Biases in discrete CH4 and N2O sampling protocols associated with temporal variation of gas fluxes from manure storage systems

Manure slurry storage systems are possible sources of methane (CH4) and nitrous oxide (N2O), both of which are strong greenhouse gases (GHG). The most commonly used methods to measure these emissions rely on chamber techniques deployed at discrete intervals (i.e. non-continuously). Due to long-term and diurnal variations in GHG emissions, discrete sampling may yield biased estimates when integrating fluxes over time. This research quantified the effect of sampling interval and the ‘time-of-day’ of sample collection on total emission estimates from discrete sampling compared to continuous monitoring, and characterized temporal flux variations in relation to environmental conditions. Methane and N2O emissions were measured continuously over the 6 mo warm storage season in 2010 from 6 pilot-scale dairy manure slurry storage tanks using flow-through steady state chambers. Discrete sampling was simulated by extracting data from hourly flux time series at 13 sampling intervals ranging from 1 to 21 d. For each sampling interval, 24 datasets were generated, one for each hour of the day. When there were high rates of CH4 ebullition, and crusts did not completely cover the surface, diurnal flux variations were linked with the diurnal surface temperature (T0) cycle, whereby bubbles that accumulated at the surface overnight burst after sunrise. With complete manure surface crusting, episodic and unpredictable flux events were more common. Diurnal variation in N2O emissions was strongly linked with variation in T0 when crusts >2 mo old were present. For CH4, sampling between 1800 and 0800 h at intervals ≤7 d yielded ±10% deviation between discrete and continuous monitoring with a frequency of 96%. The frequency of achieving ±10% deviation for N2O was 50% when sampling at ∼2000 h. If sampling during these times is not possible, manual measurements should be made in early morning (before 0900 h) and late afternoon (after 1700 h).

► We characterized temporal CH4 and N2O flux variations for stored dairy slurries.
► We examined how discrete sampling was affected by these flux variations.
► Discrete sampling should be done before 0900 h or after 1700 h, at intervals <=7 d.
► Gas flux variations were linked with the daily slurry surface temperature (T0) cycle.
► The strength of the relation between fluxes and T0 depended on crust conditions.