Figure 2 - uploaded by Aaron Strong
Content may be subject to copyright.
Daily discharge time series for the Raccoon River at van Meter (USGS 05484500) from 1927 to 2012.

Daily discharge time series for the Raccoon River at van Meter (USGS 05484500) from 1927 to 2012.


... These data were available at the county level from the USDA's National Agricultural Statistics Services database ( Similar to Villarini and Strong (2014), we quantified the basin-wide corn and soybean planted area time series by assuming that farmed area is uniformly distributed across each county and then using the proportion of the counties contained within the watershed to compute this time series. Soybean and corn yields were highly correlated, with soybean being the stronger predictor, and thus we did not use corn yield in our model. ...
... Since the previous year's data (yield and area) were stronger predictors of FWA-N than the analogous data from the concurrent year, we focused on the previous year's crop data as predictors for the following year's FWA-N. All predictors were normalized with respect to their mean and standard deviation, computed over the period of 1974 to 2013 to account for the very large differences in their values in accordance with Villarini and Strong (2014), while ensuring predictors were not highly correlated. ...
Improved understanding of the drivers of stream nitrate is necessary to improve water quality. This is particularly true for Iowa, a large contributor to Mississippi River Basin nitrate loads. Here we focus on the Raccoon River at Des Moines, Iowa, and develop statistical models to describe the monthly (from March to August) nitrate concentrations in terms of eight drivers representing monthly climate, monthly hydrology, and yearly cropping practices. We consider six 2-parameter distributions, linear and non-linear dependencies between the predictors, and the distributions’ parameters. Model selection was performed by penalizing more complex models. Our results show that the Weibull and Gumbel distributions are the only two selected distributions. Baseflow and the previous year’s soybean area were the two predictors most often identified as important. Our modeling results imply that increases in soybean area have led to increasing nitrate concentrations. Moreover, nitrate concentrations are related to baseflow in a non-linear way, with effects strongest when baseflow is near or below the average condition. Additional relevant predictors were precipitation and, to a lesser extent, temperature. We conclude that best management practices and improved conservation targeting soybean in a corn-soybean rotation will improve water quality in this artificially-drained system.