skip to main content
ARS Home » Research » Publications at this Location » Publication #112141
Title: MAPPING GRAIN SORGHUM GROWTH AND YIELD VARIATIONS USING AIRBORNE MULTISPECTRAL DIGITAL IMAGERY

Author
item YANG, CHENGHAI - TX A&M EXP'T STATION
item Everitt, James
item Bradford, Joe
item Escobar, David

Submitted to: Transactions of the ASAE
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/5/2000
Publication Date: N/A
Citation: N/A

Interpretive Summary: Use of remote sensing technology for precision farming has been steadily increasing over recent years. Remote sensing has many attributes that are attractive for precision agriculture. The most important is its ability to acquire timely imagery over an entire field or multiple fields during the growing season. In this study, airborne digital imagery and yield monitor data were used to map plant growth and yield variability. Images were acquired from a grain sorghum field five times during the 1998 growing season, and yield monitor data were also collected from the field during harvest. The images clearly reveal plant growth patterns over the growing season. Statistical analyses showed there exist strong correlations between grain yield and image data for each of the five dates; however, images obtained around the crop peak growth were a better yield indicator. Yield maps generated from the images agreed well with a yield map from the yield monitor data. These results indicate that airborne digital imagery can be a very useful data source for detecting plant growth and yield variations for precision farming.

Technical Abstract: Remote sensing is becoming an increasingly important tool for detecting plant growth and yield variations in precision agriculture. In this study, airborne digital imagery, in conjunction with yield monitor data, was used to map plant growth and yield variability. Color-infrared (CIR) images were acquired from a grain sorghum field five times during the 1998 growing gseason, and yield monitor data were also collected from the field during harvest. The images were georeferenced and then classified into zones of homogeneous spectral response using unsupervised classification procedures. The images and unsupervised classification maps clearly reveal the consistency and change of plant growth patterns over the growing season. Correlation analyses showed grain yield was significantly related to each of the three band components (near-infrared, red and green) of the CIR images and the normalized difference vegetation index (NDVI) for the five dates. Stepwise linear regression was also used to relate yield to the three bands for each of the five dates, and the three images obtained at and after the peak growth produced higher R-squared values (0.64, 0.66 and 0.61) than the other two early season images (0.39 and 0.37). Yield maps generated from the three best images agreed well with a yield map from the yield monitor data. These results demonstrate that airborne digital imagery can be a very useful data source for detecting plant growth and yield variability for precision agriculture.