Predicting Nitrogen Loading in Streams Under Climate Change Scenarios in the Continental United States.

Abstract:

Human actions have doubled the amount of nitrogen in the terrestrial biosphere. Although nitrogen application in the form of fertilizers increases food production, excess nitrogen can be harmful to the environment and to human well-being. Excess nitrogen in streams is transported into downstream water bodies, which leads to increased eutrophication and associated problems such as harmful algal blooms and/or hypoxic conditions.

The amount of nitrogen exported to streams depends on several factors, including nitrogen input to the watershed, land use, and precipitation. Previous studies have developed models for predicting nitrate load using stream discharge, in order to estimate the contribution of various factors to total nitrogen load, to identify strategies for reducing nitrogen load, and to assess future changes in nitrogen load resulting from anticipated changes in precipitation patterns. Applying these models to estimate future nitrogen loads thus requires running a rainfall-runoff model driven by climate model predictions before nitrogen loading can, in turn, be estimated, thereby compounding uncertainties.

In this study, we present a statistical modeling approach that circumvents this two-step process by estimating nitrogen loading directly from precipitation predictions that can be obtained from climate model outputs. The proposed model uses net anthropogenic nitrogen input (NANI), land use type, and precipitation as input parameters.

Preliminary results show that the model explains greater than 65% of the variance in the observed annual log transformed nitrogen loads across various catchments throughout the United States. The model is applied to the watersheds comprising the Mississippi river basin to identify the spatial distribution of the sources of nitrogen loading and the inter-annual variation in nitrogen loads under current conditions. Additionally, the model is used to examine changes in magnitude and spatial patterns of nitrogen loading for precipitation patterns under future climate conditions as represented by the CMIP5 ensemble simulations.