These data provide current and future projected post-fire invasion risk by non-native short-lived grasses and forbs based on vegetation cover data from 26,729 plots in the western United States that burned prior to being sampled. Projected post-fire invasion risk was calculated using random forests with gridded climate, soil, and topographic predictor variables following Prevéy et al. (2024). Projections cover the western United States west of -100 longitude over the current time period (1981-2010), mid-century (2041–2070) and the end of the century (2071–2100) under medium (SSP245) and high (SSP585) greenhouse gas emission scenarios for non-native C3 short-lived grasses and non-native short-lived forbs. Each raster file represents the projected post-fire invasion risk for each non-native functional group ('sl_grass' = short-lived C3 grass, 'sl_forb' = short-lived forb), then the time period (current = 1981-2010, mid = 2041-2071, and late = 2071-2100), and lastly, the emissions scenario (none for current, '245' for SSP245, and '585' for SSP585). For example, 'sl_grass_mid_245.tif' is a raster file showing projected post-fire invasion risk for non-native short-lived C3 grasses for mid-century (2041-2070) under the SSP245 emissions scenario.

These data consist of plot-level plant species cover measurements from numerous targeted post-fire vegetation studies across the western United States. This data release includes two data tables. The first data table: 'postfire_vegplot_dataset.csv', consists of absolute percent live foliar cover measurements of all plant species within plots from targeted post-fire vegetation studies, or 'datasets', across the western United States. The second data table: 'dataset_information_table.csv', lists any citations, links to related publications, or other notes about data collection for specific datasets.

Exotic annual grasses are one of the most damaging biological stressors in western North America and increase the susceptibility of landscapes to wildfire occurrence. Here we couple estimates of long-term rangeland component fractions (e.g. exotic annual grasses) with remote sensing, climate data, and machine learning techniques to estimate the long-term (1985 to 2019) probability of wildfire occurrence (30-m spatial resolution) in sagebrush-dominated landscapes of the western United States.

Post-fire vegetation status and condition have multiple implications. They are indicative of burn severity and the lasting impacts of fire the land; they also help inform post-fire debris flow modeling and related risk analyses, hydrology and water quality assessments, and vulnerability to invasive species. Monitoring vegetation recovery over time enables continuous re-evaluation of various post-fire hazards, thereby facilitating informed and timely responses to post-fire risks by land managers at the local level. Structure metrics were derived from spaceborne Global Ecosystem Dynamics Investigation (GEDI) lidar data and used to map pre- and post-fire structure. Pre- and post-fire Landsat or Sentinel satellite data were obtained from the Monitoring Trends in Burn Severity (MTBS; https://www.mtbs.gov/) program. GEDI data were intersected with each satellite band and XGBoost models were built using band values as independent variables and GEDI vegetation structure values as dependent values. The models were used to generate spatially continuous maps of structure, providing vegetation structural estimates throughout the fire perimeter and beyond.

Post-fire vegetation status and condition have multiple implications. They are indicative of burn severity and the lasting impacts of fire the land; they also help inform post-fire debris flow modeling and related risk analyses, hydrology and water quality assessments, and vulnerability to invasive species. Monitoring vegetation recovery over time enables continuous re-evaluation of various post-fire hazards, thereby facilitating informed and timely responses to post-fire risks by land managers at the local level. Structure metrics were derived from spaceborne Global Ecosystem Dynamics Investigation (GEDI) lidar data and used to map pre- and post-fire structure. Pre- and post-fire Landsat or Sentinel satellite data were obtained from the Monitoring Trends in Burn Severity (MTBS; https://www.mtbs.gov/) program. GEDI data were intersected with each satellite band and XGBoost models were built using band values as independent variables and GEDI vegetation structure values as dependent values. The models were used to generate spatially continuous maps of structure, providing vegetation structural estimates throughout the fire perimeter and beyond.

Post-fire vegetation status and condition have multiple implications. They are indicative of burn severity and the lasting impacts of fire the land; they also help inform post-fire debris flow modeling and related risk analyses, hydrology and water quality assessments, and vulnerability to invasive species. Monitoring vegetation recovery over time enables continuous re-evaluation of various post-fire hazards, thereby facilitating informed and timely responses to post-fire risks by land managers at the local level. Structure metrics were derived from spaceborne Global Ecosystem Dynamics Investigation (GEDI) lidar data and used to map pre- and post-fire structure. Pre- and post-fire Landsat or Sentinel satellite data were obtained from the Monitoring Trends in Burn Severity (MTBS; https://www.mtbs.gov/) program. GEDI data were intersected with each satellite band and XGBoost models were built using band values as independent variables and GEDI vegetation structure values as dependent values. The models were used to generate spatially continuous maps of structure, providing vegetation structural estimates throughout the fire perimeter and beyond.

Post-fire vegetation status and condition have multiple implications. They are indicative of burn severity and the lasting impacts of fire the land; they also help inform post-fire debris flow modeling and related risk analyses, hydrology and water quality assessments, and vulnerability to invasive species. Monitoring vegetation recovery over time enables continuous re-evaluation of various post-fire hazards, thereby facilitating informed and timely responses to post-fire risks by land managers at the local level. Structure metrics were derived from spaceborne Global Ecosystem Dynamics Investigation (GEDI) lidar data and used to map pre- and post-fire structure. Pre- and post-fire Landsat or Sentinel satellite data were obtained from the Monitoring Trends in Burn Severity (MTBS; https://www.mtbs.gov/) program. GEDI data were intersected with each satellite band and XGBoost models were built using band values as independent variables and GEDI vegetation structure values as dependent values. The models were used to generate spatially continuous maps of structure, providing vegetation structural estimates throughout the fire perimeter and beyond.

Post-fire vegetation status and condition have multiple implications. They are indicative of burn severity and the lasting impacts of fire the land; they also help inform post-fire debris flow modeling and related risk analyses, hydrology and water quality assessments, and vulnerability to invasive species. Monitoring vegetation recovery over time enables continuous re-evaluation of various post-fire hazards, thereby facilitating informed and timely responses to post-fire risks by land managers at the local level. Structure metrics were derived from spaceborne Global Ecosystem Dynamics Investigation (GEDI) lidar data and used to map pre- and post-fire structure. Pre- and post-fire Landsat or Sentinel satellite data were obtained from the Monitoring Trends in Burn Severity (MTBS; https://www.mtbs.gov/) program. GEDI data were intersected with each satellite band and XGBoost models were built using band values as independent variables and GEDI vegetation structure values as dependent values. The models were used to generate spatially continuous maps of structure, providing vegetation structural estimates throughout the fire perimeter and beyond.

The Existing Vegetation Cover (EVC) product depicts percent canopy cover by life form and is an important input to other LANDFIRE mapping efforts. EVC is generated separately for tree, shrub and herbaceous life forms using training data and a series of geospatial predictor layers. Plots from the Forest Inventory and Analysis (FIA) program of USDA Forest Service (https://www.fia.fs.usda.gov/) were used as the training data for tree canopy cover mapping, with canopy cover of the plots estimated from stem-mapped tree data and calibrated with line intercept field measurements of canopy cover (Toney and others 2009). Shrub and herbaceous canopy cover training data were also derived from plot-level, ground-based visual assessments. More information regarding contributors of field plot data can be found at http://www.landfire.gov/participate_acknowledgements.php. Regression tree models were developed separately for each life form using the training data and a combination of multitemporal Landsat data, terrain data from a digital elevation model, and biophysical gradient data layers. Cubist software was used for modeling. The derived regression tree equations were then applied to the geospatial predictor data to create 30-m resolution, life form specific data layers (i.e., separate data layers are generated for tree, shrub and herbaceous vegetation cover). Each of the derived data layers (tree, shrub, herbaceous) has a potential range of 0-100 percent canopy cover. Tree, shrub and herbaceous values were binned into discrete classes (up to 10 bins at 10 percent intervals for tree, shrub and herbaceous canopy cover). The final EVC layer was evaluated and rectified through a series of QA/QC measures to ensure that the life form of the canopy cover code matched the life form of the LANDFIRE Existing Vegetation Type (EVT) layer. EVC is used in the development of subsequent LANDFIRE data layers. LF 2014 (lf_1.4.0) used modified LF 2010 (lf_1.2.0) data as a launching point to incorporate disturbance and its severity, both managed and natural, which occurred on the landscape 2013 and 2014. Specific examples of disturbance are: fire, vegetation management, weather, and insect and disease. The final disturbance data used in LANDFIRE is the result of several efforts that include data derived in part from remotely sensed land change methods, Monitoring Trends in Burn Severity (MTBS), and the LANDFIRE Events data call. Vegetation growth was modeled where both disturbance and non-disturbance occurs. Urban, agriculture, and wetlands were refined to reflect a 2012 landscape using the National Conservation Easement Database, National Wetlands Inventory (NWI), and Common Land Unit database (CLU) data.

The Existing Vegetation Cover (EVC) product depicts percent canopy cover by life form and is an important input to other LANDFIRE mapping efforts. EVC is generated separately for tree, shrub and herbaceous life forms using training data and a series of geospatial predictor layers. Plots from the Forest Inventory and Analysis (FIA) program of USDA Forest Service (https://www.fia.fs.usda.gov/) were used as the training data for tree canopy cover mapping, with canopy cover of the plots estimated from stem-mapped tree data and calibrated with line intercept field measurements of canopy cover (Toney and others 2009). Shrub and herbaceous canopy cover training data were also derived from plot-level, ground-based visual assessments. More information regarding contributors of field plot data can be found at http://www.landfire.gov/participate_acknowledgements.php. Regression tree models were developed separately for each life form using the training data and a combination of multitemporal Landsat data, terrain data from a digital elevation model, and biophysical gradient data layers. Cubist software was used for modeling. The derived regression tree equations were then applied to the geospatial predictor data to create 30-m resolution, life form specific data layers (i.e., separate data layers are generated for tree, shrub and herbaceous vegetation cover). Each of the derived data layers (tree, shrub, herbaceous) has a potential range of 0-100 percent canopy cover. Tree, shrub and herbaceous values were binned into discrete classes (up to 10 bins at 10 percent intervals for tree, shrub and herbaceous canopy cover). The final EVC layer was evaluated and rectified through a series of QA/QC measures to ensure that the life form of the canopy cover code matched the life form of the LANDFIRE Existing Vegetation Type (EVT) layer. EVC is used in the development of subsequent LANDFIRE data layers. LF 2014 (lf_1.4.0) used modified LF 2010 (lf_1.2.0) data as a launching point to incorporate disturbance and its severity, both managed and natural, which occurred on the landscape 2013 and 2014. Specific examples of disturbance are: fire, vegetation management, weather, and insect and disease. The final disturbance data used in LANDFIRE is the result of several efforts that include data derived in part from remotely sensed land change methods, Monitoring Trends in Burn Severity (MTBS), and the LANDFIRE Events data call. Vegetation growth was modeled where both disturbance and non-disturbance occurs. Urban, agriculture, and wetlands were refined to reflect a 2012 landscape using the National Conservation Easement Database, National Wetlands Inventory (NWI), and Common Land Unit database (CLU) data.