Statistical analysis of effects of measures against agricultural pollution

Abstract

The Norwegian Government has initiated a plan to reduce agricultural pollution. One of the projects in this plan is aimed at investigating the effects of different measures in order to evaluate their effects and costs.

A set of experiments has been designed to estimate the effects of measures to reduce or control the use of fertilizers and erosion. The project started in 1985. It comprises continuous measurements in two water courses in each of four counties: one test drainage area where the relevant measures were implemented at the end of 1986, and one reference area where no specific measures are carried out. A series of chemical parameters are measured together with runoff and other hydrological and meteorogical data.

The paper provides a preliminary analysis of the data collected in one of the counties during the period June 1985 to April 1988. It contains examples of analysis of covariance to show possible effects of the measures carried out in the test area.

Natural variations in precipitation and pollution are large, making it difficult to see the effects of the measures without using statistical techniques to take the multivariability of the problem into account. Some effects can be shown with analysis of covariance. However, the relatively short measurement period makes it neccessary to be careful when interpreting the results.