River Nitrogen Loads and Landscape Evapotranspiration as Influenced by Climate and Land Cover Changes in the Midwestern United States

Thumbnail Image

Persistent link to this item

View Statistics

Journal Title

Journal ISSN

Volume Title


River Nitrogen Loads and Landscape Evapotranspiration as Influenced by Climate and Land Cover Changes in the Midwestern United States

Published Date




Thesis or Dissertation


The Northern Gulf of Mexico is threatened by the yearly development of a hypoxic zone (dissolved oxygen <2 mg/L); also called the “dead zone”. The areal extent of the dead zone has been linked to increased outflow of freshwater and nutrients, especially soluble nitrogen from the Mississippi River Basin to the Gulf. Much of the expansion of the hypoxic zone has been blamed on land cover changes in the last two centuries; especially the installation of subsurface drainage and adoption of row crops over native vegetation. The research presented in this thesis set out to understand (1) how climate and land cover changes have impacted flow and N loads in rivers of the Midwestern United States; the main source of soluble N to the Mississippi River, and (2) the extent of changes in landscape evapotranspiration (ET) from adoption of corn and soybeans relative to native prairies. In the first chapter of the thesis, we evaluated the role of precipitation and land cover changes on streamflow, baseflow, flow weighted N concentrations (FWNC), and N loads in various rivers of the Midwestern United States. This quantification was done using the stepwise regression analysis both at the annual and monthly scales. For annual analysis, the predictor variables were the Ln(annual streamflow), Ln(annual baseflow), annual FWNC and Ln(annual N loads) and the explanatory variables were the annual precipitation in the current year (P1), and previous years (P2, P3), and area under soybean production (Soy). For monthly analysis, the predictor variables were the Ln(monthly streamflow), Ln(monthly baseflow), monthly FWNC, and Ln(monthly N Loads) and the explanatory variables were the area under soybean production, precipitation in the previous year, and precipitation of the current, and all previous months up to that point in the season. The annual analysis showed that streamflows and baseflows were often controlled not only by the current year precipitation, but also by the previous year precipitation. The previous year precipitation effects were in terms of increased or decreased soil wetness (fillable porosity). In some instances, previous year precipitation effects on streamflow or baseflow lingered for up to 2-3 years. The area under soybean production was generally not a significant explanatory variable in the annual analysis likely due to its small variations during the study periods. In the monthly analysis, precipitation in both the current month and previous months along with the previous year precipitation were important in controlling Ln(monthly streamflow), Ln(monthly baseflow), and Ln(monthly N loads). Area under soybean production was significant in some months but had much lower statistical power than the precipitation variables. Besides the direct effects of precipitation (increased or decreased streamflow and baseflow) on N loads, there were also some indirect effects. These included substantial N left in the soil profile during dry years, which on subsequent normal precipitation years lead to a large spike in river N loads. Conversely, there was less N left in soils in wet years, which lessened the N losses in subsequent years. Analysis of the combined data from seven rivers in the Midwestern United States showed similar relationship between Ln(annual streamflow) with current year and previous year precipitations and the area under soybean production. However, the area under soybean production only explained 1% of the variability in Ln(annual streamflow). Compared to Ln(annual streamflow), the combined plots of Ln(annual baseflow) versus annual precipitation showed some differences among watersheds. These differences appeared to be related to differences in landscape steepness, extent of tile drainage, soil water holding capacity, etc. among various watersheds. Because of the differences in annual baseflow among various rivers, there were also some differences in Ln(annual N yield) versus annual precipitation relationships among various rivers. These differences appeared to be related to watershed characteristics such as steepness and soil available N both from fertilizer input and mineralization from soil organic matter. This analysis showed that less water percolation through the soil will help stretch the N yield versus precipitation curves such that there are much lower N losses even at higher annual precipitations. Two possible ways to reduce water percolation is through increased overland flow (via surface inlets) and through introduction of high ET crops. However, the presence of surface inlets in the landscape increases the potential of more sediment and phosphorus losses to rivers. High ET crops include Miscanthus and switch grass. However, there is conflicting information in the literature on ET from these crops relative to corn and soybeans. Furthermore, there are challenges in establishing the Miscanthus in cold climate due to winterkill. Another way to reduce N losses to rivers may be through less N input on the landscape. However, applied inorganic N fertilizer was not a significant variable in explaining variability in river N loads. This may be partially due to limited variation in applied N fertilizer rates as well as limited number of years over which both the N loss and N fertilizer data were available. In the second chapter, we evaluated the evapotranspiration from irrigated and non-irrigated corn, non-irrigated soybeans, and native prairies at various stage of management in Western Minnesota. ET evaluation was done using the “Mapping Evapotranspiration at High Resolution using Internalized Calibration” (METRIC) model for the 2015 growing season. The inputs included satellite imagery and weather data. The results from this study showed that estimated ET from 8 June through 30 September 2015 followed the trend: wetland (671 mm)> non-irrigated corn (627 mm)>irrigated corn (601 mm)>non-irrigated soybeans (534 mm)>previously burned prairie (532 mm)>recently burned prairie (397 mm). Comparatively for the period 1 May to 30 September 2015, the estimated ET of wetland, previously burned and recently burned prairie grasses were 810 mm, 633 mm and 485 mm, respectively. These modeling estimates were similar to the literature values and showed that ET of native prairie grasses is nearly similar or marginally higher than that of corn and soybeans. Considering that soybeans also replaced low transpiring small grains such as oats and wheat suggests that large changeover of vegetative cover replacing native prairie and small grains with soybeans starting in 1940s in the Upper Midwestern United States likely had a very small to minimal impact on landscape evapotranspiration.


University of Minnesota M.S. thesis.June 2018. Major: Water Resources Science. Advisor: Satish Gupta. 1 computer file (PDF); xiv, 185 pages.

Related to



Series/Report Number

Funding information

Isbn identifier

Doi identifier

Previously Published Citation

Suggested citation

Baeumler, Nathaniel. (2018). River Nitrogen Loads and Landscape Evapotranspiration as Influenced by Climate and Land Cover Changes in the Midwestern United States. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/200157.

Content distributed via the University Digital Conservancy may be subject to additional license and use restrictions applied by the depositor. By using these files, users agree to the Terms of Use. Materials in the UDC may contain content that is disturbing and/or harmful. For more information, please see our statement on harmful content in digital repositories.