High‐Frequency Sensor Data Reveal Across‐Scale Nitrate Dynamics in Response to Hydrology and Biogeochemistry in Intensively Managed Agricultural Basins

Abstract

Excess nitrate in rivers draining intensively managed agricultural watersheds has caused coastal hypoxic zones, harmful algal blooms, and degraded drinking water. Hydrology and biogeochemical transformations influence nitrate concentrations by changing nitrate supply, removal, and transport. For the Midwest Unites States, where much of the land is used for corn and soybean production, a better understanding of the response of nitrate to hydrology and biogeochemistry is vital in the face of high nitrate concentrations coupled with projected increases of precipitation frequency and magnitude. In this study, we capitalized on the availability of spatially and temporally extensive sensor data in the region to evaluate how nitrate concentration (NO 3 À ) interacts with discharge (Q) and water temperature (T) within eight watersheds in Iowa, United States, by evaluating land use characteristics and multiscale temporal behavior from 5-year, high-frequency, time series records. We show that power spectral density of Q, NO 3 À , and T, all exhibit power law behavior with slopes greater than 2, implying temporal self-similarity for a range of scales. NO 3 À was strongly cross correlated with Q for all sites and correlation increased significantly with drainage area across sites. Peak NO 3 À increased significantly with crop coverage across watersheds.Temporal offsets in peak NO 3 À and peak Q, seen at all study sites, reduced the impact of extreme events. This study illustrates a relatively new approach to evaluating environmental sensor data and revealed characteristics of watersheds in which extreme discharge events have the greatest consequences.Plain Language Summary Nitrate export from the Midwest United States has caused water quality degradation extending from the Midwest, where concentrations exceed drinking water limits, to the northern Gulf of Mexico, where it contributes to the formation of the Dead Zone. Climate models predict that precipitation in the Midwest will increase; therefore, understanding the response of nitrate to streamflow (discharge) is vital. In this study, we analyzed the patterns within nitrate, discharge and temperature sensor data over time for eight agricultural watersheds to determine how nitrate concentration is related to discharge or water temperature. From this, we evaluated the relative importance of hydrology vs. nitrate removal or release in controlling nitrate export. We show that nitrate, discharge, and temperature all exhibit self-similar characteristics across a range of temporal scales and that the percent of land used for agriculture was predictive of peak nitrate concentration. We found that nitrate was strongly related to discharge and that the similarity between nitrate and discharge increased with watershed size. However, peak nitrate concentration generally lagged behind peak discharge resulting in maximum peak nitrate loads often occurring at intermediate-sized, not extreme, events.Key Points:• Discharge, nitrate concentration, and temperature exhibited power law behavio...

Temporal Variability in Nitrate‐Discharge Relationships in Large Rivers as Revealed by High‐Frequency Data

“…Further, the practice of defining c‐Q slopes in log–log space to describe hydrologic controls on solute export is often dependent on the data fitting a power law function, but additional analysis is needed to ensure this is accurate for high‐frequency water quality and quantity data sets. Therefore, a comparison of c‐Q slopes with alternative metrics may help to provide additional insight into c‐Q relationships and system behavior (Hansen & Singh, ; Musolff et al, ; Thompson et al, ). Here we discuss our findings in the context of such metrics and provide suggestions for necessary future research directions.…”

“…Temperature was less correlated with c‐Q dynamics in the Mississippi River (Table ). Hansen and Singh () did not find a correlation between NO 3 − concentrations and temperature in a suite of Iowa Rivers within the Mississippi River Basin and suggested that in‐stream NO 3 − processing was not as strong as terrestrial processing and inputs. More research is needed to fully understand how river temperature can be used as a predictor for seasonal c‐Q dynamics.…”

Monitoring the riverine pulse: Applying high‐frequency nitrate data to advance integrative understanding of biogeochemical and hydrological processes

“…Another set of approaches broadly described as data mining includes regression tree analysis to examine climatic variables associated with variation in stream NO 3 − concentrations (Rusjan & Mikoš, ) and neural network modeling (Shrestha et al, ). A promising approach is to explore high‐frequency NO 3 − data in the spectral domain by first removing temporal variation, which allows exploration of the relative roles of terrestrial and in‐stream processes across spatial and discharge scales, including the response to extreme hydrologic events (Hansen & Singh, ). For example, Aubert et al () followed an earlier approach of Kirchner, Feng, and Neal () and demonstrated 1/ f fractal scaling of NO 3 − and other constituents in an agricultural catchment consistent with an earlier hypothesis that downslope advective‐dispersive transport dominates across a range of solutes (Kirchner, Feng, & Neal, ).…”

“…Recent investigations suggest that the HI is being rapidly adopted by those who study nutrient runoff patterns during high flow (Baker & Showers, 2019;Blaen et al, 2017;Dupas et al, 2016;Fovet et al, 2018;Vaughan et al, 2017). Limitations of the HI have been discussed and a pattern recognition method using machine learning was recently pioneered to delineate suspended sediment hysteresis typology during events (Hamshaw, Dewoolkar, Schroth, Wemple, & Rizzo, 2018). Further advances in NO 3 − hysteresis investigations may necessitate detailed modeling that includes empirical representation of controls on net supply applied to formally test process-based hypotheses as developed in turbidity studies (Mather & Johnson, 2014).…”

Wetland Dynamics Inferred from Spectral Analyses of Hydro-Meteorological Signals and Landsat Derived Vegetation Indices

“…A robust estimation of the scaling exponent β can be achieved by computing the slope of the linear regression fitted to the estimated PSD plotted on log-log scales [40]. The strength of these scaling exponents provides useful information about the inherent memory of the system [26,41,42]. Witt and Malamud [26] found PSD analysis to be a more accurate method to quantify persistence of self-affine time series than other empirical methods such as Hurst rescaled range (R/S) analysis, detrended fluctuation analysis, and semi-variogram analysis.…”