Devendra Dahal Stephen P. Boyte Sujan Parajuli Neal J. Pastick Logan Megard Kory Postma Dinesh Shrestha 20230627 raster digital data https://doi.org/10.5066/P9351ZTZ Devendra Dahal Neal J. Pastick Stephen P. Boyte Sujan Parajuli Michael J. Oimoen Logan J. Megard 2022 publication Remote Sensing n/a n/a MDPI AG These datasets provide early estimates of 2023 fractional cover for exotic annual grass (EAG) species and one native perennial grass species on a weekly basis from May to early July. The EAG estimates are developed typically within 7-13 days of the latest satellite observation used for that version. Each weekly release contains four fractional cover maps along with their corresponding confidence maps for: 1) a group of 16 species of EAGs, 2) cheatgrass (Bromus tectorum); 3) medusahead (Taeniatherum caput-medusae); and 4) Sandberg bluegrass (Poa secunda). These datasets were generated leveraging field observations from Bureau of Land Management (BLM) Assessment, Inventory, and Monitoring (AIM) data plots; Harmonized Landsat and Sentinel-2 (HLS) based Normalized Difference Vegetation Index (NDVI); other relevant environmental, vegetation, remotely sensed, and geophysical drivers; and artificial intelligence/machine learning techniques. A total of 28,901 AIM plots from years 2016–2022 were used to train an ensemble of five-fold regression-tree models using a cross-validation approach (each observation was used as test data once and as training data four times) that developed all the fractional cover maps. The geographic coverage includes arid and semi-arid rangelands in the western U.S classified as grassland/herbaceous by the 2019 National Land Cover Database at or below 2350-m elevation. Note: Maps of May 1st, 2023 were developed using satellite observation data no later than April 28. Maps of May 8th, 2023 were developed using satellite observation data no later than May 05. Maps of May 15th, 2023 were developed using satellite observation data no later than May 11. Maps of May 22nd, 2023 were developed using satellite observation data no later than May 19. Releases: First Release: May 01, 2023 (ver. 1.0) Revised: May 08, 2023 (ver. 2.0) Revised: May 15, 2023 (ver. 3.0) Revised: May 22, 2023 (ver. 4.0) The purpose for these releases is to provide land managers and researchers with near real time estimates of spatially explicit exotic annual grasses percent cover in the study area. Appropriate use of the data should be defined by the user; however, this data comes with caveats. First, these estimates should be viewed as relative abundances. Second, comparing this dataset to similar datasets with different spatial resolutions or different dates can lead to substantial differences between dataset values. This release includes percent cover maps for exotic annual grass, cheatgrass, medusahead, and Sandberg Bluegrass as well as their associated confidence maps. The values for percent cover maps range from 0 to 100. However, values for confidence maps range from 0 to 10. A high value on the confidence map denotes high confidence with the corresponding EAG mapped pixel. 202305 202307 publication date Complete Annually western United States -124.9400 -109.0000 49.0000 31.1700 exotic species annual grass annual herbaceous cheatgrass exotic Harmonized Landsat Sentinel Texas brome medusahead NDVI noxious red brome sagebrush rye brome soft brome Poa secunda perennial grass USGS Thesaurus nonindigenous species invasive species remote sensing USGS Metadata Identifier USGS:64709b34d34e4e58932d1d82 Common geographic areas Arizona California Colorado Great Basin Idaho Kansas Montana Nebraska Nevada New Mexico North Dakota Oklahoma Oregon South Dakota Texas Utah Washington Wyoming western United States United States None. Please see 'Distribution Info' for details. None. Users are advised to read the dataset's metadata thoroughly to understand appropriate use and data limitations. Devendra Dahal (CTR) U.S. Geological Survey, LAND RESOURCES mailing address

47914 252nd Street

Sioux Falls SD 57198 US 605-594-2716 ddahal@contractor.usgs.gov Denali and Tallgrass High Performance Computing (HPC) systems with the Linux operating system were used to create the HLS NDVI data and to build the EAG and Sandberg bluegrass models and maps. Open-source geospatial software [namely Geospatial Data Abstraction Library (GDAL version 2.4.2, https://gdal.org/download.html), NumPy, and pandas with anaconda Python (version 3.6, https://www.anaconda.com/distribution/)] were used to produce these datasets. However, these datasets can be read using proprietary software such as ArcGIS and ERDAS Imagine. Files names are: ExoticAnnualGrass_2023_PercentCover.tif Cheatgrass_2023_PercentCover.tif Medusahead_2023_PercentCover.tif SandbergBluegrass_2023_PercentCover.tif ExoticAnnualGrass_2023_Confidence.tif Cheatgrass_2023_Confidence.tif Medusahead_2023_Confidence.tif SandbergBluegrass_2023_Confidence.tif Data size is 4.97 GB for all eight datafiles. Stephen P. Boyte Bruce K. Wylie 2019 dataset https://www.sciencebase.gov U.S. Geological Survey https://doi.org/10.5066/P96PVZIF Stephen P. Boyte Bruce K. Wylie 2018 dataset https://www.sciencebase.gov U.S. Geological Survey https://doi.org/10.5066/P9RIV03D Devendra Dahal Bruce K. Wylie Neal J. Pastick Sujan Parajuli 2020 dataset https://www.sciencebase.gov U.S. Geological Survey https://doi.org/10.5066/P9ZZSX5Q Devendra Dahal Neal J Pastick Sujan Parajuli Bruce K Wylie 2020 dataset https://www.sciencebase.gov U.S. Geological Survey https://doi.org/10.5066/P91NJ2PD Devendra Dahal Stephen P. Boyte Neal J. Pastick Sujan Parajuli Michael J. Oimoen Dinesh Shrestha 2022 dataset https://www.sciencebase.gov U.S. Geological Survey https://doi.org/10.5066/P9FVYOGD Below is the modelling accuracy for each mapped species: Species Training Dataset Testing Dataset MAE RAE r MAE RAE r ExoticAnnualGrass 1.63 0.3 0.93 3.27 0.57 0.74 Cheatgrass 1.36 0.3 0.93 2.69 0.58 0.73 Medusahead 0.03 0.25 0.93 0.04 0.66 0.67 SandbergBluegrass 1.02 0.32 0.92 1.98 0.59 0.71 MAE = median absolute error (MAE measures the average magnitude of the errors in a set of predictions, without considering their direction.) RAE = relative absolute error (RAE is very similar to the relative squared error; however, it takes the total absolute error and normalizes it by dividing by the total absolute error of the simple predictor) r = correlation coefficient No formal logical accuracy tests were conducted. Data set is considered complete for the information presented, as described in the abstract. Users are advised to read the rest of the metadata record carefully for additional details. A formal accuracy assessment of the horizontal positional information in the data set has not been conducted. A formal accuracy assessment of the vertical positional information in the data set has either not been conducted or is not applicable. Nathaniel W. Chaney Eric F. Wood Alexander B. McBratney Jonathan W. Hempel Travis W. Nauman Colby W. Brungard Nathan P. Odgers 201607 publication Geoderma vol. 274 n/a Elsevier BV pgs. 54-67 https://doi.org/10.1016/j.geoderma.2016.03.025 Digital and/or Hardcopy 2016 publication date POLARIS Soil Data Input variable to model Dean B. Gesch Gayla A. Evans Michael J. Oimoen Samantha Arundel 2018 tabular digital data pgs. 83–110 Digital and/or Hardcopy 2018 publication date The National Elevation Dataset Input variable to model Jon Dewitz U.S. Geological Survey 2021 dataset https://www.sciencebase.gov U.S. Geological Survey https://doi.org/10.5066/p9kzcm54 Digital and/or Hardcopy 1992 2019 ground condition National Land Cover Database (NLCD) 2019 Products Input variable to model M.M. Thornton R. Shrestha Y. Wei P.E. Thornton S-C. Kao B.E. Wilson 2022 publication https://daac.ornl.gov/cgi-bin/dsviewer.pl?ds_id=2130 ORNL Distributed Active Archive Center Thornton, M.M., R. Shrestha, Y. Wei, P.E. Thornton, S-C. Kao, and B.E. Wilson. 2022. Daymet: Annual Climate Summaries on a 1-km Grid for North America, Version 4 R1. ORNL DAAC, Oak Ridge, Tennessee, USA. https://doi.org/10.3334/ORNLDAAC/2130 https://doi.org/10.3334/ornldaac/2130 Digital and/or Hardcopy 1989 2019 publication date Climatic variables Calculated climate normals from 30 years of annual climate data and used as an input variable to model Bruce McCune Dylan Keon 20020224 publication Journal of Vegetation Science vol. 13, issue 4 n/a Wiley pgs. 603-606 https://doi.org/10.1111/j.1654-1103.2002.tb02087.x Digital and/or Hardcopy 2002 publication date potential annual direct incident radiation (PADR) Input variable to model Gordon R. Toevs Jason J. Taylor Carol S. Spurrier W. Craig MacKinnon Mathew R. Bobo 2019 tabular digital data Denver, CO. Bureau of Land Management, National Operations Center, Denver https://doimspp.sharepoint.com/sites/blm-oc/drs/SitePages/BLM Terrestrial AIM Data (TerrADat and LMF).aspx Digital and/or Hardcopy 2016 2022 observed BLM AIM Used for training models Field observations of grass fractional cover include individual species information (2016 – 2022) that were quantified by the BLM AIM project (https://aim.landscapetoolbox.org/). We compiled first hit (FH) Terrestrial AIM Database (TerrADat) and Landscape Monitoring Framework (LMF) databases of BLM AIM to aggregate exotic annual species with the annual grass cover. The AIM plots used in this study were divided into five random subsets. 20230521 Independent variables used in the modeling process include: Biophysical (i.e. soil texture fraction, available water capacity and organic matter content in the first 30 cm of the soil from POLARIS Soil Data [Chaney et al., 2016]); National Elevation Dataset [Gesch et al. 2018]; National Land Cover Database (NLCD) 2019 Shrub Component products [Dewitz, J. (2021)]; climatic variables (annual precipitation normal, annual mean temperature normal, summer precipitation normal, summer maximum temperature normal, winter precipitation normal, and winter mean temperature normal of 1985 – 2019 from Daymet [Thornton et al. 2020]); potential annual direct incident radiation (PADR) [McCune and Dylan 2002]; and phenocurves (HLS based cloud free weekly NDVI composites [Dahal et al. 2022]) were extracted at each field site. Spectral data were extracted coincident with the year of the field observations because annual grass cover can vary substantially year-to-year based on local weather and site conditions. 20230521 An ensemble of five regression models was developed and optimized by withholding one of the unique randomized subsets as test each time. The ensemble of models was constructed using scikit-learn xgboost machine learning algorithm with GridsearchCV hyperparameters optimization, and multioutput variable wrapper. The regression models were validated against the independent test samples (20% of the total samples) by calculating Pearson’s r, median absolute errors, and relative absolute errors. 20230522 The optimized models were then used to develop predictions of multiple species grass cover maps (EAG, cheatgrass, medusahead, and Sandberg Bluegrass) for each permutation. The final annual grass cover maps, and the Sandberg Bluegrass map, represent the median of the five predicted maps for each species, or collection of species in the case of EAG. The median absolute error (MAE) of the five predicted maps serves as an associated confidence map for each species. The final percent cover maps are comparable to first hit values of AIM plots. 20230522 We applied a mask to areas above 2350m elevation because the AIM data used to train the models did not include enough points above this elevation to effectively model EAG. To target likely rangeland ecosystems, the mask also covered pixels classified by the 2019 National Land Cover Dataset (NLCD) as something other than shrub or grassland/herbaceous. 20230522 Raster Grid Cell 80862 65613 1 Albers Conical Equal Area 29.5 45.5 -96.0 23.0 0.0 0.0 row and column 30.0 30.0 meters North_American_Datum_1983 GRS 1980 6378137.0 298.257222101 Exotic Annual Grasses Percent Cover Raster geospatial data file. Producer Defined OID Internal object identifier. Producer Defined Sequential unique whole numbers that are automatically generated. Value Per pixel percent cover of exotic annual grasses Producer Defined 255 Areas that are not classified as grassland/herbaceous or shrub by NLCD 2019 or with elevations above 2350 meters. Producer defined 0 100 Count Number of raster cells with this value. Producer Defined 3.0 3191319766.0 Exotic Annual Grass Confidence Raster geospatial data file. Producer Defined OID Internal object identifier. Producer Defined Sequential unique whole numbers that are automatically generated. Value Confidence on predicted exotic annual grasses percent cover Producer Defined 255 Areas that are not classified as grassland/herbaceous or shrub by NLCD 2019 or with elevations above 2350 meters. Producer defined 0 10 unitless Count Number of raster cells with this value. Producer Defined 10.0 3191319765.0 Cheatgrass Percent Cover Raster geospatial data file. Producer Defined OID Internal object identifier. Producer Defined Sequential unique whole numbers that are automatically generated. Value Per pixel percent cover of cheatgrass Producer Defined 255 Areas that are not classified as grassland/herbaceous or shrub by NLCD 2019 or with elevations above 2350 meters. Producer defined 0 100 Percentage Count Number of raster cells with this value. Producer Defined 6.0 3191319765.0 Cheatgrass Confidence Raster geospatial data file. Producer Defined OID Internal object identifier. Producer Defined Sequential unique whole numbers that are automatically generated. Value Confidence on predicted cheatgrass percent cover Producer Defined 255 Areas that are not classified as grassland/herbaceous or shrub by NLCD 2019 or with elevations above 2350 meters. Producer defined 0 10 unitless Count Number of raster cells with this value. Producer Defined 12.0 3191319765.0 Medusahead Percent Cover Raster geospatial data file. Producer Defined OID Internal object identifier. Producer Defined Sequential unique whole numbers that are automatically generated. Value Per pixel percent cover of medusahead Producer Defined 255 Areas that are not classified as grassland/herbaceous or shrub by NLCD 2019 or with elevations above 2350 meters. Producer defined 0 100 Percentage Count Number of raster cells with this value. Producer Defined 1.0 3191319765.0 Medusahead Confidence Raster geospatial data file. Producer Defined OID Internal object identifier. Producer Defined Sequential unique whole numbers that are automatically generated. Value Confidence on predicted medusahead percent cover Producer Defined 255 Areas that are not classified as grassland/herbaceous or shrub by NLCD 2019 or with elevations above 2350 meters. Producer defined 0 10 unitless Count Number of raster cells with this value. Producer Defined 1.0 3191319765.0 Sandberg Bluegrass Percent Cover Raster geospatial data file. Producer Defined OID Internal object identifier. Producer Defined Sequential unique whole numbers that are automatically generated. Value Per pixel percent cover of Sandberg Bluegrass Producer Defined 255 Areas that are not classified as grassland/herbaceous or shrub by NLCD 2019 or with elevations above 2350 meters. Producer defined 0 100 Percentage Count Number of raster cells with this value. Producer Defined 7.0 3191319765.0 Sandberg Bluegrass Confidence Raster geospatial data file. Producer Defined OID Internal object identifier. Producer Defined Sequential unique whole numbers that are automatically generated. Value Confidence on predicted Sandberg Bluegrass percent cover Producer Defined 255 Areas that are not classified as grassland/herbaceous or shrub by NLCD 2019 or with elevations above 2350 meters. Producer defined 0 10 Confidence on predicted Sandberg Bluegrass percent cover Count Number of raster cells with this value. Producer Defined 15.0 3191319765.0 This metadata represents datasets for all listed species, i.e., exotic annual grass, cheatgrass, medusahead, and Sandberg Bluegrass. The datasets provide weekly percent cover early estimates for 2023 generated from 1st week of May to 1st week of July 2023. Devendra Dahal, Stephen P. Boyte, Sujan Parajuli, Neal J. Pastick, Logan Megard, Kory Postma, and Dinesh Shrestha, 20230522, Early Estimates of Exotic Annual Grass (EAG) in the Sagebrush Biome, USA, 2023: U.S. Geological Survey data release, https://doi.org/10.5066/P9351ZTZ. GS ScienceBase U.S. Geological Survey mailing address

Denver Federal Center, Building 810, Mail Stop 302

Denver CO 80225 United States 1-888-275-8747 sciencebase@usgs.gov Unless otherwise stated, all data, metadata and related materials are considered to satisfy the quality standards relative to the purpose for which the data were collected. Although these data and associated metadata have been reviewed for accuracy and completeness and approved for release by the U.S. Geological Survey (USGS), no warranty expressed or implied is made regarding the display or utility of the data for other purposes, nor on all computer systems, nor shall the act of distribution constitute any such warranty. Digital Data https://doi.org/10.5066/P9351ZTZ None 20230627 Dinesh (Contractor) Shrestha U.S. Geological Survey, CORE SCIENCE SYSTEMS mailing address

47914 252Nd Street

Sioux Falls SD 57198 US 605-594-6028 dshrestha@contractor.usgs.gov FGDC Biological Data Profile of the Content Standard for Digital Geospatial Metadata FGDC-STD-001.1-1999