Features of the
e experimental design features of a monitoring program include spatial and
temporal coverage, control and reference sites, number of samples needed,
Experimental Design
preliminary sampling, and properties of estimators.
Experimental Designs
Monitormg designs that
a control and a rrearment are discussed by
Spooner et al. (1985) and Spooner (1991).
Paired Watershed Design
Due to the presence of an experimental control for year-to-year hydrologic
variability, the most effective practical design for monitoring BMP effectiveness
is the paired watershed design. This design consists of monitoring downstream
The most effective practical
from two or more drainages where at least one drainage has BMP implementation
(treatment) and at least one does not (control). The paired drainages must have
design for monitoring BMP
similar precipitation and precipitation response patterns. In addition, the paired
is the paired
watersheds should be relatively homogeneous with similar land use. The two
watersheds do not have to be identical, but their paired watershed measurements
watershed design.
must be highly correlated. Figure 4.1 shows monitoring stations for an idealized
station
be used to monitorthe control
paired watershedstudy.
site and monitoring station B would be used to monitor the treatment watershed.
Ideally the paired watershed design has the following characteristics: a) simulta-
neous (i.e., paired) monitoring below each drainage; b) monitoring at all sites prior
Figure 4.1. Paired watershed
design.
to any land treatment to establish
relative responses of the drainages (calibra-
tion or pre-treatment period); and c) subsequent monitoring where at least one
drainage area continues to serve as a control (i.e., accounts for climatological
variability) throughout the land treatment period.
The calibration period is generally one to three years, depending on cropping
patterns and the number of runoff events. The calibration period should include
a full range of weather conditions to reduce the possibility that a post-implemen-
tation event will be out of the range of the calibration equation.
4.8
Previous Page
