The time evolution of constituents in runoff: Sampling, combinatorial analysis, and scale response

Paul Stephen Miller, Purdue University

Abstract

The primary objective for the collection of water quality constituent data is the quantification of the mass load leaving the system under analysis. The usefulness, representative quality, and information content of water quality data and their calculated mass loads are often questioned but are difficult to quantify. The objectives of this proposed work are to quantify the effect that common sampling procedures may have upon mass load estimation, to examine the sensitivity of mass load calculation from collected constituent data, and, finally, to understand the statistical aspects of constituent mass load calculation as affected by changes in scale. To extract this information from watershed water quality data sets, the general approach of the work was to reduce fine resolution simulation and observed water quality data in a generalized, systematic fashion. This reduction in information is representative of sampling schemes and patterns commonly found in the practice of environmental monitoring of watersheds; however, the methodologies presented in this work are applicable to any data set from any source used to calculate loading rates in conjunction with appropriate watershed hydrology data. By reducing the information contained within a data, an error distribution is calculated. This distribution can then be summarized statistically and probabilistically. Results indicated that frequency and type of sampling scheme significantly affected mass load accuracy on simulated data. The appropriate scheme should be chosen based upon project goals balanced by economic constraints. Historical water quality data were systematically processed applying a destructive combinatorial algorithm for many different constituents at multiple scales. The results of the analysis revealed a significant correlation between increased sampling frequency, mass load estimation accuracy, and decreased risk at all scales with convergence occurring more rapidly using the volumetric criterion. The analysis technique was then utilized to determine the influence of watershed scale on mass load estimation accuracy with results indicating that, although mass load magnitudes are significantly affected as scale increases, when standardized, very little scale influence is exhibited in the accuracy of mass load estimation.

Degree

Ph.D.

Advisors

Mohtar, Purdue University.

Subject Area

Agricultural engineering|Environmental engineering

COinS