U.S.-EPA-BELD4-Equivalent Landuse Database for Canada – Version 1

• Main Authors: Zhang, Junhua, Moran, Michael D.
• Format: Dataset
• Language: English
• Published: Zenodo 2018
• Subjects: BELD4; Biogenic; Emissions; Landuse; Canada; Arctic; British Columbia; Chapel Hill; Climate change; Tundra
• Online Access: https://dx.doi.org/10.5281/zenodo.2231047; https://zenodo.org/record/2231047

Air Quality Research Division, Environment and Climate Change Canada, 4905 Dufferin Street, Toronto, Ontario, M3H 5T4, Canada Email: [e-mail] Volatile organic compounds (VOCs) released from terrestrial vegetation are an important source of VOC emissions to the atmosphere. Globally, natural sources of VOC emissions have been estimated to be much larger than anthropogenic sources and to account for 80-90% of total global VOC emissions. The accurate estimation of biogenic VOC emissions requires input of landuse data with a detailed description of vegetation types at high resolution. The Biogenic Emissions Landuse Database, Version 3 (BELD3), which contains 230 vegetation classes at 1-km resolution, was compiled by the U.S. Environmental Protection Agency (EPA) for most of North America (Pierce et al., 2000) and has been widely used by many regional air quality models for the last two decades to estimate biogenic emissions. However, some of the BELD3 vegetation fields were based on satellite imagery from the early 1990s and are now outdated for many parts of North America, particularly for areas undergoing rapid, large-scale development. In addition, since the release of the BELD3 database a few issues have been identified for the Canadian part of the BELD3 database, such as (1) detailed crop types not being available for some Canadian provinces, (2) discontinuities at international and provincial borders and discontinuities within Canadian provinces, (3) unrealistically large coverage of some tree species for some areas, such as balsam fir and black spruce for the province of British Columbia, and (4) a large and questionable fraction of unknown tree species in eastern Canada that will have zero emissions. Recently, the U.S. EPA updated BELD from Version 3 to Version 4 (BELD4), with 286 landuse categories at 1-km resolution (see https://www.epa.gov/air-emissions-modeling/biogenic-emission-sources). However, these updates were mostly made for the contiguous United States. For Canada, only 20 broad landuse types based on MODIS satellite retrievals are contained in BELD4, which degrades the representation of landuse for Canada from that in the BELD3 database. In order to address the issues with the Canadian landuse data in BELD3 noted above and to improve the representation of Canadian landuse categories in the current U.S. EPA BELD4 database, a U.S.-equivalent BELD4 dataset has been extended to Canada. This extended BELD4 dataset is based on a recent version of the Canadian National Forest Inventory (NFI), with fractional coverage of 109 tree species fields gridded at 250-meter resolution (see Beaudoin et al., 2014 and https://nfi.nfis.org/en/) and on the 2016 Canadian Annual Crop Inventory (ACI), with 62 vegetation and other landuse fields gridded by dominant category at 30-meter resolution (see http://www.agr.gc.ca/atlas/aci). Two major steps were required to transform these two land-cover inventories into a BELD4-compatible format and then combine them. The first major step was to build a metadata or schema crosswalk. In order to harmonize with the vegetation and other landuse categories in the EPA BELD4 database, the 109 NFI and 62 ACI species and other landuse categories were mapped to 87 U.S. BELD4 landuse categories as shown in spreadsheet “ACI_NFI_BELD4_species_match_final.xlsx”. This step facilitated the direct usage of BELD4-tailored biogenic emissions factors in the recent version of Biogenic Emission Inventory System (BEIS) for estimating VOC emissions from vegetation and soil nitric oxide (NO) emissions (https://www.epa.gov/air-emissions-modeling/biogenic-emission-inventory-system-beis). In a few cases the NFI or ACI provided more specific species categories relative to a more generic BELD4 category; to address these cases, the multiple NFI or ACI categories were summed and assigned to the single BELD4 category (e.g., Yellow birch, White Birch, and Gray birch (wire birch) in NFI vs. Birch in BELD4). In a few other cases, a single NFI or ACI category could be matched with multiple BELD4 categories; to address these cases, the NFI or ACI category was assigned to the BELD4 category that provided the best fit (e.g., Corn in ACI vs. CornGrain, CornGrain_ir, CornSilage, and CornSilage_ir in BELD4). The second major step was to harmonize grid resolutions by aggregating and resampling the Canadian NFI and ACI gridded data fields from their native spatial resolution to the 1-km BELD4 resolution. This step was straightforward for the NFI, since the 1-km BELD4 grid is an integer multiple of the NFI 0.25-km grid. For the ACI, however, it was first necessary to aggregate from the native 30-meter resolution to 990-meter resolution and then to resample to the 1-km BELD4 grid using the “nearest neighbour” option. After the ACI and NFI fields were remapped to the 1-km grid, all-category sums of the 1-km gridded ACI and NFI data were larger than unity for some grid points because the ACI and NFI databases were compiled independently, which results in “double-counting” of the ACI and NFI fields for some areas. To avoid such double-counting, the combined regridded NFI and ACI fields were renormalized for the grid points for which the total fraction of the ACI and NFI fields was larger than unity. For the rest of the grid points, no change was made to their ACI and NFI fractions, but any grid points whose total fractional coverage was less than unity were gap filled to unity using MODIS land-cover data as described below. The Canadian ACI data set has the significant limitation that it only covers southern Canada while the NFI data set only considers forested areas. One BELD4 land use category, tundra, which is important for northern Canada, particularly for the Arctic region, is not available in either the NFI or ACI. To address these limitations, the Land Cover Classification System surface hydrology (LCCS3) dataset contained in the Collection 6 MODIS Land Cover product MCD12Q1 (https://lpdaac.usgs.gov/sites/default/files/public/product_documentation/mcd12_user_guide_v6.pdf) was used for filling the gaps, particularly in northern Canada. The crosswalk used to link the LCCS3 categories to the BELD4 categories is described in the attached Excel file “MODIS_LCCS3_BELD4_mapping.xlsx”. With the use of the LCCS3 dataset, five of the matched ACI landuse fields (Barren Land (Rock/Sand/Clay), Open Water, Woody Wetlands, Grassland/Herbaceous, and Shrub/Scrub) were extended to the Canadian Arctic and an additional five BELD4 landuse categories, including tundra (plus Perennial Ice/Snow, Moss, Lichens, and Emergent Herbaceous Wetlands), were added. In total, gridded fractional coverage fields for 92 U.S.-EPA-BELD4-equivalent landuse categories have been compiled for Canada. Plots of the 92 matched Canadian and U.S. BELD4 vegetation species and other landuse categories are shown in the accompanying file “CAN_US_Matched_BELD4_Species_Plots.pdf”. Note that 24 Canadian landuse categories, including 10 MODIS LCCS3 categories, are mapped to 15 BELD4 NLCD (U.S. National Land Cover Database) categories. This does not mean that the NLCD covers Canada, but rather this was done just for the purpose of using proper BEIS emission factors for those land-cover types. Overall, it can be seen from this set of plots that the extended BELD4 database provides a seamless extension of the U.S. BELD4 fields across the international border (e.g., see Douglas fir and Wheat Spring plots). Most of the BELD3 issues for Canada noted above have also been addressed, including the high fraction of the “unknown” vegetation category in eastern Canada. Additional analyses are provided in a recent conference presentation (Zhang et. al., 2018, https://www.cmascenter.org/conference//2018/slides/zhang_extension_version4_2018.pdf; also provided in this package as file “Canadian_BELD4_Poster_CMAS_2018_22Oct2018_final.pdf”) The updated Canadian BELD4 database in GeoTIFF format at 1-km resolution with Lambert conformal conic projection (+proj=lcc +lat_1=49 +lat_2=77 +lat_0=0 +lon_0=-95 +x_0=0 +y_0=0 +ellps=GRS80 +units=m +no_defs) is available in zip file “CAN-BELD4_tif.7z”. Note that to use this dataset, all MODIS landuse categories in the original EPA BELD4 database must be removed for Canada to avoid double-counting. REFERENCES Beaudoin, A., Bernier, P.Y., Guindon, L., Villemaire, P., Guo, X.J., Stinson, G., Bergeron, T., Magnussen, S., and Hall, R.J.: Mapping attributes of Canada’s forests at moderate resolution through kNN and MODIS imagery. Can. J. For. Res. , 44 , 521–532, http://dx.doi.org/10.1139/cjfr-2013-0401, 2014. Pierce Jr., T.E., Kinnee, E.J., and Geron, C.D.: Development of a 1-km vegetation database for modeling biogenic fluxes of hydrocarbons and nitric oxide. Sixth International Conference on Air Surface Exchange of Gases and Particles , July 3-7, Edinburgh, Scotland, https://www.epa.gov/sites/production/files/2015-08/beld3_web.ppsx, 2000. Zhang, J., Moran, M.D., and He, Z.: Extension of Version 4 of the Biogenic Emissions Landuse Database (BELD4) to Canada, 17 th Annual CMAS Conference , Oct. 22-12, Chapel Hill, North Carolina, USA, https://www.cmascenter.org/conference//2018/slides/zhang_extension_version4_2018.pdf, 2018. ACKNOWLEDGEMENTS We are very grateful for the dedicated assistance of Dr. Zhuanshi He of SOLANA Networks Inc. in preparing this dataset. Help from Ms. Lee Benson with the crosswalk between the BELD4 vegetation species and other landuse fields and the Canadian ACI/NFI species fields is also much appreciated. RELATED DATA SETS AND MATERIALS CAN-BELD4_tif.7z – New extended BELD4 GeoTIFF file “ACI_NFI_BELD4_species_match_final.xlsx” – NFI/ACI-BELD4 landuse category crosswalk file “MODIS_LCCS3_BELD4_mapping.xlsx” – MODIS-BELD4 landuse category crosswalk file “CAN_US_Matched_BELD4_Species_Plots.pdf” – plots of 92 BELD4 category fields “Canadian_BELD4_Poster_CMAS_2018_22Oct2018_final.pdf” – conference poster on this project