Excess nitrogen and phosphorus from numerous sources delivered through complex transport pathways have accelerated eutrophication and increased the incidence of harmful algal blooms, particularly in the coastal and Great Lakes states in the United States. To address these issues, many watershed models have been developed to simulate nutrient inputs and their consequent deliveries to aid in the establishment of water quality management plans. However, these models typically lack spatially explicit descriptions of nutrient sources and attenuation factors, and they often do not take the joint of surface and subsurface transport pathways into consideration. In addition, spatially explicit tile drainage areas are still lacking, which limits the accuracy of hydrology and water quality modeling. This dissertation presents a spatially explicit model to simulate nutrient loading, sources, and pathways annually (Chapter 2) and during two distinct hydrologic seasons (chapter 3), as well as uses machine learning algorithms that translate remotely sensed Earth observations and environmental datasets into tile drainage maps (Chapter 4). In Chapter 2, I enhance and apply the Spatially Explicit Nutrient Source Estimate and Flux (SENSEflux) model, to simulate nitrogen and phosphorus loads from the US Great Lakes Basin (USGLB) on an annual basis. The results show that agricultural sources are the dominant sources for both total nitrogen (58%) and phosphorus (46%) deliveries to the US Great Lakes. In addition, this study reveals that the surface pathways (sum of overland flow and tile field drainage) dominate nutrient delivery, transporting 66% of total nitrogen and 76% of total phosphorus loads to the US Great Lakes coastline. Building on the annual SENSEflux simulation in Chapter 2, Chapter 3 focuses on seasonal variations in nutrient fluxes, sources, and pathways. Two distinct hydrologic seasons were examined, the first being snowmelt where high flow conditions are present in early spring and the second being baseflow where streamflow originates from groundwater discharge in mid-to-late summer. Results indicate that total nitrogen loading during snowmelt periods is four times greater than annual average deliveries. The contribution of agricultural sources (chemical agricultural fertilizer, manure, and N fixation) is substantially higher (15% for TN and 5% for TP) during melt than baseflow, while point sources, septic tanks, and atmospheric deposition become more prominent contributors to nutrient delivery during baseflow. Thus, seasonal variation of nutrient transport should be considered when establishing nutrient criteria and reduction targets. In Chapter 4, the focus is shifted to tile drainage systems by developing a random forest classifier to map tile drainage in the US Midwest in 2017, as spatially explicit tile drainage information are still lacking. Thirty-one satellite-derived and environmental variables sampled at 60,938 tile and non-tile ground truth points collected from a variety of sources were used to train and validate the classifier. The classifier achieved a good performance (Overall accuracy: 96%; F1 score: 0.9). Then, the classifier was used to do classifications for other regions that lack ground truth data within the study region. The classified tile drainage area correlated reasonably well with the county-level area reported by the USDA National Agricultural Statistics Service (r2 = 0.68). The overall importance revealed that the maximum nighttime land surface temperature in the summer ranked highest, followed by climate- and soil-related variables. This dissertation has produced many spatially explicit data products, including nitrogen and phosphorus loadings, sources, pathways, and hotspots. These are available not only on an annual basis but also for two hydrological seasons. Besides, a significant nutrient pathway, agricultural tile drainage, has been mapped and extended to the study region from the USGLB to the 14 states across the US Midwest. These products and insights could help watershed managers and decision-makers implement nutrient reductions at the right time and place more effectively.
Read
