Precision Agriculture, 1, 291᎐300 1999
ᮊ 2000 Kluwer Academic Publishers. Manufactured in The Netherlands.
Relationship Between Spectral Data from an Aerial
Image and Soil Organic Matter and
GARY E. VARVEL,
MICHAEL R. SCHLEMMER, AND JAMES S. SCHEPERS
USDA-ARS and Dep. of Agronomy, Uni
. of Nebraska, Lincoln, NE 68583
Abstract. Early ventures into site-specific management involved fertilizer management decisions based
on soil chemical properties characterized by some form of grid sampling. This is both labor and capital
intensive and practitioners quickly began investigating other methods to get a measure of spatial
variability. Aerial photographs, which were mainly used to evaluate and assess crop status, allow for the
collection of whole-field data at relatively low cost. Our objective is to determine what relationships
exist between aerial spectral data and intensive grid soil test results and whether this information can be
used to improve future soil sampling strategies. Soil-test organic matter OM and Bray-1 P concentra-
tions were measured on soil samples taken using an alternating 12.2- by 24.4-m grid in late March 1994
from a quarter section under center pivot irrigation. Spectral data were collected in the spring of 1996
prior to planting using a multispectral network of digital cameras. Correlations of brightness values
from the blue, green, and NIR bands with both OM and Bray-1 P were significant, but relatively low.
Normality tests revealed that brightness values for the spectral data sets were generally evenly
distributed while those of the soil test OM and Bray-1 P were positively skewed. Many of the very high
soil-test data values were due to past management. When those values were removed from the database,
greater correlations between spectral data and soil test data were obtained. These results substantiated
that aerial imagery can be used to improve sampling strategies, but it must be used in conjunction with
existing knowledge and past management histories.
Keywords: Grid sampling, soil test, brightness values, skewness, kurtosis
Early approaches to site-specific management involved some form of grid soil
sampling to characterize soil chemical properties. Collection and analysis costs
make this approach impractical on a large scale. Furthermore, variable-rate appli-
cation of nutrients based on grid soil sampling results often fails to correct spatial
variability in crop growth and yield. This situation has prompted interest in other
methodologies to characterize spatial variability.
One alternative was to use information in published soil surveys of soil types as
the basis for sampling strategies. This procedure tends to group similar soils based
on soil classification data and has met with varying degrees of success. Varvel et al.
1981 indicated significant improvement in soil test P results and improved P
fertilizer recommendations when soils were sampled by soil series based on soil
Joint contribution of USDA-ARS and the Nebr. Agric. Res. Div., Journal Series No. 12213.
* Corresponding author.