Sampling strategies for the estimation of salt and nitrate loads in irrigation return flows: La Violada Gully (Spain) as a case study

Appropriate estimates of salt and nitrate loads exported by irrigated agriculture are needed to ascertain the potential negative impacts on the quality of the receiving water bodies. The irrigation return flows of the La Violada irrigation district (NE Spain), collected by La Violada Gully, were monitored for 4 years (1995-1998) in order to estimate salt and nitrate loads. The existence of a daily cycle in the Gully flow during the April to September 1995 irrigation season, associated with a cycle in the electrical conductivity, made it necessary to analyse the influence of the sampling hour on the daily estimates of salt and nitrate loads. Interpolation and regression methods were evaluated using nine 2-days surveys with samples taken at all even hours. The interpolation approach proved to be reliable for daily load estimations, whereas the regression approach was unacceptable due to the poor salinity-flow and nitrate concentration-flow relationships. The sampling time was important for the determination of salt load by the interpolation method, but resulted almost irrelevant for the estimation of nitrate load. Although the biases represent a small percentage of the mean (< 7% for both salt and nitrate load estimates), the lack of consideration of the sampling hour will result, depending on the time of sampling, in a systematic over- or under-estimation of the measured loads. The number of samples needed to estimate salt and nitrate loads within a pre-specified margin of error was determined using the 1995 and 1996 irrigation seasons series and confirmed through a Monte Carlo test for the four irrigation seasons. A systematic sampling interval of 9 days (i.e. 21 samples along the irrigation season) allowed estimating salt loading within a 5% error and nitrating loading within a 10% error at the 95% confidence level.