On the natural spatio-temporal heterogeneity of South Pacific nitrous oxide

Andrew R Babbin,Ray F Weiss,Elisabeth L Boles,Jens Mühle

doi:10.1038/s41467-020-17509-6

Abstract

Nitrous oxide (N2O) is a powerful greenhouse gas and ozone depleting substance, but its natural sources, especially marine emissions, are poorly constrained. Localized high concentrations have been observed in the oxygen minimum zones (OMZs) of the tropical Pacific but the impacts of El Niño cycles on this key source region are unknown. Here we show atmospheric monitoring station measurements in Samoa combined with atmospheric back-trajectories provide novel information on N2O variability across the South Pacific. Remarkable elevations in Samoan concentrations are obtained in air parcels that pass over the OMZ. The data further reveal that average concentrations of these OMZ air parcels are augmented during La Niña and decrease sharply during El Niño. The observed natural spatial heterogeneities and temporal dynamics in marine N2O emissions can confound attempts to develop future projections of this climatically active gas as low oxygen zones are predicted to expand and El Niño cycles change.

Highlights

In order to estimate from where air parcels arriving at the monitoring stations have traveled and help determine the influence of the OMZs, NOAA’s HYSPLIT4 model was used to calculate back-trajectories from the four Pacific stations (Supplementary Fig. 3)
Each trajectory was matched with the corresponding station’s N2O measurement taken nearest to the time at which the back-trajectory arrived at the station. It is unclear how far back in time the trajectories remain valid before inaccuracies lead to severe divergences between the calculated path and the actual, and both errors from the climatological data and from the HYSPLIT model itself must be considered in evaluating the certainty of the trajectories
HYSPLIT trajectories incur a distance error on order of 20% of the track length[53], and both errors from the climatological data and from the HYSPLIT model itself must be considered in evaluating the certainty of the trajectories

Summary

Introduction

De-trending was achieved by finding a two-degree polynomial fit and subtracting this general curve from the measured concentrations (Supplementary Fig. 1). De-seasonalization of the time series was achieved by averaging all de-trended data by month, interpolating these points to create a continuous sinusoidal seasonal cycle, and subtracting this oscillation from the de-trended series (Supplementary Fig. 2).

Results

Conclusion