Mapping forest-based natural climate solutions

Ecoregion boundary

The coastal temperate rainforests of western North America are defined by a cool and wet maritime climate. Subnational ecoregion datasets were compiled to define the most accurate boundaries. In Alaska, USA, the extent was defined by Nowacki et al. 2001⁵⁵ Level 2 Coastal Rainforests. In British Columbia, Canada, the extent was defined by the Cool Hypermaritime and Highlands Ecodivision⁵⁶. The Washington and Oregon extent was defined by combining ecoregional data with fire regime groups⁵⁷. Fire regime groups were used to identify areas least susceptible to frequent, high-severity fires due to climate change. We included fire regime groups with long fire return intervals as well as those with shorter return intervals but with low-mixed severity. Specifically, we included five ecoregions across Oregon and Washington—Pacific Northwest Coast, Willamette Valley / Puget Trough, West Cascades, North Cascades and the portions of the East Cascades and Okanagan Ecoregions with a high density of fire regime groups V-A, V- B, I-C and III-B. The Washington and Oregon extent was further refined by combining data on forest productivity^58,59, and Biophysical Settings (BPS) / Potential Vegetation Types (PVT)⁶⁰ to identify areas that have high carbon potential but which may currently be in agricultural or other human modified land uses. We resampled the forest productivity data, which provides estimates of bolewood productivity (gC/m²/yr) based on climate data and site index, from 800 m to 30 m and calculated the mean productivity for each BPS/PVT type across the region^58,59. Within the area defined by ecoregions and fire regime group, we retained BPS/PVT cells with a mean productivity GTE 175 gC/m²/yr as well as any cells where forest productivity was within the upper 50% of productivity values (>136 C/m²/yr). These productivity/BPS/PVT data were overlaid with NRCS 6th level (12-digit) Hydrologic units and any watersheds overlapping the retained productivity/BPS/PVT cells were selected to form the final Washington and Oregon extent boundaries.

Spatial data hierarchy

A spatial data hierarchy was developed to consistently compare opportunities for forest carbon conservation with a jurisdictionally nested database structure by country, state/province, landowner type, and land management designations that define the scope of NCS actions available in a given project location. Recent and publicly available subnational land ownership and land designation spatial datasets were synthesized starting with the most authoritative government datasets that are publicly available online. These data were supplemented, verified, and augmented with finer scale datasets obtained directly from local agencies. In cases where spatial data were not readily available, as was the case with some administratively protected areas (e.g., riparian buffers), the regulations were mapped using GIS analysis.

We developed a classification framework based on existing land use designations for assigning categories of action related to implementation of forest-based NCS (Table 1; Fig. 4). Areas available for NCS action were classified as NCS Action Category 1 (NCS1). In general, NCS1 areas are open to timber harvesting or other extractive activities where conservation and management actions related to enhanced carbon storage or sequestration are additional to the status quo. Examples include private forest lands, public lands held under forest management licenses to corporations, State Forests, National Forests, or woodlots. This category also includes Indigenous reserves, settlement areas for Indigenous land claims, or private lands held by Indigenous communities. NCS Action Category 2 (NCS2; Not currently available for action) include lands with administrative but not necessarily permanent protections that preclude additional measures for carbon conservation, such as riparian buffers in public lands that exclude timber harvest, special management zones such as wildlife reserves, and recreation areas. NCS Action Category 3 (NCS3; Not available for action) include permanently protected areas, such as National Parks, Wilderness Areas, and private lands protected by non-governmental organizations or land trusts. Each of these is derived from publicly available data so that potential users can further refine NCS actions to the specific circumstances and management constrains relevant to each designation. A detailed table of NCS Categories and land use designations can be found in Supplementary Table 2.

The generalizable framework developed to identify areas available for Natural Climate Solution (NCS) action and quantify the average annual flux and standing forest carbon stocks in the coastal temperate rainforests of western North America.

Alaska, USA- The primary ownership dataset used for Alaska was the Bureau of Land Management (BLM) National Surface Management Agency Area Polygons – National Geospatial Data Asset (NGDA)⁶¹. The areas available for NCS action (NCS1) were defined as privately owned (e.g., Alaska Native Corporation lands), have a timber management program, or are a local municipality. The Tongass National Forest harvestable land base (NCS1) was defined by the suitable and available stands dataset⁶². The Haines State Forest harvestable land base (NCS1) was defined by the commercially viable vegetation polygons⁶³. The Southeast Alaska State Forest (NCS1) was defined by the Alaska State Forest boundaries⁶⁴. Similarly, the University of Alaska Land Grant Trust and Alaska Mental Health Trust Authority investment lands (NCS1) were defined with their parcel database⁶⁵. Administratively protected harvest exclusions (NCS2) on non-Federal lands were approximated with a 30-m buffer on all perennial streams, lakes, and coastlines using the National Hydrography Dataset (NHD)⁶⁶ in accordance with the Alaska Forest Resources and Practice Act⁶⁷. Forest carbon projects (NCS2) on the American Carbon Registry, Climate Action Reserve, and the University of Alaska Land Management Carbon Credit Program⁶⁸ were also excluded from potentially harvestable lands. Lastly, we mapped permanently protected areas (NCS3) with the Protected Areas Database (PAD-US) 2.1 Vector Analysis File⁶⁹ GAP 1 lands; GAP 2 lands were classified as administratively protected (NCS2). Lands with an unclear forest management plan (e.g., NGDA “Alaska-Other” designation) were designated as unknown.

British Columbia, Canada- The primary ownership dataset used for British Columbia was the Generalized Forest Cover Ownership⁷⁰. This dataset was supplemented with Aboriginal Lands of Canada Legislative Boundaries⁷¹, Tree Farm Licenses⁷², and Timber Supply Areas⁷³. The areas available for NCS action (NCS1) were defined by private ownership or by Crown Tenure—Forest Management Unit, Timber License, Woodlot License Schedule A, Timber License in a TFL, Christmas Tree License and Crown Lease – Miscellaneous Leases. Indian Reserves, Land Claim Settlement Areas and Municipal Parcels were also classed as NCS1. Riparian harvest exclusions (NCS2) were approximated with a 30-m buffer on watercourse hydrographic features using the Topographic Data of Canada—CanVec Series⁷⁴, freshwater lakes⁷⁵, and coasts⁷⁰ in accordance with the Fish Protection Act⁷⁶. Additional administratively protected areas (NCS2) included Forest Recreation Reserves, Wildlife Management Areas, Miliary Reserves, Recreation Areas, Federal Reserves, and Heritage Sites. NCS2 areas also included all non-private land inside the Great Bear Rainforest agreement area. Permanently protected areas (NCS3) were defined by the Canadian Protected and Conserved Areas Database IUCN Categories I-III⁴⁶.

Washington, USA- A suite of ownership datasets was combined for Washington. These included the NGDA⁶¹, PAD-US⁴⁷, Northwest Forest Plan (NWFP)⁷⁷, and county tax parcels⁷⁸. The areas available for NCS action (NCS1) were defined by private ownership, Indigenous treaty lands and tribal ownership, NGO, or NWFP Primary Land Use Allocations (LUAs): Adaptive Management Areas, No NWFP Designation, and Others (Supplementary Table 2). From the NWFP, administratively protected areas (NCS2) were defined by the LUAs: Administratively withdrawn, Adaptive management areas in reserves, Late-successional Reserve, Late-successional reserves for marbled murrelet and the northern spotted owl, Managed late-successional reserves. Congressionally Reserved areas in the NWFP were mapped as permanently protected (NCS3). The PAD-US GAP 2 (NCS2) was mapped as administratively protected and GAP 1 (NCS3) as permanently protected. From each data source, we prioritized the most limiting land use category. Lastly, we mapped permanent protections (NCS3) on riparian zones and wetlands with a 30-m buffer around all fish-bearing streams and state shorelines and a 15-m buffer around all other streams based on stream typing for fish-bearing designations. Stream, wetland, and waterbody data sources include Washington watercourses and waterbodies⁷⁹, salmon fish distributions⁸⁰, and the National Wetlands Inventory⁸¹. We designed these buffer widths as simplified representations of those described for the Washington Forest Practice Rules⁸².

Oregon, USA- The primary ownership dataset used on Oregon federal lands was the BLM Oregon Management Ownership Dissolve Polygon⁶¹. State managed lands were delineated with Oregon Department of Forestry (ODF) ownership land management data⁸³. These datasets were augmented with the USFS Region 6 boundaries⁸⁴. Private and tribal land boundaries were compiled directly from tax lot information ParcelPoint Tax Lot Database⁸⁵, Regrid⁸⁶ and the U.S. Census Tigerlines for U.S. American Indian, Alaska Native, and Native Hawaiian Areas⁸⁷. Management status of lands in Oregon were determined from the PAD-US⁴⁷ and Northwest Forest Plan LUAs⁷⁷. The areas available for NCS action (NCS1) were defined by private ownership, NWFP LUAs, BLM multiple-use public land, USFS multiple-use public land, and Indigenous lands (Supplementary Table 2). Additional administratively protected areas (NCS2) included BLM wilderness areas, FWS study areas, NWFP LUAs, and USFS reserves. Permanently protected areas (NCS3) were defined by US-PAD GAP 1 lands and riparian buffers on the NHD⁶⁶ defined by the Oregon Forest Practices Act Private Forest Accord⁸⁸.

Forest carbon and flux models

The best available high-resolution aboveground forest carbon stock and flux (emissions and sequestration) maps at global, national, and regional scales were analyzed to compare the resulting patterns of climate mitigation benefits from improved forest management and conservation. To start, a forest vegetation mask was generated from the 30-m North America Land Change Monitoring System (NALCMS) Classes 1–10 for the year 2015⁸⁹. These data were also resampled to a 90-m resolution using the nearest neighbor technique to match the global forest carbon model. This mask included forest, shrubland, and grassland classes because recently harvested clearcuts important to our analysis were classified as shrubland and grassland. The excluded classes included: snow and ice, water, urban, barren islands, cropland, wetland, and sub-polar and polar lichen-moss classes.

The global AGB model for 2018 v3^15,90 was downloaded in WGS84 tiles of average Mg ha⁻¹, mosaiced, projected to BC Albers with a 90-m cell using bilinear interpolation, and masked/snapped to the NALCMS mask. The AGB Mg ha⁻¹ were converted into AGB Mg/pixel by multiplying by 0.81 corresponding with the cell area (8,100 m²). The AGB Mg/pixel were converted to MgC/pixel by multiplying by 0.47. Lastly, the MgC/pixel was converted to MgCO₂e/pixel by multiplying by 3.67 for summing totals by ownership polygons. The corresponding AGB Standard Error (SE) rasters^15,90 were processed the same as the AGB data. To summarize carbon stocks at the level of State/Provincial ownership types, the MgCO₂e SE was summed and then squared by individual parcels before taking the square root of the resulting totals for each subnations ownership type categories. To obtain 95% confidence intervals around the mean, the SE was multiplied by 1.96 on either side of the mean.

The Canada⁴⁰, U.S. Pacific Northwest^42,91, and contiguous U.S.⁴¹ AGB models in average Mg ha⁻¹ were downloaded with 30-m cells that were projected into BC Albers and masked/snapped to the NALCMS mask. Data for AGB Mg ha⁻¹ were converted into AGB Mg/pixel by multiplying by 0.09 corresponding with pixel area (900 m²). The AGB Mg/pixel were converted to MgC/pixel by multiplying by 0.47. Lastly, the MgC/pixel was converted to MgCO₂e/pixel by multiplying by 3.67 and used to sum total biomass by ownership polygons. For calculating the forested area (ha) available for NCS action and average MgCO₂e ha⁻¹, the spatial hierarchy ownership polygons were clipped to the 30-m NALCMS mask.

We estimated forest carbon sequestration and losses for 2001–2021 based on a global map of carbon flux⁹. This map includes emissions from all biological carbon pools for stand-replacing disturbances and removals into aboveground and belowground biomass by standing and new forest. It spatializes estimates of flux based on Intergovernmental Panel on Climate Change national greenhouse gas inventory rules by overlaying forest extent, loss and gain maps with emission and removal factors from the literature. Sequestration³² and emission³³ rasters were downloaded in WGS84 tiles of average MgCO₂e pixel ⁻¹, mosaiced, and left unprojected for calculating zonal statistics. The feature class data was unprojected into WGS84 for calculating zonal statistics. For the purposes of this analysis, the gross forest greenhouse gas emissions and gross carbon loss data were used. The MgCO₂e/pixel was converted to the average MgCO₂e pixel ⁻¹ yr⁻¹ by dividing the 21-year time period for summing totals by ownership polygons. These historical fluxes are assumed to apply to future trends, and we assume that historical patterns and causes of disturbances will continue through 2030. Uncertainty estimates for these flux estimates are not available for this data at this level of analysis.

2030 natural climate solutions opportunity

To demonstrate the scale of potential NCS opportunity from an increase in improved forest management and conservation⁷ in the coastal temperate rainforests of western North America, we compared the historical range of variability in timber harvest³ (Supplementary Table 4) to the 175 MtCO₂e yr⁻¹ 2030 land-based climate commitments made by the United States and Canada^30,31. The 10% coefficient of variation (CV) in historical timber harvest (harvested roundwood only represents a portion of the emissions detected with the flux model used in this study⁹) was then used as a conservative emissions reduction target based on the average annual carbon emissions for the study region (91 MtCO₂e yr^–1).