Get reliable science

We developed a time series (2000 � 2016) of spatially explicit herbaceous annual maps in 1% increments that emphasize annual grass cover in the sagebrush ecosystem. The study area includes the Great Basin, the Snake River Plain, the state of Wyoming, and contiguous areas. We integrated remotely sensed data (250-m eMODIS NDVI) with land cover (National Land Cover Database), biogeophysical (e.g. soils, topography), and 30-year precipitation data into regression-tree software to develop an herbaceous annual mapping model. In sagebrush ecosystems, annual grasses, especially cheatgrass (Bromus tectorum), result in grass-fire cycles that endanger human-built structures, compromise recreational and hunting resources, destroy wildlife habitat, and reduce air quality. We applied a mask to areas above 2250-m elevation because cheatgrass is unlikely to exist at substantial cover above this threshold. To target likely sagebrush ecosystems, the mask also hid pixels classified as something other than shrub or grassland/herbaceous by the NLCD. We validated the model in two ways. We used 1) accuracy assessment data that the regression-tree software generated for training and independent test data sets. This assessment includes a correlation coefficient (r) value, an average absolute error, a relative absolute error, and a mean average error (MAE) and standard deviation based on nine model boot-strap randomizations. 2) High resolution remotely sensed herbaceous annual data spatially averaged to 250 m and BLM Terrestrial Assessment Inventory and Monitoring (AIM) data. These data were compared to the herbaceous annual cover maps in corresponding years, and we calculated r-values and the MAE. This time series of maps illustrates the spatio-temporal variability of annual grasses in the sagebrush ecosystem and can be accessed at doi:10.5066/F71J98QK.