Terrestrial Wetlands

Prepared by Carl Trettin,¹ Wenwu Tang,² and Steven Campbell^3
1USDA Forest Service; ²University of North Carolina, Charlotte; ³USDA Natural Resources Conservation Service

13A.1 Introduction

This appendix provides the methodologies and data used to estimate the area and carbon pools of terrestrial wetlands in North America. Since the First State of the Carbon Cycle Report (SOCCR1; CCSP 2007), several developed geospatial databases have provided the opportunity to improve the estimation of carbon pools beyond what is feasible using area density factors. The development of the Gridded Soil Survey Geographic (gSSURGO) database by the U.S. Department of Agriculture’s (USDA) Natural Resources Conservation Service (NRCS) was a particularly important advancement, availing gridded soil survey information for the United States and Puerto Rico. Similarly, the USDA Forest Service’s Forest Inventory and Analysis (FIA) database uses forest biomass data for the United States, thereby facilitating its incorporation into carbon pool assessments. Sections 13A.2–13A.6 detail the data and methods used to obtain the reported wetland area and carbon pools.

13A.2 Conterminous United States

13A.2.1 Approach

The U.S. Fish and Wildlife Service’s (USFWS) National Wetlands Inventory (NWI) was used as the basis for identifying terrestrial (i.e., nontidal) freshwater wetlands within the conterminous United States (CONUS) and for distinguishing between forested and nonforested wetlands. Subsequently, geospatial databases were used to calculate the carbon pools in soils and forests. Specifically, the gSSURGO database was used to calculate soil carbon, and the FIA database was used to calculate forest carbon based on the reported biomass. A carbon pool density factor was used for the nonforest vegetation biomass because an appropriate geospatial database was not available.

13A.2.2 Data

The datasets used for analyses of the wetland area and carbon pool computations are summarized in Table 13A.1.

13A.2.3 Results

Wetland Area

According to NWI data, there are 395,197 km² of terrestrial freshwater wetlands in CONUS, 54% of which are forested and 46% nonforested (see Table 13A.2). The estimate of forested freshwater wetlands is within 2% of the most recent NWI report; the total area of freshwater forested wetlands is calculated as 213,914 km², compared with 208,912 km² for 2009 from Dahl (2011). This area is smaller than the wetland area used in SOCCR1 (405,670 km²; CCSP 2007) because that report also included tidal wetlands. Mineral soils compose 79% of the terrestrial wetlands, with 21% being organic or peat soils (see Table 13A.2). The distribution of wetlands among soil (organic and mineral) and vegetation (forest and nonforest) categories among states is presented in Figure 13A.1.

The accuracy of the NWI data is considered to be over 90% for large wetlands (i.e., those > 1 hectare); uncertainties increase with smaller wetlands (Nichols 1994). Independent field-based studies also have been conducted to evaluate the accuracy of the NWI data for wetland mapping. The reported accuracies ranged from over 90% of overall accuracy in Michigan, Maine, and Massachusetts (see Kudray and Gale 2000; Nichols 1994; Swartwout et al., 1981) to underestimation of wetland area by 39% in Vermont (see Morrissey and Sweeney 2006). With these issues considered, the NWI data are recognized as a reasonable source for estimating wetland area, particularly at large spatial extents, and thus are the source for national-level reporting.

Wetland Carbon Stock Estimation

Carbon stocks were calculated based on soil carbon content calculated from gSSURGO, forest biomass extracted from the FIA database, and a biomass density factor for nonforest vegetation. Forest vegetation consists of a carbon stock of about 0.878 petagrams of carbon (Pg C), with 79% occurring on mineral soils; nonforest vegetation contributed approximately 0.093 Pg C (see Table 13A.3). Integrating forest biomass and soil carbon pools yields approximately 13.5 Pg C in terrestrial wetlands (see Table 13A.4). The breakdown of carbon within forested and nonforested wetlands and of mineral and organic soils by state is summarized in Table 13A.4.

13A.3 Alaska

13A.3.1 Approach

The NWI and traditional soil surveys of Alaska are not available for the entire state. Fortunately, Clewley et al. (2015) recently published an inventory of wetlands based on remote-sensing data that used the Cowardin Classification system for representing the distribution of wetland types. Similarly, NRCS has produced a gSSURGO dataset for Alaska. Accordingly, those datasets were used as the basis for estimating the terrestrial wetland categories and carbon stocks following the same general approach used for CONUS. The combination of the wetland and carbon stock assessment with the distribution of frozen wetlands is considered to provide a comprehensive assessment of wetlands for the state.

13A.3.2 Data

Table 13A.5 presents the principal datasets used in this study that include information on soil, wetlands, soil organic carbon, permafrost, and elevation.

13A.3.3 Results

Wetland Area

The total area of freshwater wetlands in Alaska, based on the Clewley et al. (2015) database, is 579,645 km² (see Table 13A.6). The wetland data were classified from ALOS PALSAR² remote-sensing data using a random forest-based classifier. The data were processed using the adjustment factor employed by Clewley et al. (2015) to calculate the total area of freshwater wetlands, and data that overlapped into Canada were excluded. The overall accuracy of the classification is 84.5% for distinguishing specific wetland types and 94.7% for distinguishing wetlands with uplands (Clewley et al., 2015). The NWI class was used to aggregate the areas into forested and nonforested types.

Also calculated was the total area of wetlands in Alaska from STATSGO2 data using the percent in hydric soil attribute (“hydric_pct”; i.e., the percent in hydric soil). The total area is 587,143.9 km² based on the STATSGO2 percentage of hydric soils, which is very close to that provided by the Clewley et al. (2015) dataset.

Soil organic carbon data from STATSGO2 were employed to estimate the area of organic soils in Alaska, using the variable named “hydric_org_pct” (i.e., the percent in hydric organic soil) as the basis. This variable was multiplied by the area of map units (polygons) in the STATSGO2 dataset to obtain the area of peatland within each map unit. The total area of peatlands estimated from STATSGO2 using the hydric organic soil attribute is 107,057 km².

Incorporating the distribution of organic soils into the overlay analyses yielded the distribution and area of the four wetland categories (see Figure 13A.2). The total area of the four wetland categories of freshwater wetlands in Alaska are summarized in Table 13A.6.

Assessing the overlap of wetlands and permafrost areas provided a basis for distinguishing carbon stocks. The use of the USGS probability map of permafrost provided a cut-off threshold of 60% to permafrost occurring within 1 m of the surface (with a 30-m spatial resolution). The resultant area of permafrost is 405,891 km², compared with 548,503 km² based on permafrost 2 m in depth from STATSGO2 data. Overlaying the USGS permafrost area with the wetlands shows that the total area of wetlands within the permafrost region is 267,887 km², which is approximately 46% of the total wetland area. The areas of the four types of freshwater wetlands in Alaska within permafrost or nonpermafrost regions are presented in Table 13A.7.

Wetland Carbon Stocks

Ecosystem carbon stocks for the four wetland categories were derived from soil carbon stocks from USDA STATSGO data, biomass carbon data from FIA for forests, and a density factor for nonforested wetlands (see Table 13A.8).

Partitioning the ecosystem carbon pools among wetlands in permafrost and nonpermafrost zones is provided in Table 13A.9. Approximately 46% of the wetland carbon pool occurs within the permafrost areas.

13A.4 Puerto Rico

13A.4.1 Approach

The approaches to quantifying the distribution of terrestrial wetlands and the associated carbon pools for Puerto Rico follow those of CONUS, where a suite of datasets was used, including gSSURGO, NWI, Value-Added Look Up Table Dataset, Cartographic Boundary Shapefile, and FIA Forest Biomass Dataset. An overlay analysis was conducted between NWI and gSSURGO to identify vegetation and soil types for wetlands. Cartographic Boundary identified the boundary of Puerto Rico. The FIA Forest Biomass dataset provided the forest biomass information. Soil Data Development Tools for ArcGIS were used to extract the soil class of freshwater wetlands.

13A.4.2 Data

Datasets used in this study are summarized in Table 13A.10.

13A.4.3 Results

Wetland Area

The total area of terrestrial wetlands derived from NWI data is 311.4 km². However, gSSURGO data coverage was missing for approximately 9.8% of the terrestrial wetland area. Distributing the area of missing soil data among the forested and nonforested categories yields the final area of the four wetland categories (see Table 13A.11).

Ecosystem Carbon Pool

Ecosystem carbon pools, including soil and biomass, for freshwater wetlands in Puerto Rico are summarized in Table 13A.12.

13A.5 Canada

13A.5.1 Approach

Canadian terrestrial freshwater wetlands were estimated based on a combination of spatial data because there was not a single wetland database that could produce estimates of organic and mineral soil wetlands and of forest and nonforest vegetation.

13A.5.2 Data

Datasets in this study are summarized in Table 13A.13.

13A.5.3 Results

Organic and Mineral Soil in Forested and Nonforested Terrestrial Wetlands in Canada

Organic and mineral soils for forested and nonforested wetlands were estimated by overlaying land-cover datasets (GLWD and North America land-cover data) with soil datasets (FAO soil data, Peatland Database of Canada, and Soil Landscape of Canada). Those analyses routinely underestimated wetland area compared with estimates in published reports, especially for organic soils (Tarnocai 2006; Warner 2005; see Table 13A.14 for examples of the differences in wetland area based on data sources).

Because the accepted area of peatlands is 1,135,610 km² as reported by Tarnocai (2006), it was used as the basis for the total peatland area; the 16,375 km² of permafrost peatlands (Tarnocai et al., 2011) were excluded from the final area table (see Table 13A.15). Wetland-specific soil types from the Peatlands of Canada and the Soil Landscapes of Canada datasets were used to identify mineral and organic soil wetlands. The analysis of wetland area in Canada is based on the Peatlands of Canada database, which was updated from its previous version. The accuracy of the wetland area estimated using this database is equal to or greater than 66%, as suggested by Tarnocai (2009). The distribution of terrestrial freshwater wetlands in Canada is presented in Table 13A.15. For comparison, Warner (2005) reported 1.056 million km² of peatland area (organic soil wetland) for Canada, a difference of 7%.

Carbon Pools

Carbon pools of the Canadian wetlands were calculated using the area carbon density factors for the four wetland categories, derived from CONUS (see Table 13A.16).

13A.6 Mexico

13A.6.1 Approach

An assessment of terrestrial wetlands in Mexico was used as the basis for identifying wetland areas and soil types. The North American Land Cover dataset (see Table 13A.17) and a recent dataset from Mexico were used to segregate the wetlands into vegetation categories. Area carbon density factors were used to develop the estimates of wetland carbon pools.

13A.6.2 Data

The datasets used to estimate the area of terrestrial wetlands in Mexico are presented in Table 13A.17.

13A.6.3 Results

Organic and Mineral Soil in Forested and Nonforested Wetlands in Mexico

This estimate of freshwater wetlands is greater than other reported values (e.g., 31,000 km²; Bridgham et al., 2006). A review of the map units from the Mapa Potencial de Humedales could not ensure that selected wetlands were adequately constrained to freshwater systems (due to problems with data code translations). Accordingly, the calculated wetland area was reduced by 25% to provide a conservative estimate (see Table 13A.18), thereby reducing the accuracy to at least 75%. The metadata for the database did not provide an estimate of the mapping error.

Acknowledgments

Spatial analyses were conducted at the Center for Applied GIScience, University of North Carolina, Charlotte, by Yu Lan, Jiyang Shi, Wenpeng Feng, Yu Lan, Douglas Shoemaker, Minrui Zheng, and Xiang Zhao. Julie Arnold, USDA Forest Service Southern Research Station, assisted with the literature and preparation of data tables and spreadsheets. Also used were computational facilities at the Portland office of the USDA Natural Resources Conservation Service.