A geostatistical approach to identify and mitigate agricultural nitrous oxide emission hotspots

Highlights

• Geospatial analyses resolved N₂O emissions at fine spatial scales.

• Hotspots emitted N₂O at rates > 2-fold greater than non-hotspot locations.

• Targeted management of N₂O hotspots could reduce emissions by 17%.

Abstract

Anthropogenic emissions of nitrous oxide (N₂O), a trace gas with severe environmental costs, are greatest from agricultural soils amended with nitrogen (N) fertilizer. However, accurate N₂O emission estimates at fine spatial scales are made difficult by their high variability, which represents a critical challenge for the management of N₂O emissions. Here, static chamber measurements (n = 60) and soil samples (n = 129) were collected at approximately weekly intervals (n = 6) for 42-d immediately following the application of N in a southern Minnesota cornfield (15.6-ha), typical of the systems prevalent throughout the U.S. Corn Belt. These data were integrated into a geostatistical model that resolved N₂O emissions at a high spatial resolution (1-m). Field-scale N₂O emissions exhibited a high degree of spatial variability, and were partitioned into three classes of emission strength: hotspots, intermediate, and coldspots. Rates of emission from hotspots were 2-fold greater than non-hotspot locations. Consequently, 36% of the field-scale emissions could be attributed to hotspots, despite representing only 21% of the total field area. Variations in elevation caused hotspots to develop in predictable locations, which were prone to nutrient and moisture accumulation caused by terrain focusing. Because these features are relatively static, our data and analyses indicate that targeted management of hotspots could efficiently reduce field-scale emissions by as much 17%, a significant benefit considering the deleterious effects of atmospheric N₂O.

Introduction

Nitrous oxide (N₂O) is a potent greenhouse gas (Hartmann et al., 2013) and the leading cause of stratospheric ozone loss (Ravishankara et al., 2009). In response to its deleterious environmental effects, efforts to mitigate agricultural emissions, which account for nearly 75% of the national anthropogenic source (US Department of State, 2014), are in development. Such efforts often focus on N management improvements (e.g., optimizing the source, depth, and timing of fertilizer) at the field or farm scale. Yet, the findings from these mitigation strategies have been highly variable (Venterea et al., 2016), in part because episodic and spatially variable emissions hinder accurate budget estimates (Mathieu et al., 2006, Velthof et al., 2000). For instance, field-scale N₂O emission measurements with chambers can yield a coefficient of variation (CV) as high as 500% (Folorunso and Rolston, 1984, van den Pol-van Dasselaar et al., 1998), suggesting that our ability to accurately determine the outcome of mitigation practices is cause for concern. At fine sub-field spatial scales (< 1 m² to 1000 m²), N₂O “hotspots” appear to be disproportionately strong sources (Parkin, 1987, van den Heuvel et al., 2009), yet their influence over cumulative field-scale emissions remains uncertain because high-resolution data are rarely available. For farmers to manage N₂O emissions effectively, subfield-scale emission estimates are necessary to identify potential hotspots and to benchmark their effects on field-scale mitigation practices.

Light detection and ranging (LiDAR) digital elevation models (DEMs) are powerful tools that can help guide precision agriculture and conservation strategies (Galzki et al., 2011, Wan et al., 2014). When coupled with geospatial techniques, this emerging technology helps generate high-resolution maps of agriculturally relevant information such as the presence of hydric soils (Fink and Drohan, 2016), moisture content (Moore et al., 1993, Murphy et al., 2009), and soil nitrogen status (Weintraub et al., 2014) that allow farmers to focus extra attention and resources on critical areas. Furthermore, complex processes like methane emissions (Sundqvist et al., 2015) have been characterized using DEMs, suggesting that this technology can better resolve the field-scale spatial distribution of N₂O emissions.

Indeed, differences in topography and landscape position have a strong influence on N₂O emissions (Ambus, 1998, Ball et al., 1997) because terrain gradients redistribute moisture and nutrients that are necessary for the production of N₂O. Consequently, N₂O emission frequency distributions are typically positively skewed by a few strong sources (Parkin, 1987, Velthof et al., 2000) observed at topographically low positions (Ambus, 1998). Here, terrain focusing enables the development of hotspots by concentrating organic matter, moisture, and nitrate (NO₃⁻) into localized, but potentially predictable areas. Taken together, these soil characteristics can support disproportionately high rates of denitrification (Groffman et al., 2009) that we posit are capable of sustaining high N₂O emissions. However, field-scale emission distribution maps remain coarse, since an unrealistic number of static chambers are required to resolve the high variability, implying poor constraints on hotspots.

With the aid of DEMs and geospatial analyses, denitrification hotspots can be isolated and mapped by pinpointing locations with the highest probability of moisture and NO₃⁻ accumulation (Anderson et al., 2015). We propose that a similar approach can resolve the distribution of N₂O emissions at a high spatial resolution that will guide targeted mitigation practices. Here, we examine the spatial distribution of N₂O fluxes and cumulative emissions in a strip-tilled cornfield to address three questions: 1) can DEMs help predict where N₂O hotspots will develop on the landscape; 2) how significant are hotspots in the cumulative field-scale budget; and 3) how can DEMs be used to guide N management and N₂O mitigation?