The persistence of atrazine was monitored in three fields at different
sites in Spain during two consecutive years (1990 and 1991). Laborato
ry assays for determining the influence of temperature and soil moistu
re content on the rate of herbicide degradation were carried out on so
il samples from the same fields. The degradation constants derived fro
m these assays, together with weather records for the period of the fi
eld experiments, were used in a computer program which simulated herbi
cide persistence in the field. Some adjustments were made to adapt the
model to Spanish conditions. The model predicted with reasonable accu
racy the persistence of the herbicide in two soils, although there was
a tendency to overestimate the residues at early dates. Discrepancies
between predicted and measured residues were greater in the third soi
l, due to rapid initial losses that were not predicted by the program.
In this case, the agreement was improved if the program was run takin
g time zero to be one month after herbicide application. Possible reas
ons for these discrepancies are discussed.