Page Banner

United States Department of Agriculture

Agricultural Research Service

Title: Predicting in-Season Nitrogen Status of Hard Red Spring Wheat Using Multispectral Satellite Imagery

Authors
item Long, Daniel
item Westcott, M. - MONTANA STATE UNIV.
item Whitmus, J. - MONTANA STATE UNIV.

Submitted to: Agronomy Abstracts
Publication Type: Abstract Only
Publication Acceptance Date: July 29, 2004
Publication Date: October 31, 2004
Citation: Long, D.S., Westcott, M.P., Whitmus, J.D. 2004. Predicting in-season nitrogen status of hard red spring wheat using multispectral satellite imagery. Agronomy Abstracts.

Technical Abstract: Late-season topdressing of nitrogen (N) fertilizer may further improve yield and quality of final grain. The objective of this study was to determine whether information from remote sensing is useful to decide whether late-season topdressing of N is needed. Relationships between image spectral reflectance, and flag leaf nitrogen (FLN), leaf chlorophyll (LC), or leaf area index (LAI) were investigated in Montana wheat fields. Coefficients for the correlation between spectral reflectance, and FLN were significant, but moderate in value (r less than 0.63) and favored the NIR band in all farm fields. The association between LAI and all spectral bands was relatively strong (r less than 0.7) indicating that spectral reflectance was also a function of crop density, especially in the dryland fields where crop growth was water limited. The regression of FLN on each of the individual spectral bands accounted for up to 75% of the variance in this crop attribute, but farm field exlained much of the variance that had been included in a model to partial location effects. Coefficients of multiple determination (R^2) were less than 0.34 indicating that only a modest amount of the variance in FLN or LC could be explained by spectral bands as for multiple regression models selected by backward elimination. Drought and luxury soil N limited the crop's response to applied N and apparently reduced the ability to predict crop N. When applied to a validation data set, the regression models produced SEP values that ranged from ±0.32% to ±0.58% for FLN and from ±1.88 to ±7.02 SPAD meter units for LC.

Last Modified: 11/1/2014
Footer Content Back to Top of Page