254
Again we see the same general shaped curves as in
Figure 2 with the ERTS image curve being about 5% higher than
the laboratory curve. These shifts can be attributed to
atmospheric effects. For the lakes sampled it is apparent
that an excellent correlation exists between reflectance and
turbidity, solids, and the lake's classification.
Madison Area Lakes : During August of 1973, these
techniques were applied to a series of lakes in the Madison,
Wisconsin area. Water sampling was done almost simultaneously
with the 5 August 1973 ERTS overpass. The lakes near Madison
are generally warm water lakes, and in several lakes accelerated
eutrophication is occuring due to large nutrient influxes
from agricultural and urban storm runoff.
Water samples were collected and Secchi disc readings
taken to assure that the bottom reflection was not affecting
the images. The water samples were analyzed for turbidity,
solids and reflectance. ERTS image brightness values were
determined for all of the lakes in question.
This image analysis was accomplished for the overpass
of 5 August 1973 and for an overpass a year earlier in 1972.
Figure 4 represents two ERTS images of these lakes. Figure 5
shows the resulting correlation curves. The fact that the
curve for 5 August 1973 is higher than the curve for 28 August
1972 merely indicates that there were more atmospheric effects
on 5 August 1973. The greater scatter for the 1972 image data
is because water sampling was not simultaneous with the 1972 ERTS
overpass.
Both the tannin lakes near Ely, Minnesota; and the
non-tannin lakes near Madison, Wisconsin, appear to have good
correlation between reflectance and algal-caused turbidity.
It appears that it is not possible to classify a tannin lake
with a curve for non-tannin lakes (See Figures 3 and 5). Like
wise it is not possible to classify a non-tannin lake by use
of a curve for tannin lakes. Further research is needed to
identify tannin and non-tannin lakes on an ERTS image before
they are classified as to eutrophic level. A method must also
be devised to eliminate the effects of bottom reflection in
shallow lakes and to identify those lakes where these effects
are significant. Weed mats also cause image interpretation
errors and some method must be developed for identifying or
handling such situations. It is hoped that by computer wave
length ratioing techniques all these types of lakes can be
identified, and once the identification is made they can be
appropriately classified as to eutrophic level.