A widely-used eddy covariance gap-filling method creates systematic bias in carbon balance estimates.

Climate change mitigation requires, besides reductions in greenhouse gas emissions, actions to increase carbon sinks in terrestrial ecosystems. A key measurement method for quantifying such sinks and calibrating models is the eddy covariance technique, but it requires imputation, or gap-filling, of missing data for determination of annual carbon balances of ecosystems. Previous comparisons of gap-filling methods have concluded that commonly used methods, such as marginal distribution sampling (MDS), do not have a significant impact on the carbon balance estimate. By analyzing an extensive, global data set, we show that MDS causes significant carbon balance errors for northern (latitude [Formula: see text]) sites. MDS systematically overestimates the carbon dioxide (CO[Formula: see text]) emissions of carbon sources and underestimates the CO[Formula: see text] sequestration of carbon sinks. We also reveal reasons for these biases and show how a machine learning method called extreme gradient boosting or a modified implementation of MDS can be used to substantially reduce the northern site bias.

Carbon and climate implications of rewetting a raised bog in Ireland.

Peatland rewetting has been proposed as a vital climate change mitigation tool to reduce greenhouse gas emissions and to generate suitable conditions for the return of carbon (C) sequestration. In this study, we present annual C balances for a 5-year period at a rewetted peatland in Ireland (rewetted at the start of the study) and compare the results with an adjacent drained area (represents business-as-usual). Hydrological modelling of the 230-hectare site was carried out to determine the likely ecotopes (vegetation communities) that will develop post-rewetting and was used to inform a radiative forcing modelling exercise to determine the climate impacts of rewetting this peatland under five high-priority scenarios (SSP1-1.9, SS1-2.6, SSP2-4.5, SSP3-7.0 and SSP5-8.5). The drained area (marginal ecotope) was a net C source throughout the study and emitted 157 ± 25.5 g C m-2  year-1 . In contrast, the rewetted area (sub-central ecotope) was a net C sink of 78.0 ± 37.6 g C m-2  year-1 , despite relatively large annual methane emissions post-rewetting (average 19.3 ± 5.2 g C m-2  year-1 ). Hydrological modelling predicted the development of three key ecotopes at the site, with the sub-central ecotope predicted to cover 24% of the site, the sub-marginal predicted to cover 59% and the marginal predicted to cover 16%. Using these areal estimates, our radiative forcing modelling projects that under the SSP1-1.9 scenario, the site will have a warming effect on the climate until 2085 but will then have a strong cooling impact. In contrast, our modelling exercise shows that the site will never have a cooling impact under the SSP5-8.5 scenario. Our results confirm the importance of rapid rewetting of drained peatland sites to (a) achieve strong C emissions reductions, (b) establish optimal conditions for C sequestration and (c) set the site on a climate cooling trajectory.

Identifying main uncertainties in estimating past and present radiative forcing of peatlands.

Reconstructions of past climate impact, that is, radiative forcing (RF), of peatland carbon (C) dynamics show that immediately after peatland initiation the climate warming effect of CH4 emissions exceeds the cooling effect of CO2 uptake, but thereafter the net effect of most peatlands will move toward cooling, when RF switches from positive to negative. Reconstructing peatland C dynamics necessarily involves uncertainties related to basic assumptions on past CO2  flux, CH4 emission and peatland expansion. We investigated the effect of these uncertainties on the RF of three peatlands, using either apparent C accumulation rates, net C balance (NCB) or NCB plus C loss during fires as basis for CO2 uptake estimate; applying a plausible range for CH4 emission; and assuming linearly interpolated expansion between basal dates or comparatively early or late expansion. When we factored that some C would only be stored temporarily (NCB and NCB+fire), the estimated past cooling effect of CO2 uptake increased, but the present-day RF was affected little. Altering the assumptions behind the reconstructed CO2  flux or expansion patterns caused the RF to peak earlier and advanced the switch from positive to negative RF by several thousand years. Compared with NCB, including fires had only small additional effect on RF lasting less than 1000 year. The largest uncertainty in reconstructing peatland RF was associated with CH4 emissions. As shown by the consistently positive RF modelled for one site, and in some cases for the other two, peatlands with high CH4 emissions and low C accumulation rates may have remained climate warming agents since their initiation. Although uncertainties in present-day RF were mainly due to the assumed CH4 emission rates, the uncertainty in lateral expansion still had a significant effect on the present-day RF, highlighting the importance to consider uncertainties in the past peatland C balance in RF reconstructions.

Insect herbivory dampens Subarctic birch forest C sink response to warming.

Climate warming is anticipated to make high latitude ecosystems stronger C sinks through increasing plant production. This effect might, however, be dampened by insect herbivores whose damage to plants at their background, non-outbreak densities may more than double under climate warming. Here, using an open-air warming experiment among Subarctic birch forest field layer vegetation, supplemented with birch plantlets, we show that a 2.3 °C air and 1.2 °C soil temperature increase can advance the growing season by 1-4 days, enhance soil N availability, leaf chlorophyll concentrations and plant growth up to 400%, 160% and 50% respectively, and lead up to 122% greater ecosystem CO2 uptake potential. However, comparable positive effects are also found when insect herbivory is reduced, and the effect of warming on C sink potential is intensified under reduced herbivory. Our results confirm the expected warming-induced increase in high latitude plant growth and CO2 uptake, but also reveal that herbivorous insects may significantly dampen the strengthening of the CO2 sink under climate warming.

Early snowmelt significantly enhances boreal springtime carbon uptake.

We determine the annual timing of spring recovery from space-borne microwave radiometer observations across northern hemisphere boreal evergreen forests for 1979-2014. We find a trend of advanced spring recovery of carbon uptake for this period, with a total average shift of 8.1 d (2.3 d/decade). We use this trend to estimate the corresponding changes in gross primary production (GPP) by applying in situ carbon flux observations. Micrometeorological CO2 measurements at four sites in northern Europe and North America indicate that such an advance in spring recovery would have increased the January-June GPP sum by 29 g⋅C⋅m-2 [8.4 g⋅C⋅m-2 (3.7%)/decade]. We find this sensitivity of the measured springtime GPP to the spring recovery to be in accordance with the corresponding sensitivity derived from simulations with a land ecosystem model coupled to a global circulation model. The model-predicted increase in springtime cumulative GPP was 0.035 Pg/decade [15.5 g⋅C⋅m-2 (6.8%)/decade] for Eurasian forests and 0.017 Pg/decade for forests in North America [9.8 g⋅C⋅m-2 (4.4%)/decade]. This change in the springtime sum of GPP related to the timing of spring snowmelt is quantified here for boreal evergreen forests.

The uncertain climate footprint of wetlands under human pressure.

Significant climate risks are associated with a positive carbon-temperature feedback in northern latitude carbon-rich ecosystems, making an accurate analysis of human impacts on the net greenhouse gas balance of wetlands a priority. Here, we provide a coherent assessment of the climate footprint of a network of wetland sites based on simultaneous and quasi-continuous ecosystem observations of CO2 and CH4 fluxes. Experimental areas are located both in natural and in managed wetlands and cover a wide range of climatic regions, ecosystem types, and management practices. Based on direct observations we predict that sustained CH4 emissions in natural ecosystems are in the long term (i.e., several centuries) typically offset by CO2 uptake, although with large spatiotemporal variability. Using a space-for-time analogy across ecological and climatic gradients, we represent the chronosequence from natural to managed conditions to quantify the "cost" of CH4 emissions for the benefit of net carbon sequestration. With a sustained pulse-response radiative forcing model, we found a significant increase in atmospheric forcing due to land management, in particular for wetland converted to cropland. Our results quantify the role of human activities on the climate footprint of northern wetlands and call for development of active mitigation strategies for managed wetlands and new guidelines of the Intergovernmental Panel on Climate Change (IPCC) accounting for both sustained CH4 emissions and cumulative CO2 exchange.

Micrometeorological measurements of methane and carbon dioxide fluxes at a municipal landfill.

Continuous and area-integrating monitoring of methane (CH4) and carbon dioxide (CO2) emissions was performed for 6 and 9 months, respectively, at a municipal landfill in Finland with the micrometeorological eddy covariance (EC) method. The mean CH4 emission from June to December was 0.53 mg m(-2) s(-1), while the CO2 emission between February and December averaged 1.78 mg m(-2) s(-1). The CH4 emissions from the summit area of the landfill, where active waste deposition was going on, were 1.7 times as high as from the slope area with a better surface cover. The variation in emissions over the source area of the measurement was high. Significant seasonal variation, linked to air and soil temperature, was only seen in the CO2 release rates. Results obtained with the EC method were comparable to those measured with closed static chambers. According to the EC measurements, the gas recovery system decreased CH4 fluxes by 69-79%. The ratio of the measured CH4 and CO2 emissions roughly indicated the route of the landfill gas emission, resembling the ratio of the gases measured in the gas wells (1.24) when the emission originated from the area with no oxidizing cover layer and being smaller when CH4 oxidation had taken place.

Nitrous oxide emissions from a municipal landfill.

The first measurements of nitrous oxide (N20) emissions from a landfill by the eddy covariance method are reported. These measurements were compared to enclosure emission measurements conducted at the same site. The average emissions from the municipal landfill of the Helsinki Metropolitan Area were 2.7 mg N m(-2) h(-1) and 6.0 mg N m(-2) h(-1) measured bythe eddy covariance and the enclosure methods, respectively. The N20 emissions from the landfill are about 1 order of magnitude higher than the highest emissions reported from Northern European agricultural soils, and 2 orders of magnitude higher than the highest emissions reported from boreal forest soils. Due to the small area of landfills as compared to other land-use classes, the total N20 emissions from landfills are estimated to be of minor importance for the total emissions from Finland. Expressed as a greenhouse warming potential (GWP100), the N2O emissions make up about 3% of the total GWP100 emission of the landfill. The emissions measured by the two systems were generally of similar magnitude, with enclosure measurements showing a high small-scale spatial variation.

Monitoring and modelling of biosphere/atmosphere exchange of gases and aerosols in Europe.

Monitoring and modelling of deposition of air pollutants is essential to develop and evaluate policies to abate the effects related to air pollution and to determine the losses of pollutants from the atmosphere. Techniques for monitoring wet deposition fluxes are widely applied. A recent intercomparison experiment, however, showed that the uncertainty in wet deposition is relatively high, up to 40%, apart from the fact that most samplers are biased because of a dry deposition contribution. Wet deposition amounts to about 80% of the total deposition in Europe with a range of 10-90% and uncertainty should therefore be decreased. During recent years the monitoring of dry deposition has become possible. Three sites have been operational for 5 years. The data are useful for model development, but also for model evaluation and monitoring of progress in policy. Data show a decline in SO(2) dry deposition, whereas nitrogen deposition remained constant. Furthermore, surface affinities for pollutants changed leading to changes in deposition. Deposition models have been further developed and tested with dry deposition measurements and total deposition measurements on forests as derived from throughfall data. The comparison is reasonable given the measurement uncertainties. Progress in ozone surface exchange modelling and monitoring shows that stomatal uptake can be quantified with reasonable accuracy, but external surface uptake yields highest uncertainty.

Assessing vegetation exposure to ozone: is it possible to estimate AOT40 by passive sampling?

A statistical model for estimating AOT40 from time-integrated concentration data is presented and validated. AOT40 is a numerical index that describes the ozone exposure of ecosystems in terms of the hourly accumulated exposure over a threshold of 40 ppb. A rather simple model formulation was achieved by approximating the frequency distribution of hourly concentrations by the Gaussian probability distribution. The model represents a relationship between a time-averaged ozone concentration and the corresponding AOT40. Time-averaged concentration data are obtained when employing the passive sampling technique. For testing the method, passive sampling data with a 14-day sampling time were simulated by averaging continuous monitoring data. Data from eight background stations in Europe were used for calibration and testing. In spite of the simplifying assumption, the model performed well for the accumulated exposure when tested against independent data. The results show that it is possible to obtain a reasonable estimate of AOT40 even in the absence of continuous data, if the time-averaged concentration can be measured reliably.