Geoscience Reference
In-Depth Information
Observed vs. Predicted Yield
8
7
6
5
4
1:1 line
Observed = 1.13 × Predicted -0.70
R
2
= 0.57
3
3
4
5
6
7
8
Predicted Cotton Yield (Mg/ha)
fIGURe 16.3
Observed versus predicted cotton yield estimates using Equation (16.3). Dotted line is a
1:1 relationship. (Taken from Corwin, D.L., Lesch, S.M., Shouse, P.J., Soppe, R., Ayars, J.E.,
Agron. J.
, 95,
352-364, 2003. With permission.)
regression model, while all other parameters were significant near or below the 0.05 level. The
R
2
value for Equation (16.2) is 0.61, indicating that 61 percent of the estimated spatial yield variation
is successfully described by Equation (16.2). However, the residual variogram plot indicates that
the errors are spatially correlated, which implies that Equation (16.2) must be adjusted for spatial
autocorrelation.
Using a restricted maximum likelihood approach to adjust for spatial autocorrelation, the most
robust and parsimonious yield response model for cotton was Equation (16.3):
Y
= 19.28 + 0.22(EC
e
) - 0.02(EC
e
)
2
- 4.42(LF)
2
- 1.99(pH) + 6.93(θ
g
) + ε
(16.3)
Figure 16.3 shows the observed versus predicted cotton yield estimates from Equation (16.3).
Figure 16.3 suggests that the estimated regression relationship has been reasonably successful at
reproducing the predicted yield estimates with an
R
2
value of 0.57. Sensitivity analysis reveals that
LF is the most significant factor influencing cotton yield with the degree of predicted yield sensitiv-
ity to one standard deviation change resulting in a percentage yield reduction for EC
e
, LF, pH, and θ
g
of 4.6, 9.6, 5.8, and 5.1 percent, respectively. The point of maximum yield with respect to salinity is
calculated by setting the first partial derivative of Equation (16.3) to zero with respect to EC
e
, which
results in a value of 7.17 dS m
−1
, which is similar to the salinity threshold for cotton of 7.7 dS m
−1
reported by Maas and Hoffman (1977).
16.4 delIneAted SIte-SpeCIfIC MAnAGeMent UnItS
From Equation (16.3) and scatter plots of cotton yield versus properties (Figure 16.2), management
recommendations were made that spatially prescribed what could be done to increase cotton yield at
those locations with less than optimal yield. Four recommendations can be made to improve cotton
productivity at the study site: (1) reduce the LF in highly leached areas (i.e., areas where LF > 0.5),
(2) reduce salinity by increased leaching in areas where the average root zone (0 to 1.5 m) salinity is
>7.17 dS m
−1
, (3) increase the plant-available water in coarse-texture areas by more frequent irriga-
tion, and (4) reduce the pH where pH > 7.9.
