ENV 455:  Environmental Health Seminar
Instructor: Charles V. Shorten, Ph.D., P.E.

Professor, Environmental Health Program
Department of Health, West Chester University, West Chester, PA 19383
Office: 319 Sturzebecker Health Sciences Center
Phone: (610) 436-2360/2125 Fax: (610) 436-2860  Email:   cshorten@wcupa.edu

A Linear Regression Study of Nitrate-Nitrogen and Phosphorus as a Function of Population in the Big Thompson Watershed, CO

Jeff Thomas
Email: JT300157@wcupa.edu

  • Abstract

  • Introduction

  • Methods

  • Results

  • Discussion

  • Conclusion

  • Literature Cited

  • Acknowledgements

  • Appendices

  • Abstract

    As population rises so does the amount of impervious surfaces. These surfaces cause storm water runoff to travel quickly to streams with increased loads of nitrite nitrogen and phosphorus among many other chemicals. The flux amounts of these chemicals should increase with the change in population. This increase in population is seen throughout the world with more and more people moving and building in cities. This can cause a severe upset to stream chemistry by adding in large amounts of non-point-source pollutants.

    In Colorado, the Big Thompson watershed runs through 4 cities that have experienced major growth in the past 30 years. With this expansion, the fluxes of nitrite nitrogen and phosphorus should increase. An evaluation of historical stream chemistry and census data did not support this relationship. The P-values of the line slope do not meet the .05 coefficient confidence level with some seasonal exceptions. Other factors have influenced flux levels of stream nutrients more than that of population and impervious surfaces. These may include, but are not bound to, more point source pollution, increased motor vehicle traffic, and larger area of agricultural development.


    Water quality in watersheds is a vary dynamic condition. The factors that will influence the levels of contaminants in rivers very from land use to point source pollution, from rainfall to seasons. These factors influence the amount of nutrients, like phosphorus, nitrate-nitrogen, and pollutants, like PCB’s, sediment, etc, loaded into a waterway. Once in the rivers and streams contaminates can limit the primary productivity and reproduction of many animals and plants by causing explosive growth of algae. Excess nitrate-nitrogen affects a river by stimulating massive growth of photosynthetic algae. As the alga dies and decomposes, with the help of aerobic decomposers, the dissolved oxygen level in the river will drop. This drop will disrupt the aquatic system and decrease biodiversity (Carpenter 1997, Smil 1997). Nitrite nitrogen has had a major effect on the estuary zone at the mouth of the Mississippi River. After years of high nitrate-nitrogen flux the estuaries in the Gulf of Mexico are severely hypoxic (Mitsch 2001). When phosphorus levels increase the amount of Cyanobacteria and algae will grow until phosphorus has been depleted. This can cause a mat of new growth on the surface that will block sunlight from reaching benthic plants (Carpenter 1997). The death of the benthic plants and decomposition of the new algae and cyanobacteria decomposing will cause the dissolved oxygen in the river to be decreased.

    The Big Thompson watershed is a river system in Colorado. This river system runs from Rocky Mountain National Park, at the continental divide, down the mountain passes though the city of Estes Park and out on the Great Plains again flowing though three other major cities; Loveland, Fort Collins and Greeley. A map of the area can be seen in Figure 1.  (Click on the figure to enlarge it.)

     COMap.jpg (42961 bytes)

    The total drainage basin is 835.15 square miles. The majority of the watershed is in Larimer County with the addition of the city of Greeley in Weld County. In the past 50 years each county has under gone high levels of growth.

    Urbanization has thrived in areas that have well defined waterways. This trend goes back to hundreds of years ago when water was a way of life. Historically water was a way of travel, source of power and sewer. In this century, however, our waterways have discovered a new threat. Impervious surfaces are threatening the land, aquifers, and rivers. Recent satellite images show that areas such as Baltimore and Washington DC are 10-15% impervious (Bodamer 2001). There is an increase in the area of impervious surfaces with an increase in population (Harbor 1994). A study done by Peter Guerrero (2001) found that in the last half of the 20th century urban development has increased by 327% and road area has been increased 278%. Denver alone has increased its impervious surfaces from 30 to 39% in the past 12 years (American Forests 2001).

    Water is able to soak into the spongy green layer when it rains in the forest. This water slowly percolates to the aquifer and to the streams and rivers. However, in the same rainstorm centered over a highly urbanized area (large percent impervious surface), the rain will follow the route city planners have decided it will take. Water will flow down over roofs, streets, and parking lots. This water will wash over the pollutants of our daily lives on these surfaces. These pollutants include pesticides from our lawns, hydrocarbons from our cars, particulate matter, and other chemicals from industry that are deposited on the ground. When the fast moving runoff flow picks up chemicals, it can act to drastically change the stream chemistry. If a major city experiences a dry spell and then receives a major storm, the run off can act as a dose of poison to a river (Benyus 1998).

    This paper presents an analysis of the effects of population growth in the Big Thompson watershed for the past 40 years on the fluxes of Nitrate-nitrogen (mg/s as N) and total phosphorus dissolved (mg/s as P). The analysis will be done though several linear regressions of the nutrient fluxes in correlation to population with the tenet of, impervious surface through population has an influence on phosphorus and Nitrate-nitrogen fluxes in the Big Thompson watershed.


    The USGS’s National Water Information System (NWISWeb) database (USGS 2001) provided historical nutrient data for the Big Thompson River from 1970-1995. The historical nutrient data for Nitrate-nitrogen and phosphorus was received from the gauging station number 6744000, near the city of La Salle. This station was chosen because it is located down stream from the major growth centers of the watershed. This station is a good monitoring point for the concentrations and fluxes of phosphorus and nitrate- nitrogen from the total watershed because La Salle is located at the terminal end of the river. Therefore, water sampled at this station will contain contaminates from the entire watershed. From the USGS’s National Water Information System (NWISWeb) database (USGS 2001) a historical flow rate was added for each day of a historical nutrient sampling. The flow rate was in ft3/Second while the nutrient data is in mg/L. To obtain the flux of the nutrients the formula is, flow multiplied by concentration. The resulting quantity is in mg/second.

    Population data was received from the US Census bureau (Census 2001). An overlay of the commissioner district map of Larimer county and a watershed boundary map shows that the 2nd and 3rd commissioner districts are comparable to the majority of the Big Thompson watershed. The final large city added into the population was the city of Greeley. The confined populations of the 2nd and 3rd commissioner districts of Larimer County and the city of Greeley comprised a close approximation of the historical population of the Big Thompson watershed.

    The analysis of the data was done using linear regressions on the full data set. This analysis is used because both population and nutrient fluxes are continuous functions. By comparing the two functions a statistical best fit line is described by a slope and a constant. The slope has a confidence interval known as the P-value. If this P-value is equal to or less than .05 the slope has a 95% confidence level and the correlation of population and nutrient fluxes is 95% confident that the slope is positive. The regression compared the populations of the watershed to the nutrient fluxes of, phosphorus, and nitrate- nitrogen. This raw data can be seen in Appendix A Table 1. To further analyze the data the database was divided into the different seasons. Seasons were as follows:

    A seasonal distribution is used since the average flow rates in each season had different trends seen in Table 1. Eight regressions were then done one for each nutrient for each season.

    Table 1 Average seasonal flow rates of the Big Thompson River at La Salle, Co in L/S






    Average Flow






    The null hypothesis is impervious surface through population have no influence on phosphorus and nitrate-nitrogen fluxes in the Big Thompson watershed. A 95 percent confidence rate was needed to reject the null hypothesis. In other words a P-value grater than or equal to of .05 was needed on the slope the line to reject the null.


    Census data for each decade were used to set the population size. Since the census is conducted only once every ten years, it is assumed that the population size from the mid-point of one decade to the mid-point of the next was equal to the size indicated by the size of the 1970, 1980, and 1990 census, as appropriate. The combined nitrate-nitrogen and, separately, the phosphorus flux rates were plotted as a function of the appropriate population size. These are shown for the entire data sets in Appendix B Figures 1 and 2. Least-squares linear regression (Microsoft Excel) was used to calculate the best fit line through these data; regression statistics are shown in Appendix B Tables 1 and 2. For the nitrogen relationship the equation of the line is y=0.0325x+3022.7. This relationship is significant; the p-value for the slope coefficient is 0.03477. The null hypothesis of no relationship between the nitrogen nutrient levels and the population size is therefore rejected. For the phosphorus relationship the equation of the line is y=0.0001x+11.795. This relationship is not significant; the p-value for the slope coefficient is 0.0504. The null hypothesis of no relationship between the nitrogen nutrient levels and the population size is therefore accepted. Phosphorus flux must have other factors than population that influenced the flux amount. These and other factors will be discussed later in the discussion section.

    In each season and each nutrient, the combined nitrate-nitrogen and, separately, the phosphorus flux rates were plotted as a function of the appropriate population size.

    These are shown for each seasonal data set in Appendix B Figures 3 - 10. Least-squares linear regression (Microsoft Excel) was used to calculate the best-fit line though these data; regression statistics are shown in Appendix B Tables 3 - 10.  For each season and nutrient the equation for the line of best fit had a coefficient of greater than 0.05. This trend can be seen in Table 2.

    The null hypothesis of no relationship between the nutrients and population size was therefore accepted. The only exemption to this statement was the regression for the winter season phosphorus flux, which can be seen in Appendix B Table 4 and Figure 4. In the winter season the coefficient of the line is 0.001249. This value rejects the null Hypothesis with a 99.875 confidence that population is correlated to the flux of phosphorous in the Big Thompson watershed.


    The increases in impervious surfaces have been linked to the increases in population. As populations increase the amount of land taken up by, roads, houses, parking lots, and industry increase. In a study done by Carpenter (1997) found an increase of population caused an increase in the impervious surface. The percentage increase of impervious surface can not be calculated since the type of land use predicts the amount of impervious surface used. For example, an acre of farmland has is much more permeable than an acre of dense city.

    Of the ten regressions preformed only two, the regression for total nitrite- nitrogen and the regression for phosphorus in winter rejected the null hypothesis. With a confidence level of 95%, population did not have a statistical effect on stream nutrients. However population can have an indirect effect on stream nutrients. Arnold and Gibbons (1996) wrote that populations cause increases in impervious surfaces. However impervious surfaces do not generate pollution, they are a critical contributor to the hydrologic changes in the degradation of the waterway. Many other factors may exist that influence the nutrients in the stream more than impervious surface did.

    In the regression regarding the total nitrite nitrogen the P-value was .003477. This value had a sufficient confidence level to reject the null hypothesis. In this instance the population/impervious surface did have an effect on the flux of the nitrite nitrogen. This supports the original hypothesis that impervious surface through population has an influence on nitrate-nitrogen fluxes in the Big Thompson watershed. When looking at the regressions season by season, however, there is no effect of population/impervious surface and flux levels. So as a whole the nitrite-nitrogen levels have increased due to population and other factors.

    The winter regression of phosphorus also was able to reject the null hypothesis with a line coefficient of 0.001249. While the total regression was not able to reject the null, the seasonal regression was able to. With a 99.875% confidence level the slope of the line of best fit was positive. In the winters, the total discharge of the river is typically less than that of the rest of the year. The average flow rates for Fall, Summer, Spring and Winter are 2742, 5456, 4721, and 2211 L/S respectively. In this situation an increase in the concentration of phosphorus would be less diluted and have a higher flux. Only in the winter months was this evident there for the flux of the nutrient is more dependent on the flow of the river in this case than the concentration of the nutrient. Further studies could be done to support this hypothesis.

    There are many other influences on the nutrient fluxes in a watershed other than impervious surfaces. These can include point source pollution, agriculture, silviculture, climate, weather and more. The first five have been shown to be the top influences on nutrient flux, along with impervious surfaces. It is easy to see how point source pollution can increase the flux of nutrients. The constant adding of pollution to a river will cause an increase in the fluxes of nutrients. Regulations are established (CDPHE 2001) and enforced to make sure the increase is not of a significantly lethal amount. The Big Thompson watershed has very little point source discharges of nitrate-nitrogen and no reported discharges of phosphorus. The amount of nitrate-nitrogen added by industries has, in the past, rarely been greater than a total of 2542 lbs./year (EPA 2001). With such a small amount added directly from industries this nitrite nitrogen can be easily diluted and cause little to no effect on stream chemistry. Agriculture can add phosphorus to the soil through fertilizer and pesticides. A highly cultivated land with little to no buffers of runoff has been shown to markedly increase phosphorus levels in rivers (Suszkiw 2001). Another threat from large-scale farming is the nitrite nitrogen increase from pesticides and fertilizers on the crops. Silviculture has a net good effect on reducing stream nutrients. This effect comes from the filtering effect of open ground and the slowing of runoff water (Carmon and Shamir 1997). The open permeable ground will cause the water to filter down though the ground to recharge the aquifer. A possible negative effect can, however, come when large amounts of pesticides are used in silviculture. These chemicals again can leach into the ground causing increases in nutrient flux. Finally, climate and weather can play an important role in the flow of the river, which will have an effect on the dilution of a nutrient in that river. In the winter months the land is dry and cold; this causes less water to flow in the river and ice to form, which traps more water for release later in the winter when spring is near, thus cutting down on the dilution of nutrients. Weather is tied to climate. Typically in every region of the world a location will have its dry time of the year and its wet times of the year. The wet time in Colorado is in the spring when the ice and snow melt from the mountains and rain is prevalent. Summer has also been shown to be equally as wet as spring and even more so in some years. Fall and winter have flow rates that match each other to within 500 L/second of each other.


    The Big Thompson watershed statistically has had little contamination of nutrients from population increases. Nutrient flux data suggests no major effect of population from the cities on the watershed and the effects of the nutrients, nitrite-nitrogen and phosphorus. Impervious surface through population have no major influence on phosphorus and nitrate-nitrogen fluxes in the Big Thompson watershed. The findings are not conclusive to support that population and impervious surfaces have no effect on increasing stream nutrients in the Big Thompson River. However, Colorado is an ever growing state. Its cities have plenty of room to grow. With good planing Colorado’s water will be clean and beautiful for generations to come.

    As urbanization continues to increase, impervious surface will continue to increase. With this increase in impervious surfaces, there will be an increase in the amount of nutrients in the rivers of America. These nutrients will come from the runoff of the cities. Without the natural filtering processes of an aquifer these nutrients will make their way directly to streams and rivers. With increased nutrients from an increase in urbanization there will be more concerns of the water quality of the nation’s waters. These concerns should be dealt with in a timely manner. The best protection is prevention. It can be more costly and less healthy to clean up after the fact than spending the extra time and money to prevent the contamination. New technologies like treatment of runoff and new permeable surfaces can have a positive impact on the water quality of the nations waters. The best way, however, to decrease runoff from impervious surfaces is planing the layout of new urbanization to maximize the permeable area and to create what Marwedel (1998) has called green areas. These green areas will let water take a more natural journey to the river. This journey will be slowed by aquifers and a general slowing of the runoff over natural surfaces, and not through manmade channels and pipes. When looking to the future of city development, planners should take advantage of all the tools new technology has provided to create a city that is not only beautiful but also healthy for the nearby rivers.

    Literature Cited

    American Forests: Developing Denver Needs Trees. American Forests. Vol. 107 Issue 2:16 (2001)

  • Arnold Jr., Chester L.; Gibbons, C. James: Impervious Surface Coverage; the Emergence of a Key Environmental Factor. J. of the American Planing Assoc. Vol. 62 Issue 2: 243-259 (1996)

  • Bodamer, David: New Satellite Maps Pinpoint Storm water Runoff. Civil Engineering. Vol. 71 Issue 7:33-34 (2001).

  • Booth, Derek B.; Leavitt, Jennifer: Field Evaluation of Permeable Pavement Systems for Improved Stromwater Management. J. of the American Planing Assco. Vol. 65 Issue 3:314-326 (1999).

  • Benyus, Janine. Saving for a Rainy Day. American Forests. Vol. 104 Issue 3:24-26 (1998)

  • Carmon, Naomi; Shamir, Uri: Water-Sensitive Urban Planning: Protecting Groundwater. J. of Environmental Planning and Management. Vol. 40 Issue 4:413-435 (1997).

  • Carpenter, S: Nonpoint Pollution of Surface Waters with Phosphorus and Nitrogen. Ecological Applications. Vol 8: 559-568 (1997)

  • Colorado Department of Public Health and the Environment (CDPHE) 2001,

  • http://www.cdphe.state.co.us/op/waterqualityregs.asp
  • Environmental Protection Agency (EPA), 2001 http://oaspub.epa.gov/surf/surffac?huc=10190006&ldip=17&name=Big%20Thompson

  • Guerrero, Peter: Better Data and Evaluation of Urban Runoff Programs Needed to ASSEDD Effectiveness. FDCH government Account Reports. 06/29/2001

  • Harbor J.,M. A Practical Method for Estimating the Impact of Land-Use Changes on Surface Runoff, Groundwater Recharge and Wetland Hydrology. J. of the American Planing Assoc. Vol. 60 Issue 1: 95-109 (1994).

  • Marwedel, James: Opting for Performance: An Alternative to Conventional Zoning For Land Use Regulations. J. of Planning Literature. Vol. 13 Issue 2: 220-232 (1998).

  • Mitsch, William J; Day, John W.; Gilliam, Wendell; Groffman, Peter M.; Hey, Donald L.; Randall, Gyles W.; Wang, Naiming: Reducing Nitrogen Loading to the Gulf of Mexico from the Mississippi River Basin; Strategies to Counter a Persistent Ecological Problem. BioScience. Vol 51 Issue 5: 373-388 (2001)

  • Smil, Vaclav: Global Population and the Nitrogen Cycle. Scientific American. July Vol 277 Issue 1: 76-82 (1997)

  • Suszkiw, Jan: Rainfall Simulators Target Runoff. Agricultural Research. October:18-20 (2001)

  • United States Census Bureau (Census), 2001 http://quickfacts.census.gov/qfd/states/08000.html

  • United States Geological Society (USGS), 2001 http://water.usgs.gov/co/nwis/qw

  • Vitousek, Peter M, et al: Human Alteration of the Global Nitrogen Cycle: Sources and Consequences. Ecological Applications. Vol 7: 737-748 (1997)

  • Acknowledgements

    • Dr. Charles Shorten for his guidance

    • Jim Galloway, Abby Hinkle, Michelle Filling, and Mike Wayock, for constructive criticism on reviews

    Thank you to these people, with out you I would not have completed this.