Mapping methane reduction potential of tidal wetland restoration in the United States

The study evaluates how restoring tidal exchange to impounded coastal wetlands in the contiguous United States can reduce methane (CH4) emissions and yield net climate benefits. Natural climate solutions (NCS) in blue carbon ecosystems (salt marshes, mangroves, seagrasses) are considered promising due to rapid implementation, cost-effectiveness, and co-benefits. Saline tidal wetlands typically emit minimal CH4 because sulfate reduction suppresses methanogenesis, making them favorable for climate mitigation. Many coastal wetlands have been impounded by dikes, roads, and railroads, often freshening conditions and increasing CH4 emissions. The research aims to: (1) map potential restoration opportunities using GIS and public datasets; (2) estimate emissions factors based on current and likely post-restoration salinity; (3) quantify which restoration types offer net-cooling on 100-year time horizons; (4) provide a spatially explicit national estimate and parcel-level mapping of CH4 reduction potential. The work updates and refines earlier national estimates that relied on regional upscaling and lacked mapped project locations.

The paper builds on literature indicating high soil carbon storage rates in tidal wetlands and low CH4 emissions under saline conditions. Prior national assessments (e.g., Fargione et al., Kroeger et al.) estimated large CH4 reduction potential by assuming extensive artificially freshened impoundments and near-elimination of CH4 after restoration. Methane emissions factors by salinity derive from Poffenbarger et al. (2011), updated for SOCCR-2, and converted to CO2e using sustained GWP. Additional references include IPCC Wetlands Supplement Tier 1 factors for drained lands, and studies on tidal restoration outcomes, estuarine salinity dynamics, and uncertainties in NCS scalability. The authors reassess previous assumptions about the extent of artificially freshened wetlands and post-restoration salinity using new mapping and expert surveys.

- Data integration: Combined three geospatial datasets: (1) National Wetlands Inventory (NWI) for wetland classes and impoundment status (using special modifier 'h'); (2) NOAA Coastal Change Analysis Program (C-CAP, 2006–2010) for land cover and a two-class salinity scheme (estuarine >5 ppt; palustrine 0–5 ppt); (3) Probabilistic coastal lands map for tidal elevation (below mean higher high water spring tide). - Mapping impoundments: Developed an algorithm to include tidal wetlands and open-water features potentially impounded. Included freshwater emergent/scrub-shrub wetlands intersecting coastal lands, intertidal waters, and certain estuarine deepwater features flagged as impounded. Added lakes/ponds within 90 m of mapped tidal features (iterated 3 times to capture chains). Included adjacent estuarine deepwater and rivers within 90 m buffers, constrained inland by HUC8 watershed boundaries; manually removed non-tidal inland artifacts. - Independent accuracy assessment: Used stratified random sampling within areas intersecting the USGS Protected Areas Database (PAD-US). Collected 436 high-quality expert-labeled control points via an R Shiny app, applying fuzzy set classifications to address ambiguity. Estimated unbiased areas and calculated user’s accuracy and area underestimation/overestimation, testing effects of definition strictness and time (1990 vs 2019). - Reference-wetland salinity mapping: Assigned potential post-restoration salinity to impoundments and low-elevation uplands by nearest-neighbor extrapolation from adjacent non-impounded reference wetlands (C-CAP salinity), filtering out small parcels (<1 pixel) and non-relevant classes. - Activities (area) estimation: Classified potentially restorable land covers below tidal elevation (agriculture, forest, grass/brush, open space/bare). Propagated classification uncertainty via Monte Carlo by simulating accuracy assessment matrices (treated as multinomial) and computing unbiased areas. Weighted areas by probability of being below tidal elevation where appropriate. - Expert surveys for scenarios: Conducted two surveys of site experts: (1) impoundment status/attributes; (2) current and post-restoration salinity across six classes (fresh, oligohaline, mesohaline, polyhaline, saline, brine), including data source (measurement vs expert judgment), impoundment drivers, and restoration plans. Derived frequencies of potential salinity changes to weight emissions factors when GIS classes were ambiguous or unchanged. - Emissions factors: Compiled CH4 emissions by salinity class from literature (skew-normal fits, bootstrap to derive medians and SE). Computed two-class factors (estuarine, palustrine) and drained-land CH4 factors (IPCC Tier 1 for agriculture, forest, grassland; bare assumed ~0). Applied -401±52 gCO2e m−2 y−1 for tidal wetland CO2 burial (log-normal). Converted CH4 to CO2e using sustained GWP of 45. Constructed weighted emissions factors for each conversion, combining estimated salinity change probabilities with CH4 differences. - Net-cooling likelihood and zero-inflation: Ran 10,000-iteration Monte Carlo sampling of emissions/removals factors and area to estimate the probability that each conversion yields net cooling over 100 years. Applied a censored (zero-inflated) approach: net-warming scenarios set to zero for NCS capacity. - Total flux and uncertainty: Summed conversion-type-specific CO2e fluxes (factor × area) to compute national totals with 95% credible intervals from Monte Carlo. - Candidate parcel mapping: Converted raster outputs (current class, reference salinity, tidal elevation probability, protection status, mean and 95% CI of emissions reduction) to polygons; retained polygons with net-cooling potential, in protected lands, ≥900 m², and ≥1 metric tonne CO2e y−1 potential. Joined parcels to PAD-US to summarize by state and managing entity; computed distances to nearest reference-salinity wetlands.

- National extent of impoundments: Estimated 0.53 ± 0.12 million ha of impounded features in the CONUS coastal zone. Existing maps (NWI) underestimate impoundments by ~50%; user’s accuracy for mapped impoundments was 88.4 ± 3.5%. - Areas by potential conversion (median, 95% CI): Impounded open water → palustrine had the greatest area (199,981 ha; 158,923–244,022). Impounded palustrine → estuarine area: 57,133 ha (45,255–69,780). Potential drained agriculture/uplands → estuarine each <50,000 ha. - Emissions factors (weighted, zero-inflated; gCO2e m−2 y−1): Impounded palustrine → estuarine offered the largest reduction (median 1727 ± 561). Open water → estuarine: 203 ± 80. Impounded palustrine → palustrine and open water → palustrine: ~46 ± 50 each on average. Drained bare and forest → estuarine often not net-cooling; all drained upland → palustrine conversions were net-warming in simulations. - Total CH4 reduction potential: 0.91 Tg CO2e y−1 (95% CI 0.42–1.60) nationally from reconnecting tidal impoundments and select drained lands. Despite smaller area, impounded palustrine → estuarine yielded the largest share (median 0.59 Tg CO2e y−1; 0.22–1.04). Impounded open water → estuarine: median 0.11 Tg CO2e y−1 (0.04–0.23). Palustrine-ending conversions and estuarine→estuarine had lower contributions due to lower factors/areas. Including drained upland→estuarine had marginal effect on totals. - Spatial distribution and management: 54% of potential occurs within protected areas. Identified 1,796 candidate parcels (≥1 t CO2e y−1) with median probability of occurrence on protected lands. States with highest shares: Louisiana 48%, California 26%, South Carolina 11%. Managing entities with large shares: US Fish & Wildlife Service 45%, State Fish & Wildlife 42%, State Departments of Natural Resources 11%. - Survey insights: Restoration planned for ~18% of identified impoundments overall; by coast: Pacific 60%, Gulf 13%, Atlantic 7%. 67% of impoundments are purposeful (e.g., waterfowl management, agriculture, salt production); 11% incidental (e.g., roads/railroads). Using only measurement-based survey responses slightly increased potential salinizing restorations (from 10.5% to 16.7%). - Comparison to prior estimate: New national estimate (0.91 Tg CO2e y−1) is an order of magnitude lower than earlier 12.0 Tg CO2e y−1, primarily due to reduced estimates of artificially freshened area (median 0.057 M ha here vs 0.48 M ha previously) and different emissions factors (e.g., post-conversion palustrine 172 gCO2e m−2 y−1 vs previously near-zero assumed).

The refined, spatially explicit assessment indicates that methane reduction from tidal reconnection is substantial but smaller than previously reported due to improved, conservative area estimates and updated emissions factors. The largest per-area benefits arise when impounded palustrine wetlands are restored to estuarine salinity; however, such opportunities are relatively rare compared to palustrine-ending conversions. The mapping suggests that many high-potential projects lie within protected areas, facilitating implementation and management coordination, with notable concentrations in Louisiana, California, and South Carolina. The analysis highlights key uncertainties: impoundments may be under-mapped (e.g., road-impounded wetlands), reference salinity mapping may be conservative, and expert surveys may understate salinity increases due to limited measurements. Future improvements in mapping (independent of NWI and C-CAP), emissions factor datasets (including direct measurements in impoundments/open waters), and accounting guidance (e.g., additionality of carbon burial, time horizon for CH4 GWP) could revise estimates. In the broader climate context, even full implementation of blue carbon NCS provides limited contribution relative to national emissions, but CH4 avoidance from impounded wetlands represents an anthropogenic source reduction with no non-permanence risk and meaningful co-benefits, aligning with policy priorities for rapid mitigation.

This study provides a conservative, spatially explicit national estimate and parcel-level map of methane reduction potential from tidal wetland restoration in the CONUS. By integrating multiple geospatial datasets, conducting an independent accuracy assessment, and applying expert-informed salinity-change scenarios, the authors estimate 0.91 Tg CO2e y−1 of potential avoided emissions, with the greatest benefits from restoring impounded palustrine wetlands to estuarine conditions. The work delivers a map of 1,796 candidate parcels—primarily within protected areas—and identifies states and management entities with disproportionate potential. Although the total mitigation is modest relative to national emissions, the identified opportunities are actionable and valuable. Future research should refine impoundment and salinity mapping across protected and non-protected lands, measure CH4 fluxes in impoundments/open waters, evaluate N2O and biomass changes, and clarify accounting for carbon burial additionality and time horizons to improve national blue carbon assessments.

- Mapping underestimation: Impoundments are under-mapped by ~50% in NWI due to optional special modifiers and detection challenges, especially for road/rail-caused impoundments and in non-protected areas; reference salinity mapping (C-CAP) may be conservative at sharp boundaries. - Survey biases: Expert surveys may underestimate salinity increases when measurements are lacking; measurement-only subset suggests slightly more salinizing restorations but with smaller sample size. - Scope and assumptions: Focused on protected areas for accuracy assessment and then scaled nationally; did not model feasibility, cost, or likelihood of restoration success; used a 100-year GWP; treated net-warming scenarios as zero (censoring) for NCS capacity. - Greenhouse gases and pools: Did not include N2O emissions, biomass stock changes, or potential additionality of carbon burial in impoundments/reservoirs; assumed tidal wetland CO2 burial rate constant and did not credit burial changes. - Emissions factors and data gaps: CH4 factors derived from literature on intact/natural wetlands may not fully represent impounded/open-water conditions; need for before-after-control-impact monitoring to refine factors. - Spatial processing assumptions: Nearest-neighbor salinity assignment, distance buffers, and parcel filtering thresholds introduce uncertainty; small-scale spatial autocorrelation handled additively in parcel error estimates.