on the species and either height or diameter‐at‐breast‐height, DBH (Jenkins et al., 2004). Canopy height, and various derived metrics, are often measured to verify models based on remote sensing, but field measurement with a clinometer can be time intensive and logistically impractical when vegetation is dense. Regardless, documented relationships between DBH and height for a given species can be used to extrapolate aboveground biomass to large spatial extent if canopy height and species composition are available via remote sensing (Smart et al., 2020). Though more commonly employed in forested settings, this method may also be used in herbaceous wetlands (Trilla et al., 2009).
Field sampling of soil characteristics, such as the depth of the organic profile, bulk density and its vertical profile, and chemical composition, is particularly important for estimating belowground carbon stocks. Specifically, the quantity and quality of soil pore water drive carbon accumulation and long‐term storage. For example, salinity is among the dominant drivers of vegetation change in coastal wetlands (Taillie et al., 2019; Williams et al., 1999), while hydroperiod holds this role in inland wetlands (Ross et al., 2003). While more sophisticated soil sampling may reveal changes with depth, a crude sample collected with a small shovel and a Mehlich III extraction is often sufficient to determine salinity, cation exchange capacity, and the ratio of organic to mineral soil (Taillie et al., 2019). In wetlands with standing water, contamination with surface water is inevitable, and thus is susceptible to daily fluctuation in salinity as a function of precipitation, evapotranspiration, and other water processes (Herbert et al., 2015). As such, coastal wetlands should be sampled during both dry and wet periods to best understand the variation in salinity as a function of precipitation and salinization (Herbert et al., 2015). Though evidence of salinization may persist longer in soils compared to surface waters (Chagué‐Goff et al., 2012), soil cation concentrations may not accurately reflect the history of saltwater exposure over multiple years.
1.3.2. Remote Sensing
Spaceborne observations of the land surface provide reflectance information that is related to ecosystem optical and emissivity properties. Optical remote sensing typically measures surface reflectance in wavelengths ranging from ultraviolet to the longwave‐infrared regions, whereas microwave observations (including active and passive approaches) use longer wavelengths to observe brightness temperature (as it relates to soil moisture), and active LIDAR records light travel time that is related to how ecosystem structure modifies the profile of a laser beam return. Remote sensing per se does not measure above‐ or belowground structure directly; for example, the surface retrievals, referred to as Level 2 products, include the atmospherically and geometrically corrected surface information, but require the application of algorithms to derive vegetation or soil properties (i.e., Level 3 products) or models to produce process‐level information (i.e., Level 4 products).
In this context, remote sensing observations can provide spatially consistent information on ecosystem type and composition as well as ecosystem structure (i.e., via canopy height models), both of which can be used to derive information on above‐ and below ground carbon stocks. The use of remote‐sensing derived information also requires knowledge on how to integrate temporal revisit as well as ground sampling distance (GSD), or spatial resolution, which can introduce uncertainties in terms of detection of seasonally flooded wetlands, or introduce “double counting” of wetlands when using coarse spatial resolution data (>500 m).
Some examples of how remote sensing has informed wetland carbon stock accounting are included in this chapter; for example, where remote sensing data based on the MODIS sensor aboard AQUA and TERRA has been used to map vegetation indices linked to wetland habitat. Or where passive and active microwave observations (i.e., NASA’s AMSR and SMAP missions) have been used to derive permafrost and soil moisture information used in ecosystem models. And where radar and lidar information from the German Aerospace Center (DLR) mission Tandem‐X and airborne instruments (e.g., NASA’s G‐LiHT, ECOSAR) have been used to map canopy height and to relate this to biomass models to estimate aboveground carbon stocks. Advances in remote sensing, particularly in improved spatial resolution capabilities, will contribute to better delineation of wetland types using high‐resolution optical data (see Cooley et al., 2019) and for determining aboveground structure of vegetation in the case of the International Space Station (ISS) small‐footprint waveform lidar mission “GEDI.” In addition, operational and emerging hyperspectral remote sensing missions such as on the ISS like the DLR Earth Sensing Imaging Spectrometer (DESIS), and “free flying” spacecraft proposed by the European Space Agency CHIME and NASA Surface Biology and Geology (SBG) missions, have potential to use more sophisticated algorithms relating surface reflectance to soil carbon content.
1.3.3. Ecosystem Modeling
Modeling approaches used to estimate above‐ and belowground stocks of wetland carbon can be partitioned to diagnostic and prognostic approaches. Diagnostic modeling tends to rely more heavily on observational data, such as remote‐sensing derived vegetation distributions, to initialize and update model states over time. In contrast, prognostic modeling relies on “first principles,” where fundamental relationships linking biophysical information, carbon uptake, carbon allocation, and carbon turnover determine vegetation and soil carbon stocks. Many examples of diagnostic and prognostic wetland modeling comparisons exist in the literature and the uncertainties between and among these approaches remains high (Fisher et al., 2015).
To some extent, all the estimates of above‐ and belowground carbon stocks require empirical or process‐based models either to scale site‐level measurement, or convert volumetric to mass‐based estimates, or to attribute changes over time due to climate and land use. However, the relationship of the models to underlying data reflects a gradient of data‐constrained to process‐based representation of how estimates are made. The estimates provided in this chapter span this gradient, with some emphasizing empirical approaches, where carbon density is multiplied by wetland area, others using semi‐empirical methods, where the carbon densities are spatially constrained by remote sensing observations, or where prognostic models are used to assess long‐term historical or future climate feedbacks on carbon turnover and respiration losses.
1.4. ESTIMATES OF WETLAND STOCKS BY WETLAND TYPES
1.4.1. Mangroves
Mangroves and other tidal wetlands have the highest carbon density among terrestrial ecosystems (McLeod et al., 2011). Although they only represent 0.3% of the total forest area (or 0.1 % of land area), C emissions from mangrove destruction alone at current rates could be equivalent to up to 1–10% of carbon emissions from deforestation (Donato et al., 2011; Richards et al., 2020). From 1996 to 2016, 158.4 Mt of C (1.8%) was lost from mangrove ecosystems (Richards et al., 2020), with total emissions per year ranging from 25–29 Tg CO2‐equivalent (Friess, 2019). Due to their location along highly populated coastlines, they are under significant threat from anthropogenic activity as well as sea level rise and climate extremes. In fact, it is estimated that over 50% of mangrove forests and tidal marshes have been destroyed over the past 60 years, at a rate of 1% to 2%/yr (McLeod et al., 2011), although contemporary (2000–2016) rates of loss have reduced (0.13%/yr), particularly from anthropogenic destruction (Goldberg et al., 2020). The high C sequestration coupled with the high risk of future destruction makes mangroves a prime candidate for carbon mitigation initiatives such as the United Nations Collaborative Programme on Reducing Emissions from Deforestation and Degradation in Developing Countries (UNREDD and REDD+).
One of the main challenges to implementing carbon mitigation projects is measuring carbon efficiently, effectively, and safely. In mangroves especially, the extreme difficulty of the terrain has hindered the establishment of sufficient field plots needed to accurately measure carbon on the scale necessary to relate remotely sensed measurements with field measurements at accuracies of 80 to 90% as required for REDD and other carbon trading mechanisms (UNREDD, 2010). Furthermore, most intensive mangrove sites are established in South‐East Asia, Australia, and Latin America, with a large gap in knowledge in African mangrove ecosystems.
Mangrove aboveground biomass (AGB) and aboveground carbon (AGC) are strongly related to mangrove height, thus the largest