Uncertainty in United States coastal wetland greenhouse gas inventorying

Coastal wetlands store carbon dioxide (CO2) and emit CO2 and methane (CH4) making them an important part of greenhouse gas (GHG) inventorying. In the contiguous United States (CONUS), a coastal wetland inventory was recently calculated by combining maps of wetland type and change with soil, biomass, and CH4 flux data from a literature review. We assess uncertainty in this developing carbon monitoring system to quantify confidence in the inventory process itself and to prioritize future research. We provide a value-added analysis by defining types and scales of uncertainty for assumptions, burial and emissions datasets, and wetland maps, simulating 10 000 iterations of a simplified version of the inventory, and performing a sensitivity analysis. Coastal wetlands were likely a source of net-CO2-equivalent (CO2e) emissions from 2006–2011. Although stable estuarine wetlands were likely a CO2e sink, this effect was counteracted by catastrophic soil losses in the Gulf Coast, and CH4 emissions from tidal freshwater wetlands. The direction and magnitude of total CONUS CO2e flux were most sensitive to uncertainty in emissions and burial data, and assumptions about how to calculate the inventory. Critical data uncertainties included CH4 emissions for stable freshwater wetlands and carbon burial rates for all coastal wetlands. Critical assumptions included the average depth of soil affected by erosion events, the method used to convert CH4 fluxes to CO2e, and the fraction of carbon lost to the atmosphere following an erosion event. The inventory was relatively insensitive to mapping uncertainties. Future versions could be improved by collecting additional data, especially the depth affected by loss events, and by better mapping salinity and inundation gradients relevant to key GHG fluxes. Social Media Abstract: US coastal wetlands were a recent and uncertain source of greenhouse gasses because of CH4 and erosion.


Introduction
Managing land to optimize carbon storage and mitigate degradation is one among many strategies under consideration to curb anthropogenic greenhouse gas emissions (Griscom et al 2017). Coastal wetlands-defined here as salt marshes, mangroves, tidal freshwater wetlands, and tidal freshwater forests -have received some of this attention because they can act as a net-greenhouse gas sink (Howard et al 2017), and because restoration (Kroeger et al 2017) and conservation (DeLaune and White 2012) may reduce or mitigate emissions. Regulation and market mechanisms can incentivize wetland restoration to promote emission reduction (Pendleton et al 2012, Wylie et al 2016 and myriad co-benefits (Barbier et al 2011, Doughty et al 2017, Griscom et al 2017. Coastal wetlands can bury carbon (Chmura et al 2003, Ouyang and Lee 2013, Howard et al 2017 and form new soil (Morris et al 2002) by adding organic carbon to the soil column through sub-surface root addition (Nyman et al 2006). Carbon burial is a dynamic response to sea-level rise (Kirwan andMegonigal 2013, Kirwan et al 2016). Carbon removed from the atmosphere and incorporated into soils and plant matter is referred to throughout this paper as a 'removal'. However, wetlands can also be the sources of emissions when they are eroded (DeLaune and White 2012), developed (Stein et al 2014), or drained for agriculture (Drexler et al 2009). Freshwater and brackish tidal wetlands emit methane (CH 4 ) (Bridgham et al 2006, Poffenbarger et al 2011, a more potent greenhouse gas than carbon dioxide (CO 2 ) over the course of its atmospheric lifetime (Frolking andRoulet 2006, Neubauer andMegonigal 2015). At a national scale, in order to estimate total greenhouse gas emissions or removals, researchers need to know the areal coverage of different wetland types, the areal coverage of wetland change events, and to assign annualized CO 2 equivalent (CO 2 e) stock changes to those wetland classes and change events.
Spatial data, literature review, and expert assumptions are all used to inventory greenhouse gas fluxes at national scales. These inputs introduce uncertainty (IPCC 2014), which needs to be quantified to establish both levels of confidence and priorities for future research. The Intergovernmental Panel on Climate Change (IPCC) quantifies emissions and removals with 'emissions factors' and 'activities data'. For agricultural, forested and other lands, emissions factors are values assigning greenhouse gas fluxes to land cover types and change events (equation (1)). Activities data are typically interpreted as the areal coverage of land cover type and/or land cover change events.
The IPCC published guidance for national-scale greenhouse gas inventories for coastal wetlands (IPCC 2014), and the United States incorporated these for the first time in its 2017 national greenhouse gas inventory (NGGI) conducted by the Environmental Protection Agency (EPA) (EPA 2017). Our analysis is not an official part of that NGGI. Instead, we used the accounting concepts outlined therein, as well as updated literature review and spatial data, in order to improve uncertainty estimates at the national scale and highlight areas of research that could further reduce that uncertainty.
Emissions or Removals flux Activities area Emissions.Factor flux area . 1 In the NGGI, uncertainties in emissions and removals were estimated using a basic algebraic approach (IPCC 2014, EPA 2017. We address five assumptions and approaches from the previous NGGI to improve uncertainty estimates in coastal wetlands: 1. The probability distributions of the activities and emissions data were not explicitly defined; 2. Key variables such as the uncertainty inherent in tidal-elevation maps were not included; 3. Uncertainties in many activities data and emissions factors are best described by non-normal distributions, which could not be accommodated using the basic algebraic approach; 4. Key assumptions, such as the depth affected by degradation events, were based on expert assessment and therefore treated as fixed values, not as probability distributions; and 5. Some inventory decisions, such as how to calculate the global warming potential (GWP) of CH 4 emissions and how much area to include in the inventory, have more than one recognized technique, and uncertainty from choosing among techniques was not quantified.
Our analysis expands upon the scope of the NGGI uncertainty analysis and explicitly identifies and quantifies uncertainty for key activities data and emissions factors. We update key datasets with new synthesis efforts (Windham-Myers and Cai in Revision) (supplemental information is available online at stacks.iop. org/ERL/13/115005/mmedia) and the results of NASA Carbon Monitoring Systems projects (Olofsson et al 2014, Byrd et al 2018. Our research questions are: 1. How much certainty is there that CONUS coastal wetlands were a net-source or sink of GHGs from 2006-2011? 2. Which datasets, assumptions, or mapping categories introduce the most uncertainty into the coastal wetland category of the US national GHG inventory?

Methods
We addressed our research questions by integrating multiple spatial and non-spatial datasets, explicitly defining uncertainty in each step, estimating total propagated uncertainty using a Monte Carlo analysis (Ogle et al 2003, IPCC 2006, and by quantifying the sensitivity of total emissions and removals to each input.

Time period and 2006-2011 land cover classes analyzed
As in the NGGI (EPA 2017) we quantified area using the Coastal Change Analysis Program (C-CAP; figure 1; supplemental table 1). C-CAP is a Landsatbased land cover mapping product with 23 land cover classes, including six types of intertidal wetlands defined by two types of salinity (palustrine and estuarine) and three types of vegetation (emergent, scrub/shrub, and forested) (McCombs et al2016). We did not include seagrasses in this analysis because C-CAP's 'estuarine aquatic bed' category typically represents nearshore vegetated environments, such as kelp beds, which are not a net-carbon storing system (Howard et al 2017). The coastal wetland section of the NGGI inventory also did not include palustrine forested wetlands, since they fall under the purview of forested lands. We include them because information on their contribution to uncertainty is informative regardless of their reporting subcategory.
The NGGI inventory is required to report from 1990-2015, so they linearly interpolate C-CAP changes back to 1990 and forward to 2015 (EPA 2017). Although C-CAP produces land cover change maps for five-year intervals for all US coastal states from 1996-2011, for our analysis we focus on the C-CAP 2006-2011 time step because it is currently the only version with accuracy assessment data. From 2006-2011 we mapped 240 different land cover types including, six classes of wetlands that had the same classification in 2006 and 2011, and 234 types of change to, from, and between wetland classes.

Overview of inventory calculations
We quantified total US GHG emissions and removals from coastal wetlands by mapping the area of different classes of stable wetlands and different types of change events, then multiplying that area by the summed soil, biomass and methane flux from 2006-2011 (equation (2) In which: i is a 2006-2011 land cover class in n land cover classes estimated area i is the total area of land cover class i Each flux is the mass CO 2 e emitted or stored per unit area for land cover class i. As in the US coastal NGGI, we defined the area of interest as the CONUS and included all C-CAP estuarine wetlands (figure 2) and palustrine wetlands occuring at an elevation below the highest tides. This is referred to throughout as the coastal lands definition. Since estuarine wetlands as C-CAP defines them are driven by oceanic tidal influence, we used mapped area as represented in C-CAP as fixed values (figure 2). Since palustrine wetlands can either be tidal or nontidal, we used a probabilistic map of areas falling below mean higher high water spring (MHHWS) tides to map palustrine wetland area falling within the coastal zone. Palustrine wetland mapped areas were not treated as fixed values; we estimated them as a probability distribution using a mean (μ pal,i ) and standard deviation (σ pal,i ) for each class (i), derived from the probabilistic MHHWS map (equation (3)) Our analysis made a distinction between mapped area and estimated area. Estimated area can be greater than or less than mapped area because unequal omission errors (errors of exclusion) and commission error (errors of inclusion) can cause a land cover class to be over-or under-mapped. We scaled mapped area by taking into account potential errors in 2011 classification (Olofsson et al 2014) as well as 2006-2011 change detection (figure 2). In a simplified version of this concept, accuracy assessment matrices containing counts of true classifications and misclassifications, can be simplified down to a single estimated-tomapped area ratio (r) for a classification (i) (equation (4)). This value will be less than one if a land cover class is over-mapped, and greater than 1 if a land cover class is under-mapped We estimated total emissions or removals by multiplying estimated area by the summed per area flux of soil and biomass CO 2 and CH 4 CO 2 e (figure 2). For emissions factors we treated flux data as it was reported (either positive or negative), but transformed them when necessary, so that any emissions were always represented as a negative value and removals were always represented as a positive value. For soils, if the land cover type did not change or changed but did not result in soil loss (supplemental information 2.3.1), then soil carbon flux was estimated as the annual soil carbon burial rate multiplied by the number of years that wetlands were present (equation (5)). If the 2006-2011 class changed and represented a soil loss event, such as conversion to developed, agricultural land, or open water, then emissions were estimated to be the product of mean soil carbon density, depth lost, and fraction of that returns to the atmosphere (equation (6)). We quantified biomass using three vegetation classes: forested, scrub/shrub, and emergent vegetation. We estimated biomass flux if there was a transition between vegetation types or from vegetated to unvegetated surfaces between 2006-2011 (equation (7)). We quantified CH 4 fluxes using two salinity classes, since freshwater wetlands (palustrine) emit more methane than brackish to saline wetlands (estuarine) (Poffenbarger et al 2011). We calculated methane flux for a class by determining CH 4 emissions associated with the salinity type in 2006 and 2011, summing them, and multiplying by 2.5 to normalize the flux over five years (equation (8) 2.3. Estimating area of wetland class and change events 2.3.1. Using tide and elevation data to map coastal palustrine wetlands As in the previous NGGI, we mapped a subset of palustrine wetlands categorized as coastal lands because their tidal elevation was lower than MHHWS. However, uncertainties in digital elevation model (DEM) elevations and in mapping tidal height were not previously included in the NGGI uncertainty analysis (EPA 2017). We enhanced the inventory by creating a probabilistic coastal lands map (supplemental information: section 2.1). For wetland surface elevation data we used DEMs that were created using Light Detection and Ranging (LiDAR) and were aggregated by the National Oceanic and Atmospheric Association (NOAA) for their Sea-Level Rise Viewer (supplemental table 1). DEMs were created to Federal Emergency Management Administration accuracy standards (ASPRS 2004, Coveney 2013). DEMs have a nominal root mean square error (RMSE) of 0.185 m for low-relief areas and assume no bias (supplemental table 1). However, wetland vegetation and soil introduce system-specific bias and random error (Chassereau et al 2011) not captured by the nominal accuracy reporting. We corrected for a mean error of 0.173 m and estimate a RMSE of 0.205 m for wetland surfaces based on a weighted average of results from multiple US-based studies (supplemental table 2). We created a map of MHHWS heights using empirical Bayesian kriging to interpolate between NOAA tide gauges. We also created a corresponding uncertainty map incorporating random error in LiDAR mapping, datum transformations (Schmid et al 2013, Leon et al 2014, and distance between tide gauges. We combined the DEMs, the MHHWS map, and the associated uncertainty surfaces into a single spatial layer representing the probability of elevation being below MHHWS (figures 1, 2).
For palustrine wetlands, we treated mapped area as a random variable. For each of 111 palustrine wetland categories we extracted pixel counts by probability class for the coastal lands map intersecting the C-CAP class and represented mapped area as a normal distribution approximated from the multiple binomial distributions (supplemental information: section 2.1). The means and standard deviations for all 111 palustrine wetland stable classes and palustrine wetland change events are reported in supplemental table 2.

Representing uncertainty in land cover classification and change detection
We calculated an estimated area from mapped area The accuracy assessment matrix records counts for all instances of mapped classes-what a datapoint was mapped as-and reference classes-what it actually was (supplemental tables 4-5). We converted the accuracy assessment matrix from counts to proportional areas (Olofsson et al 2014, Byrd et al 2018, and calculated the estimated proportional area for each class as the reference class' column sum in the proportional area matrix. We used estimated and mapped area at the full map scale to calculate an estimated to mapped area ratio (r). For each 2006-2011 C-CAP class, we used the appropriate r to scale mapped area by the 2011 class. We then used a second r value from the 'change' and 'no change' matrix to scale again based on change detection. Additional detail on how we calculated proportional area accuracy assessment matrices and class-specific scaling factors are available in the supplemental information: section 2.2).
We represented uncertainty in estimated to mapped area ratio by representing each mapped class in the accuracy assessment count matrix as a multiple multinomial distributions, a distribution that describes counts falling into two or more categories as a random variable (supplemental information: section 2.2).

Carbon storage and emissions data
As in the NGGI we calculated emissions factors for soils, and CH 4 based on literature review and synthesis. Unlike the NGGI we include carbon fluxes related to biomass because data is now available as part of a remote sensing calibration and validation effort (Byrd et al 2018), and a literature review that is part of continued inventory development (supplemental information: section 2.3). We did not include N 2 O emissions.

Soil flux data
We estimated soil carbon stock change in wetlands remaining wetlands and lands converted to wetlands as annual carbon burial rate from a literature review of lead-210 ( 210 Pb) dated cores (supplemental information; section 2.3.1). 210 Pb-based measurements typically integrate carbon burial over a century, compared to cesium-137 ( 137 Cs)-and artificial plot-based measurements, which integrate carbon burial over multidecadal to annual time scales; therefore we assumed 210 Pb-based rates are more representative of longterm storage rates. We described soil carbon burial using a lognormal distribution because observed removals can not be negative when strictly relying on dated sediment profiles, observed values were always greater than zero, and the data show a positively skewed distribution (figure 3; table 1).
For soil carbon stock change associated with wetland loss, we used average soil carbon density values reported by Holmquist et al (2018) to characterize the CO 2 emission rate (table 1). Holmquist et al (2018) determined that soil carbon density did not vary significantly by depth, and that the probability distribution of soil carbon density was described well by a normal distribution, truncated so that values could not be less than zero. They also determined utilizing a single average value for all wetlands was more parsimonious and precise than stock estimates based on available maps of soil carbon.
The previous NGGI (EPA 2017) made two assumptions about carbon changes during wetland conversion events that were not considered in the error propagation. First, the depth of soil lost to conversion was based on a range of values reported for aquaculture and salt production pond construction (0.5-2.5 m; IPCC 2014) but was fixed to 1 m. In the NGGI, this value was applied to wetland areas that converted to open water as indicated by C-CAP.
Because wetland to open water conversion events were dominant in our accounting and the IPCC depth intervals for degradation were largely not applicable, we represented uncertainty regarding this assumption by using a uniform distribution ranging from 0.5-1.5 m (table 1) to represent a wide distribution centered on 1 m. This uncertainty reflected a consensus from our coauthor group and reflected an expert assumption rather than data, as we could not readily locate or ingest any relevant data. The NGGI also assumed that 100% of the carbon released by conversion from coastal wetlands to open water is lost to the atmosphere. However (Lovelock et al 2017) reviewed available studies and estimated 25%-50% of terrestrial carbon delivered to the marine environment was buried in ocean sediments (Baldock et al 2004, Cai 2011, Blair and Aller 2012. Therefore we represented the fraction lost back to the atmosphere as a uniform distribution ranging from 50%-75% (table 1).

Biomass flux data
We utilized biomass data from (Byrd et al 2018) to generate emissions factors for emergent wetlands. We accounted for forested wetland biomass using a synthesis of tree diameter at breast height (DBH) for mangrove and tidal freshwater forested plots, then converting DBH to above ground biomass using allometric equations cited within the data source, or originating from a similar representative study (supplemental information: section 2.3.2). We represented scrub/shrub data using a subset of the Byrd et al (2018) biomass data, plots that were dominated by the shrub Iva frutescens, and a subset of the forested biomass dataset, plots in which average tree heights were lower than 5 m. We converted biomass to organic carbon using a conversion factor of 0.441 (Byrd et al 2018). We represented above-ground biomass with lognormal distributions because the data exhibited skewed positive distributions (table 1; supplemental figure 2). We represented CH 4 fluxes using a normal distribution for estuarine wetlands because while the vast majority of sites indicated a net emissions scenario, one oligohaline site in New Jersey displayed netuptake of CH 4 for much of the two years reported   (2)) 10 000 times, simulating the underlying data using random draws from the probability distributions for 145 random variables (supplemental information:  We also propagated uncertainty for nine emissions factors or emission factor components (table 2). For normally distributed variables we randomly drew the same number of datapoints from literature review from the probability distribution then represented the emissions factor or component as the mean of the randomly drawn data. For uniform distributions, we randomly drew a single value. For emissions factors that were lognormally distributed we randomly redrew the underlying data as in normal distributions but represented the central tendency of using the exponentiated logmean. This choice is consistent with IPCC Wetlands Supplement guidance, however arithmetic means are often used for lognormally distributed emissions factors (Levy et al 2017). Because the goal of this paper is to quantify the effect of assumptions on the inventory, we repeated the uncertainty analysis using the arithmetic mean of lognormally distributed values (supplemental information: section 3.2; supplemental figure 4).

Sensitivity analysis
We performed a one-at-a-time sensitivity analysis (Metsaranta et al 2017), meaning we categorized sensitivity of the US scale emissions and removals to assumptions, datasets, and mapping accuracies by manipulating one input at a time and recording the effect. For each random variable we re-calculated the coastal wetland total GHG emissions and removals using the 0.025 quantile and 0.975 quantile values from Monte Carlo analysis, while fixing all others at their median value. We reported sensitivity of the inventory to each input as the difference in the total flux between using the input's minimal and maximal settings.
The sensitivity analysis also helped test the effect of some of the fundamental assumptions. For example, CH 4 fluxes need to be converted to CO 2 e, and there is  1 ) which more effectively represent the system (Neubauer and Megonigal 2015). We quantified the effect of that choice by calculating the inventory using a GWP and median values for all other inputs and then recalculated changing only the GWP to SGW/CP (Neubauer and Megonigal 2015). Also, we tested the assumption of relying on the coastal lands definition for determining how much palustrine wetland area to include in the inventory compared to a tidal wetlands definition from the National Wetlands Inventory (NWI) (Hinson et al 2017. For this alternative analysis, we included all C-CAP palustrine wetlands intersecting an NWI-based tidal wetlands map  and treated all palustrine mapped areas as fixed. In the sensitivity analysis we calculated the difference in total inventory between the default settings and the NWI based mapping strategy. We also we repeated the sensitivity analysis using the arithmetic mean of lognormally distributed values, and discuss the results further in the supplemental information (section 3.2; supplemental figure 5).

Results and discussion
3.1. Initial assessment of estimated area The Monte Carlo analysis combining C-CAP and LiDAR DEMs define a total area of interest with a median of 3.56 million hectares (M ha; figure 5). Stable wetlands were the largest category (figure 5) with estuarine emergent wetlands dominating (1.82 M ha), followed by palustrine forested wetlands (0.68 M ha), palustrine emergent wetlands (0.54 M ha), and estuarine forested wetlands (0.19 M ha). Of the wetlands that changed to or from other categories, loss of emergent wetlands to open water was the most dominant classification. Conversion from open water to emergent wetlands was the next most important conversion but only made up for one third of the area converted from emergent wetlands to open water. The NWIbased strategy mapped fewer palustrine wetlands, especially palustrine forested wetlands, defining a total area of interest of 2.86 M ha.

Uncertainty in the CONUS 2006-2011 coastal wetland inventory
Coastal wetlands were likely to have acted as a netsource of GHG from 2006 to 2011 (figure 6; table 1;  supplemental table 7). Across the 10 000 Monte Carlo iterations median total net-emission was −10.3 million tonnes (M tonnes) of CO 2 e per year (yr −1 ) over five years with a confidence interval ranging from −1.6 to −21.3 M tonnes CO 2 e yr −1 . Although the confidence intervals were wide they were strictly negative, which support the conclusion of net-emissions from 2006-2011. Separating estuarine wetlands, which have lower CH 4 emissions, and palustrine wetlands, which have higher CH 4 emissions, indicates that both classes are more likely to have acted as net-emitters (table 2). However, estuarine wetlands emissions were more likely occurring due to wetland conversion events (figure 6). While overall stable and gaining estuarine wetlands acted as a net-sink and stable and gaining palustrine wetlands a net-source according to their median values, both categories had uncertainties spanning both net-emissions and net-storage scenarios.
3.3. The dominant contributions to national-scale uncertainty CONUS-scale total flux was most sensitive to inputs in four major classes: uncertainty in emissions and burial data, assumptions about how to calculate the inventory, C-CAP 2006-2011 change detection accuracy, and C-CAP 2011 classification accuracy (figure 7; supplemental table 7). Overall the inventory was most sensitive to uncertainty in the underlying emissions and storage data, and to assumptions made. Uncertainty arising from the probabilistic coastal lands mapping was not a dominant contributor to total uncertainty in this framework.
Uncertainty in palustrine CH 4 emissions, had the greatest effect on the inventory estimates for CONUS coastal wetlands, 11.6 M tonne CO 2 e yr −1 (figure 7; supplemental table 7). The average depth of soils lost to erosion, extraction, or drainage, was second most impactful and had a 9.4 M tonne CO 2 yr −1 . Estuarine CH 4 emissions were also important and had a 8.5 M tonne CO 2 e yr −1 effect. Soil carbon burial rate had a 5.2 M tonne CO 2 e yr −1 effect and assumptions made about the fraction of soil carbon lost to the atmosphere had a 3.9 M tonne CO 2 e yr −1 effect.
The decision to use GWP over SGWP/CP had a median effect of 8.8 M tonnes of CO 2 e yr −1 . The alternate choice moved the estuarine stable and gains sector from net-storing (+2.2 M tonnes CO 2 e yr −1 ) using GWP to net-emitting (−2.0 M tonnes CO 2 e yr −1 ) using SGW/CP (figure 7; supplemental table 8). Emissions from stable palustrine wetlands overtook palustrine soil and biomass losses when using SGW/CP. The SGW/CP choice increased the estimate of total CO 2 e emissions 89% over the traditional GWP model. Uncertainty in mapping also contributed to uncertainty in the inventory. 2006-2011 change detection was the most uncertain mapping category. Notably, we drew a different conclusion regarding the 2006-2011 change than the official C-CAP accuracy assessment (McCombs et al 2016). We concluded that change was under-mapped while McCombs et al concluded change was over-mapped (supplemental information: section 3.1; supplemental figure 3). This occurred because McCombs et al raw counts for the accuracy assessment matrix and we used a proportional area matrix (Olofsson et al 2014).
Sensitivity of the inventory to input uncertainty dropped precipitously for the remaining inputs. These include the decision between using a coastal lands definition to identify palustrine wetlands and the strictor NWI-based definition (2.0 M tonne CO 2 e yr −1 effect) (figure 7; supplemental table 7). The effect of uncertainty in fluxes associated with changes in forested and scrub/shrub biomass and carbon density for eroded soils range from 0.6 to 0.1 M tonnes CO 2 e yr −1 . Classification accuracy introduced uncertainty for estuarine aquatic beds, open water, unconsolidated shore and palustrine aquatic beds. In our accounting, these all indicate soil loss events.

Implications for future research
Uncertainty estimates are important components of complete and transparent GHG inventories (EPA 2017). Uncertainty information is not intended to dispute the validity of the estimates, but rather to help prioritize efforts to improve accuracy and guide future decisions. We recommend improving process models for CH 4 emissions and soil carbon burial, increasing the number of observations for key inputs, and developing more detailed and accurate maps for categories relevant to coastal wetland carbon cycling and inventory estimates.
3.4.1. Improving process models for CH 4 emissions and soil carbon burial The uncertainty and sensitivity analysis presented herein suggest that uncertainty could be reduced at the scale of the contiguous US primarily by improving data availability and process-based models for CH 4 emissions, CH 4 radiative forcing, and carbon burial rates. Net-wetland CH 4 emission combines CH 4 production by methanogenic archaea under anaerobic conditions, CH 4 oxidation and consumption by methanotrophic bacteria mainly under aerobic conditions, and CH 4 transport to the atmosphere (Conrad 1989, Whalen 2005. Major controls of these processes include: water table position; soil temperature; sulfate supply and potential production of hydrogen sulfide, a methanogen toxin, for which salinity is a proxy for; vegetation, including both biomass and species composition, which may facilitate CH 4 transport from soil production sites to the atmosphere; and primary production of vegetation, since new photosynthate may be a substrate for methanogenesis (Wang et al 1996, Walter andHeimann 2000). Large discrepancies have also been noted between chamber and eddy covariance measurements of CH 4 fluxes (Hendriks et al 2010, Krauss et al 2016), suggesting the need for additional comparisons between these two methods.
The use of GWPs serves an important policy need because GWPs are transparent and tractable. However, GWPs are an oversimplification because modeling CO 2 e in power units (W m −2 ) that relate directly to radiative forcing is several steps removed from actual climate impacts such as changes in temperature, precipitation, and sea level. The SGW/CP model is equally transparent and tractable, but more closely represents reality by acknowledging that changes in GHG emissions persist over several years (Neubauer and Megonigal 2015). Therefore we recommend that SCW/CP's should be considered for adoption by the IPCC. When considering the consequences of GHG inventory data beyond the IPCC context, ecosystem scientists and policy analysts should discuss metrics that are independent of time frames, such as switchover time, as they are more informative of the longterm impacts (Frolking and Roulet 2006). Our uncertainty analysis is focused on variables that are inputs to GWP and SGW/CP models, but there is an ongoing need to address the uncertainty introduced by using these models to underpin climate policy.
Currently, IPCC guidance recommends applying separate carbon burial rates to different wetland types and ecoregions to increase accuracy. However, multiple studies suggest other relevant geographic and methodological factors need to be considered in the US inventory. In some locations, accelerating sea-level rise is expanding the area conducive to carbon burial, potentially increasing carbon burial rates (Kirwan andMudd 2012, Hill andAnisfeld 2015). A sensitivity analysis of the marsh equilibrium model highlighted Figure 6. CONUS inventory results of 10 000 Monte Carlo simulations, shaded to distinguish simulations resulting in a net-emission (orange) or a net-storage (blue) scenario. The thick gray vertical line at 0 separates these scenarios. Points indicate medians, and black horizontal lines the upper and lower 95% confidence intervals. Top panels separate fluxes from estuarine and palustrine wetlands, and from wetlands that were lost from those that were stable or gained area. The bottom panel shows net-annualized emissions from 2006-2011. relative sea-level rise, plant productivity and relative tidal elevation as dominant drivers of carbon sequestration in stable wetlands (Morris and Callaway in Press). Elevation/inundation gradients were correlated with sediment accretion dynamics in San Francisco Bay (Callaway et al 2012). Finally, there are many ways to measure carbon burial that integrate different time scales: decades-137 Cs, centuries-210 Pb, or millennia-14 C (Turetsky et al 2004). We recommend that future studies rectify the complex interactions between regional variability in relative sea-level rise, plant productivity, local elevation/inundation dynamics, and the potential effects of measuring this carbon burial using differing methods.

Increasing data availability for key inputs
Some inputs in the inventory could be improved by targeted studies and additional data collection, including soil depth affected by conversion to open water, and percent carbon returned to the atmosphere upon loss.
Available data on elevation loss due to the diking of wetlands for agriculture (Drexler et al 2009), and the mass lost to 50 cm depth following vegetation die off Uncertainty in assumptions about carbon loss is not unique to this study and was discussed explicitly in a recent global analysis of soil and biomass loss from mangrove conversions (Sanderman et al 2018), which report that the rate and forms of carbon loss may depend on soil type and depth (Donato et al 2011). Because assumptions about loss events vary from study to study, and because of the fact that these assumptions are dominant contributors to uncertainty (figure 7), future research should prioritize empirical and modeling studies that constrain depth and percent carbon loss due to wetland conversion events.

Improving mapping capacity of tidal carbon relevant gradients
The wetlands supplement of the IPCC report provides two CH 4 emissions factors for wetlands, one for fresh to brackish conditions and another for higher salinity (18 ppt threshold) (Poffenbarger et al 2011, Bridgham et al 2013. However, C-CAP salinity categories do not match these categories, instead mapping estuarine and palustrine (5 ppt threshold; figure 3). This inconsistency limits our ability to confidently assess the true GHG balance for saline wetlands at the national scale. We propose developing maps and data to support at least three categories of salinity-saline (>18 ppt), brackish (0.5-18 ppt), and fresh (<0.5 ppt)-in order to reduce uncertainty in landscape scale CH 4 emissions from coastal wetlands (figure 4).
Existing remote sensing approaches for vegetation and inundation dynamics could improve mapping both CH 4 emissions and carbon burial rates. Recent strides in mapping coastal wetland vegetation biomass (Byrd et al 2018), vegetation species classification (Immitzer et al 2016) and seasonal dynamics (Mo et al 2015) could provide more detailed vegetation descriptions that would be a proxy for salinity zones. For inundation/elevation regimes, extensive coastal DEMs are available, but lack the accuracy to adequately map tidal flooding depth and inundation time at relevant scales and could be improved by integrating additional remote sensing and modeling (Hladik et al 2013, Parrish et al 2014, Buffington et al 2016. Future studies should quantify the precision needed for DEMs in the tidal zone. Currently soil emissions factors are calculated using tabular data, however improvements in mapping should be leveraged to support spatially-explicit approaches in future versions of the inventory incorporating trends in productivity and seasonality (Knox et al 2017), variation in carbon mineralization rates (Mueller et al 2018), edaphic factors and geomorphology (Rovai et al 2018). Many improvements may be forward-looking and hindcasting may not be appropriate (Byrd et al 2018), and spatially-explicit approaches should only be utilized only if they actually do improve precision and accuracy of inventorying compared to simpler approaches .
Biomass changes were not a top contributor to uncertainty, but changes in forested and scrub/shrub biomass were the ninth and fifteenth contributors to uncertainty respectively. This study quantified the effect of uncertainty by upscaling means and uncertainties from multiple field studies, however remote sensing approaches using LiDAR, RADAR, object based image detection, and optical remote sensing, can all be used to characterize biomass changes on local to regional scales (Byrd et al 2018). Future studies could expand the uncertainty and sensitivity analysis to capture the effect that uncertainties in genus-specific assessments of wood density (Jenkins et al 2003), biomass carbon content (Byrd et al 2018), and the contributions and decay rates of downed wood (Krauss et al 2018).
C-CAP's accuracy was not a dominant contributor to the overall uncertainty in the inventory, but we were only able to quantify this from 2006-2011. C-CAP is available for the entire CONUS coastal zone from 1996-2011, and trends were extrapolated out back to 1990 and forward to 2015 for the NGGI inventory. Future studies are needed to assess accuracy for earlier time steps.

Conclusions
Uncertainty in CONUS coastal wetland greenhouse gas inventory estimates comes mostly from lack of knowledge on CH 4 emission variability, the fate of soil carbon post-conversion, and an inability to extrapolate trends to available map products. Switching from GWP to SGW/CP increases the overall calculation of CO 2 e impacts from 2006-2011 by 89%. The underlying mapping products, C-CAP, and the probabilistic coastal lands layer for mapping tidal freshwater wetland extent, were not dominant contributors to uncertainty. However, the inventory development could benefit from improved change detection, accuracy assessments that go back further in time, and improved mapping of intermediate salinities and inundation gradients. Our analysis provides a framework to track improvements to the coastal wetland GHG inventory as more data and improved process knowledge become available. The data used here were not collected for the purpose of the inventory; future improvements will demand targeted investment in data collection, model improvements, spatial product development, and more extensive, independent accuracy assessments.