Data Collection: Data was collected from twenty-five randomly selected rivers between orders 1 and 5 (five from each order). Vernier probes and chemical kits were used to test the indicators. Three samples were collected for each of the first six indicators, either across the width of the river, or, in the case of narrow rivers, up the river, against the flow of the water. Probes were left in the water until the readings began to settle. Benthic data was also collected, using a net and two-minute agitation samples to collect a goal minimum sample of 200 organisms. These readings were recorded on data sheets and input into Fathom. The data was recorded as both an individual river file as well as a master data file.
After data collection, the class was split over how best to use the data to answer the essential question. Two methods were developed. Group A collated all the data first, using it to define each river and indicator as healthy or unhealthy. Group B sorted rivers as stressed or unstressed and then used statistical analysis to determine significant indicators.
Group A: Group A first compiled the data points of the ten indicators that were collected at twenty-five different rivers. A box-and-whisker plot was then used to determine outliers in the collected data of individual indicators. These outliers were tallied and then discarded. Additionally, any impossible readings, most notably negative turbidity readings, were discarded. Other indicators were also discarded because of a lack of data and inconsistent rating systems. “Healthy” standards for each indicator were then researched. For the purposes of this study, “healthy” was considered the point at which the level of the indicator would pose a risk to aquatic life. At this point, curves were created to rate each data point when compared to the established “healthy” standard. A scale of 0 to 100 was used (with 0 being the furthest possible from the “healthy” standard, 100 being the closest, and 50 being the cutoff for “healthy” and “unhealthy” data), and the curves were formed to fit each standard for each indicator. The scores were then compiled using two different methods and orders to reach four final scores.
Group B: Group B subjectively classified each river as either existing in a stressed or unstressed environment. This idea for procedure originated from a conversation with the Virginia Department of Environmental Quality (DEQ) employee and statistician, Jason Hill. Agricultural fields, roads, highways, parking lots, and steep banks conducive to erosion were all decided to be “stressors," or factors that would negatively impact the health of a river. Additional natural elements impacting the health of the river were also taken into account, such as the composition of the river bed, and the agricultural practice that surrounded the river. Rivers that were largely free of stressors were classified as unstressed. Two lists were compiled. The first was a list of the stressed (10) rivers and the second was a list of unstressed (15) rivers. For each list, the data for each of the ten indicators was compiled and an average was calculated. Box and whisker plots of the averages were used to compare the stressed and unstressed values of each indicator. They were also used to eliminate any outliers. A match paired statistical test was preformed on each indicator. It was concluded that nitrates, benthic, and DO were good indicators of river health based on their signifiant difference in stressed and unstressed values. We then compared the values of these three indicators with a modified rating of DEQ standards to determine the health of the Rivanna River Watershed.
After data collection, the class was split over how best to use the data to answer the essential question. Two methods were developed. Group A collated all the data first, using it to define each river and indicator as healthy or unhealthy. Group B sorted rivers as stressed or unstressed and then used statistical analysis to determine significant indicators.
Group A: Group A first compiled the data points of the ten indicators that were collected at twenty-five different rivers. A box-and-whisker plot was then used to determine outliers in the collected data of individual indicators. These outliers were tallied and then discarded. Additionally, any impossible readings, most notably negative turbidity readings, were discarded. Other indicators were also discarded because of a lack of data and inconsistent rating systems. “Healthy” standards for each indicator were then researched. For the purposes of this study, “healthy” was considered the point at which the level of the indicator would pose a risk to aquatic life. At this point, curves were created to rate each data point when compared to the established “healthy” standard. A scale of 0 to 100 was used (with 0 being the furthest possible from the “healthy” standard, 100 being the closest, and 50 being the cutoff for “healthy” and “unhealthy” data), and the curves were formed to fit each standard for each indicator. The scores were then compiled using two different methods and orders to reach four final scores.
Group B: Group B subjectively classified each river as either existing in a stressed or unstressed environment. This idea for procedure originated from a conversation with the Virginia Department of Environmental Quality (DEQ) employee and statistician, Jason Hill. Agricultural fields, roads, highways, parking lots, and steep banks conducive to erosion were all decided to be “stressors," or factors that would negatively impact the health of a river. Additional natural elements impacting the health of the river were also taken into account, such as the composition of the river bed, and the agricultural practice that surrounded the river. Rivers that were largely free of stressors were classified as unstressed. Two lists were compiled. The first was a list of the stressed (10) rivers and the second was a list of unstressed (15) rivers. For each list, the data for each of the ten indicators was compiled and an average was calculated. Box and whisker plots of the averages were used to compare the stressed and unstressed values of each indicator. They were also used to eliminate any outliers. A match paired statistical test was preformed on each indicator. It was concluded that nitrates, benthic, and DO were good indicators of river health based on their signifiant difference in stressed and unstressed values. We then compared the values of these three indicators with a modified rating of DEQ standards to determine the health of the Rivanna River Watershed.