Sharon Perkins Buck
Master's Thesis submitted to the Faculty of the Virginia Tech in partial fulfillment of the requirements for the degree of
Master of Science
Biological Systems Engineering
Mary Leigh Wolfe, Chair
Frank E. Woeste
Donald M. Vietor
John V. Perumpral, Dept. Head
January 30, 1997
A probabilistic risk assessment (PRA) for the discharge of excessive nitrogen from nonpoint sources (NPSs) to a stream was performed for a small agricultural watershed in northern Virginia. Risk, by definition, is the product of the frequency of occurrence of an event and the consequences of that event. The purpose of this research was to determine the probability of occurrence of a nitrogen discharge event (i.e., frequency). The consequences of such a discharge event were not explicitly determined but were implicitly assumed to be negative in nature.
An event tree was developed to show the basic hydrologic processes at work in a small watershed. However, the event tree could not be used to discover the causes for nitrogen loss from the watershed. Therefore, a fault tree was developed for excessive nitrogen discharge in surface runoff on any day from agricultural sources. The development of the fault tree was found to be a useful exercise in understanding the intricate cause and effect relationships between agricultural practices and NPS pollution. Based on the results, the fault tree methodology might be used as an effective teaching or communication tool.
The fault tree was also evaluated quantitatively to determine a probability of occurrence for excessive nitrogen discharge to the stream on any day. Land use, fertilization, monitoring, and long-term weather records were used in conjunction with scientific judgment and expert opinion to establish the probabilities within the fault tree and to calculate the overall probability of nitrogen discharge to the stream on any day. The results obtained from the fault tree calculations tend to underestimate the importance of cropland best management practices (BMPs) over the long term, because the fault tree was developed on a daily basis (i.e., every day in a year has the same probability of a discharge event occurring). A more accurate depiction of the NPS pollution control problem was achieved by assuming the occurrence of a runoff event. A second fault tree was presented for the discharge of excessive nitrogen to the stream during a runoff event. The quantitative assessment of the new fault tree showed more clearly the impact of BMPs on reducing the likelihood of nitrogen discharge. A 0.15 decrease in the probability of nitrogen discharge during a runoff event was calculated for the Owl Run watershed from 1987 to 1993 due to the effects of BMPs installed during that time period. A 0.20 decrease was calculated for an Owl Run subwatershed for the same time period. This subwatershed isolated two major dairy operations and the effects of the BMPs installed for those dairies.
Despite the success of the fault tree in mirroring changes within the watershed, the amount of data and time required to perform the quantitative assessment may limit its use in the NPS pollution control field. The basic nature of the fault tree technique also limits its usefulness in the field. One such limitation is that degrees of events cannot be expressed. For example, a BMP is either present or not present on a fault tree. There can be no indication of how effective the BMP is in preventing NPS pollution without substantially increasing the level of detail displayed by the tree. Another limitation is that the ultimate result of the fault tree calculations is a probability of occurrence. This value is not as easily understood as the output of NPS pollution computer models, for example, where the output has specific meaning and units (e.g., milligrams of nitrogen per liter of runoff). The qualitative fault tree, however, has the advantage over computer models when it comes to understanding the concepts behind the technique and being able to see the cause and effect relationships at work in the watershed. Laypersons can understand the fault tree more easily than the complex computer code and intricate equations of models.
Send Suggestions or Comments to email@example.com