Geoscience Reference
In-Depth Information
regression to be 0.0336. The resulting regression
had a standard error of 1.87 L kg
−1
or about a
third larger than for the K
oc
regression. Esti-
mates of K
d
from both equations are plotted ver-
sus measured K
d
in Figure 14.3. Both estimates
fall along the 1:1 line with similar scatter except
near high K
d
values where SOC gives a better
estimate of K
d
than does EC
a
, which accounts
for much of the smaller sum-of-squares.
Perhaps more important than overall agree-
ment between calculated and estimated K
d
val-
ues is whether or not the calculated values show
the same spatial patterns as the measured values.
The spatial distributions of K
d
calculated from
EC
a
and SOC are shown in Figure 14.4. Com-
pared to the measured K
d
values (Figure 14.2),
both calculated distributions are more diffuse
and do not reach the maximum values measured
within the Okoboji map unit (Figure 14.1). How-
ever, both calculated distributions accurately
recreate the low values of K
d
indicative of the
better drained Clarion map units. Underestima-
tions from EC
a
data of 3 L kg
−1
or more are con-
centrated within the Okoboji map unit and the
lower southwest corner of the grid. In contrast,
calculations from SOC greatly underestimate K
d
only in limited areas along the northern bound-
ary of the grid. Estimates based on EC
a
and SOC
both overestimate K
d
in the center of the inten-
sive grid in an area roughly bounded by Harps
soil that has a higher pH than the other soils
within the grid (Novak et al., 1997). Atrazine
sorption affinity has been shown to decrease
with increasing pH (Yamane and Green, 1972).
Neither SOC nor EC
a
alone captured this pH
interaction with atrazine K
d
.
Overall, EC
a
measurements provided reasonable estimates of K
d
, although not as accurate as
the standard procedure of estimating K
d
from SOC. Spatial patterns for estimated K
d
values were
also similar to measured patterns, although more diffuse, and estimates based on SOC were closer
to measured values in the regions of high K
d
. The real advantage to using EC
a
measurements to
estimate K
d
, however, is in the speed and ease of making many measurements over a wide area.
Measurements of EC
a
over the grid were made in about an hour, and SOC determinations took
many days of field and laboratory effort.
Once calibrated, maps of K
d
estimated from EC
a
surveys would be useful in determining the
leaching potential of herbicide applications for specific areas in fields. We can illustrate this by
mapping K
d
across the entire 32 ha field by using the EC
a
data collected along the transects and
the relationship between EC
a
and K
d
developed above. Using linear kriging to interpolate EC
a
for a
regular grid across the field and then applying the nonlinear regression relating K
d
to EC
a
, we can
develop a map of K
d
with little extra effort (Figure 14.5). This map clearly shows the spatial pat-
tern of K
d
variation across the field and could be used as the basis of computing a leaching risk for
atrazine using methods such as those proposed by Loague et al. (1990). Although not as accurate as
250
K
d
(L kg
-1
)
9.5
7.5
5.5
3.5
1.5
0
250
0
250
SOC (g kg
-1
)
52
42
32
22
12
0
0
250
250
EC
a
(mS m
-1
)
60
50
40
30
20
0
0
250
Easting (m)
fIGURe 14.2
Spatial distribution of (a) atrazine
sorption affinity, K
d
, (b) soil organic carbon fraction,
SOC, and (c) apparent electrical conductivity, EC
a
,
over the gridded area.
