2018 Trend Analysis for PA Surface Waters Prepared by: Kevin McGonigal Susquehanna River Basin Commission 4423 North Front Street Harrisburg, PA 17110 Introduction Water quality in rivers and streams undergoes continuous change for many reasons including natural changes in hydrology. Trend analysis is a statistical technique used to identify whether water quality values have increased or decreased over an established time period. Twenty-two long-term Pennsylvania Department of Environmental Protection (PADEP) Water Quality Network (WQN) monitoring stations (listed in Table 1) within the Ohio, Susquehanna, and Delaware River Basins were analyzed for short- and long-term trends. Sampling frequency at individual sites ranged from six to 20 samples per year from 1998 to 2016. Two sites, Yellow Breeches and Penns Creek, are part of the Chesapeake Bay Program’s Non-Tidal Water Quality Monitoring Network (NTNWQN). Both NTNWQN sites included routine monthly samples and high flow event sampling. Samples were analyzed for the parameters listed in Table 2. Long-term trends were calculated for water years 1999- 2016. Short-term trends were calculated for water years 2007-2016. Additional analytical results can be found at the United States Geological Survey (USGS) water Quality Changes in the Nation’s Streams and Rivers web page (https://nawqatrends.wim.usgs.gov/swtrends/). Methods The primary tool for trends analysis was the Weighted Regressions on Time, Discharge, and Season (WRTDS) model (Hirsch et al., 2010). WRTDS requires a long-term flow record obtained from a paired USGS gage. Three of the 22 sites were not paired with a gage and thus were analyzed using the Seasonal Mann-Kendall trends test within the EXCEL stat package (XLSTAT, 2017). Using measurements of stream flow coupled with water quality results from sampled events, WRTDS estimates concentration and flux (also called load, which is the amount of a substance carried by flowing water) for each day across some period of time. Because water quality relates to water flow, WRTDS adjusts the measured variation in concentration and flux according to the stream flow variation to calculate Flow Normalized Concentrations (FNC) and Flow Normalized Flux (FNF). Trends were scored according to a WRTDS likelihood factor with values between 0.66 and 0.90 corresponding to “very likely” and values > 0.90 corresponding to “highly likely”. Trend directions are listed in Tables 6-9 with hollow triangles representing less likelihood, solid triangles representing more likelihood, and solid circles signifying no likelihood for trend. Cross hatches are shown for parameters where WRTDS could not be used due to either an excessive number of censored values (>50% below the method detection limit) or some other model validation issue. XLSTAT is an add-in for Microsoft Excel that was used to complete Seasonal Mann-Kendall analyses at Lackawaxen, Conemaugh, and Erie; i.e., stations where the flow record was insufficient for WRTDS analysis. Trends determined using seasonal Mann-Kendall are not adjusted for flow such that changes cannot be directly attributed to management action. XLSTAT was also used to analyze pH and specific conductivity at all sites. Significance was based on p-value <0.10, which corresponds to a WRTDS “highly likely” trend. Note that censored values were set to half of the detection limit. Trend results for both FNC and FNF were included in this report with loads presented as yields (the load divided by the watershed acreage) to allow for better site comparability. FNC and FNF trends
18
Embed
2018 Trend Analysis for PA Surface Waters · 2018 Trend Analysis for PA Surface Waters Prepared by: Kevin McGonigal Susquehanna River Basin Commission 4423 North Front Street Harrisburg,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
2018 Trend Analysis for PA Surface Waters
Prepared by: Kevin McGonigal Susquehanna River Basin Commission 4423 North Front Street Harrisburg, PA 17110 Introduction Water quality in rivers and streams undergoes continuous change for many reasons including natural changes in hydrology. Trend analysis is a statistical technique used to identify whether water quality values have increased or decreased over an established time period. Twenty-two long-term Pennsylvania Department of Environmental Protection (PADEP) Water Quality Network (WQN) monitoring stations (listed in Table 1) within the Ohio, Susquehanna, and Delaware River Basins were analyzed for short- and long-term trends. Sampling frequency at individual sites ranged from six to 20 samples per year from 1998 to 2016. Two sites, Yellow Breeches and Penns Creek, are part of the Chesapeake Bay Program’s Non-Tidal Water Quality Monitoring Network (NTNWQN). Both NTNWQN sites included routine monthly samples and high flow event sampling. Samples were analyzed for the parameters listed in Table 2. Long-term trends were calculated for water years 1999-2016. Short-term trends were calculated for water years 2007-2016. Additional analytical results can be found at the United States Geological Survey (USGS) water Quality Changes in the Nation’s Streams and Rivers web page (https://nawqatrends.wim.usgs.gov/swtrends/). Methods The primary tool for trends analysis was the Weighted Regressions on Time, Discharge, and Season (WRTDS) model (Hirsch et al., 2010). WRTDS requires a long-term flow record obtained from a paired USGS gage. Three of the 22 sites were not paired with a gage and thus were analyzed using the Seasonal Mann-Kendall trends test within the EXCEL stat package (XLSTAT, 2017). Using measurements of stream flow coupled with water quality results from sampled events, WRTDS estimates concentration and flux (also called load, which is the amount of a substance carried by flowing water) for each day across some period of time. Because water quality relates to water flow, WRTDS adjusts the measured variation in concentration and flux according to the stream flow variation to calculate Flow Normalized Concentrations (FNC) and Flow Normalized Flux (FNF). Trends were scored according to a WRTDS likelihood factor with values between 0.66 and 0.90 corresponding to “very likely” and values > 0.90 corresponding to “highly likely”. Trend directions are listed in Tables 6-9 with hollow triangles representing less likelihood, solid triangles representing more likelihood, and solid circles signifying no likelihood for trend. Cross hatches are shown for parameters where WRTDS could not be used due to either an excessive number of censored values (>50% below the method detection limit) or some other model validation issue. XLSTAT is an add-in for Microsoft Excel that was used to complete Seasonal Mann-Kendall analyses at Lackawaxen, Conemaugh, and Erie; i.e., stations where the flow record was insufficient for WRTDS analysis. Trends determined using seasonal Mann-Kendall are not adjusted for flow such that changes cannot be directly attributed to management action. XLSTAT was also used to analyze pH and specific conductivity at all sites. Significance was based on p-value <0.10, which corresponds to a WRTDS “highly likely” trend. Note that censored values were set to half of the detection limit. Trend results for both FNC and FNF were included in this report with loads presented as yields (the load divided by the watershed acreage) to allow for better site comparability. FNC and FNF trends
present different ways to look at the data and are not always in agreement. Typically, when they were in opposition, one trend had higher confidence than the other. Table 3 aggregates sites by river basin (Ohio, Susquehanna, and Delaware) to display the average short- and long-term changes in yield and load. Tables 4 and 5 show the changes in yields for individual sites. Tables 6-9 show trend directions and Tables 10-12 show individual site concentrations, yields, and percent change. Figures 1-4 display the short-term trends geographically. Results When aggregating results to the river basin level, long-term and short-term trends in TN, hardness, and iron were all downward. Short-term alkalinity trends in the Delaware were decreasing (driven by Schuylkill) while all others were increasing. Short-term trends for TNH, TSS, and zinc were all downward while TP were all upward. Distinct trends for the Ohio included upward calcium, magnesium, and sulfate in both the long- and short-term. The same parameters were downward in the Susquehanna and Delaware. Short-term TNO and OP trends were upward in the Ohio while long-term trends for all river basins were downward. The Susquehanna had both long-term and short-term upward trends in lead and a shift in aluminum from an upward long-term trend to a downward short-term trend. Aluminum and lead were both downward in the Ohio with no trends in the Delaware. TOC showed upward trends in the Susquehanna and no trends elsewhere. The most consistent trend found in all basins and all sites was an increase in laboratory-derived pH. Upward trends in field pH were only found at 9 of 22 sites. Schuylkill, Wissahickon, and Connoquenessing had the most short-term downward trends including nitrogen and phosphorus species and iron. Beaver had a slightly different trend pattern with nitrogen and metals all decreasing and phosphorus and orthophosphate increasing. Many other sites showed increasing phosphorus trends in the short-term while simultaneously having downward nitrogen trends including Lehigh, Yellow Breeches, Tioga, Monongahela, and Beaver. Aughwick, Mahoning, and Slippery Rock had upward short-term trends in nitrogen and phosphorus while Mahoning had downward trends for all metals. Tioga, Youghiogheny, Dunkard, Allegheny, Beaver, and Mahoning all had downward trends for aluminum and iron while Loyalsock was the only site with both increasing. Slippery Rock also showed upward trends for nitrogen, phosphorus, and TSS. Schuylkill, Dunkard, Beaver, and Connoquenessing all showed downward trends for TSS. The most consistent improvements in water quality over the short-term included decreases in aluminum, iron, magnesium, lead, and zinc at Mahoning and Beaver and short-term decreases in total nitrogen and phosphorus at Schuylkill, Wissahickon, and Connoquenessing. Most consistent degradations include increases in total nitrogen and phosphorus at Aughwick, Mahoning, and Slippery Rock. Unique trend patterns:
- Connoquenessing and Schuylkill had the most short-term decreases including TN, TP, TSS, metals, and ions
- Consistent nitrogen and phosphorus decreases: Schuylkill, Wissahickon, and Connoquenessing
- Nitrogen improvements and phosphorus degradations: Yellow Breeches, Lehigh, Tioga, Monongahela, Dunkard, and Beaver
- Mahoning improvements in all metals and degradations in total nitrogen and total phosphorus
- Beaver improvements in all metals and total nitrogen but degradations in phosphorus - Schuylkill had the only decrease in alkalinity - Iron increasing at Aughwick, Loyalsock, Monongahela, and Shenango - Magnesium increasing at Wissahickon, Monongahela, and Allegheny
The “strongest” trends occur where both FNC and FNF were the same direction with high confidence for both long-term and short-term. This occurred at Schuylkill, Wissahickon, and Monongahela for total nitrogen and at Williamsport and Monongahela for TNO where all trends were improving. Wissahickon also had all downward trends in total phosphorus and orthophosphate. Orthophosphate trends were all downward at Tioga and upward at Lehigh. Alkalinity had upward trends at all Ohio River basin sites except for Mahoning and Connoquenessing. Sulfate trends were all decreasing for Beaver, Shenango, Mahoning, Connoquenessing, and Slippery Rock. Connoquenessing also had all downward trends for calcium and hardness. Dunkard showed increases in lab and field pH and specific conductivity. All trends at Beaver were downward for aluminum, magnesium, lead, and zinc. All trends were downward for magnesium at Connoquenessing and Slippery Rock.
Table 1. Pennsylvania Water Quality Network Stations Analyzed
WQN Name Gage Square Miles Watershed Sampling Date Samples/Year
110 Schuylkill 1474500 1,893 Delaware 10/1/1998 -
09/2016 12
115 Wissahickon 1474000 64 Delaware 10/1/1998 -
09/2016 6
123 Lehigh 1453000 1,279 Delaware 10/1/1998 -
09/2016 6
147 Lackawaxen 1432110 589 Delaware 10/1/1998 -
09/2016 6
212 Breeches 1571500 216 Susquehanna 10/1/1998 -
09/2016 20
229 Penns 1555000 301 Susquehanna 10/1/1998 -
09/2016 20
249 Aughwick 1564500 205 Susquehanna 10/1/1998 -
09/2016 6
324 Tioga 1518700 446 Susquehanna 10/1/1998 -
09/2016 6
402 W Br. Williamsport 1551500 5,682 Susquehanna 10/1/1998 -