B. Fagerberg et al., ESTIMATING REGIONAL CROP YIELDS IN SWEDEN WITH A SIMULATION-MODEL ANDFIELD OBSERVATIONS, Swedish Journal of Agricultural Research, 28(2), 1998, pp. 91-98
Yields of barley were predicted for four agricultural districts in cen
tral and southern Sweden during the years 1988-92. An observed time tr
end was corrected for yearly weather effects as predicted by the Europ
ean multicrop model WOFOST. Effects of pest damage estimated from smal
l plot experiments with pesticides, and from general large-scale area
surveys, were included. Original model parameters were used to predict
barley emergence and phenological development, but a parameter for th
e relative growth rate of leaf area in the model was reduced. Predicti
ons of yields for SKO-areas (approximately 30 x 30 km, Statistics Swed
en) with the same soil- and plant model parameters, were overestimated
in central and underestimated in southern Sweden. Compiling all yield
data correlation coefficients, I, for predicted versus observed yield
s ranged from 0.62 to 0.71, with no conclusive differences between are
a definition, method for calculation or method for estimating pest dam
age. Regressions as measured by r were improved when both weather and
pest damage were considered. The root mean error of prediction (RMSEP)
was also reduced, when both weather and pest damage were considered.