|Abstract or Summary
- This study evaluates the Earth Resources Technology Satellite-
One (ERTS-1) multispectral scanner (MSS) as a means of predicting
lacustrine trophic state and the magnitude of selected trophic state
Numerical classificatory methods are employed to ascertain the
trophic character of 100 lakes in Minnesota, Wisconsin, Michigan,
and New York using the trophic indicators: chlorophyll a, conductivity,
inverse of Secchi disc transparency, total phosphorus, total
organic nitrogen, and an algal assay yield. A complete linkage clustering
algorithm is first used to examine the lakes for natural clusters.
The hyper-dimensional cloud of data points is then reduced in dimensionality
through the ordination technique of principal components
analysis. The first three principal components explain, respectively,
68 percent, 14 percent, and 8 percent of the variation in the data; the
three-dimensional ordination is expressed in the form of a "ball and
wire" model. The two complementary classificatory techniques show
the existence of poorly defined clusters. A multivariate trophic state
index (PC1) is derived from the principal component analysis.
A binary masking technique is used to extract lake-related MSS
data from computer-compatible digital magnetic tapes. Data products
are in the form of descriptive statistics and photographic concatenations
of lake images. Linear digital contrast stretching is demonstrated.
MSS color ratio regression models are developed for the prediction
of Secchi disc transparency and chlorophyll a in selected
Minnesota and Wisconsin lakes. The models give good estimates of
Secchi disk transparency and fair estimates of chlorophyll a levels.
Lake area estimates derived from MSS pixel counts are
generally within ten percent of area values obtained from topographic
maps. The synoptic view of the sensor is conducive to lake enumeration
The trophic state of lakes, as defined by lake position on the
first principal component axis (PC1 value), is predicted using MSS
color ratio regression models; each date of ER TS-1 coverage has its
own model. There is a general tendency for the MSS ratios (GRNRED,
GRNIR1, GRNIRZ, REDIR1, and REDIRZ) to decrease as the manifestations
of eutrophication become more evident. A less rigorous approach to the study of MSS data-lacustrine
trophic state relationships is undertaken using three-dimensional
color ratio models. The mean IR1 intensity levels of several "hypereutrophic"
lakes exceed their mean RED levels and effectively isolate
them from other lakes in the models.
An automatic image processing technique is employed to
classify a group of Wisconsin lakes using MSS colors (GRN, RED,
and IR1) in conjunction with the lakes' trophic state index values.
The results are depicted in the form of both gray-scale and colorenhanced
The ERTS-1 MSS has utility in the assessment of the lacustrine
resource. Its usefulness is most apparent when the seasonal contrasts
between lakes at different points on the trophic scale are at a maximum.
Excessive cloud cover, faulty or missing MSS data, and the need for
some ground truth impair, but do not preclude, its use in lake monitoring
The use of computer-compatible tapes in conjunction with digital
image processing techniques is essential if the maximum benefits are
to be derived from the ER TS-1 MSS in lake-oriented studies.