Great Plains conservation
Sagebrush conservation



The Rangeland Analysis Platform's vegetation cover product provides annual percent cover estimates from 1984 to 2019 of: annual forbs and grasses, perennial forbs and grasses, shrubs, trees, and bare ground. The estimates were produced by combining approximately 60,000 field plots from the NRCS National Resources Inventory (NRI) and the BLM Assessment, Inventory, and Monitoring (AIM) datasets with the historical Landsat satellite record.

Utilizing the computation power of Google Earth Engine and Google AI Platform, annual cover estimates are produced across the western half of the U.S. at 30m resolution, an area slightly larger than a baseball diamond.

When interpreting predicted vegetation cover, it is important to consider the model error specific to each vegetation functional group, as shown in the following table:

Vegetation cover Annual forbs
and grasses
Perennial forbs
and grasses
Shrubs Trees Bare ground
Mean absolute error (%) 7.0 10.3 5.8 2.8 6.7
Root mean square error (%) 11.0 14.0 8.3 6.8 9.8

These errors provide an accuracy assessment. In basic terms, the vegetation cover value of a given pixel should be thought of as plus or minus the error. For example, the confidence of a 35% annual forb and grass pixel is +/- 7.0% (mean absolute error). Error metrics are notably higher for perennial vegetation due to the greater possible range of values, particularly in the Great Plains region.

The vegetation cover data and maps are intended to be used alongside local knowledge and on-the-ground data to inform management actions that improve rangelands and wildlife habitat. They should not be used in isolation to quantify rangeland resources, determine or define cover thresholds, or evaluate the efficacy of management practices or treatments. The RAP can be used to evaluate resources in concert with site-specific information about the area under investigation, such as past land management practices, vegetation treatments, conservation efforts, or natural disturbances.

Data are available within Google Earth Engine (ImageCollection 'projects/rangeland-analysis-platform/vegetation-cover-v2') and as GeoTIFFs from

Allred, B.W., B.T. Bestelmeyer, C.S. Boyd, C. Brown, K.W. Davies, L.M. Ellsworth, T.A. Erickson, S.D. Fuhlendorf, T.V. Griffiths, V. Jansen, M.O. Jones, J. Karl, J.D. Maestas, J.J. Maynard, S.E. McCord, D.E. Naugle, H.D. Starns, D. Twidwell, and D.R. Uden. 2020. Improving Landsat predictions of rangeland fractional cover with multitask learning and uncertainty. bioRxiv:2020.06.10.142489.

Great Plains conservation

Great Plains data are available as GeoTIFFs from

Categorical tree cover

Annual categorical tree cover across the Great Plains biome. Class categorization was performed with the rangeland cover product at 480m resolution. Agriculture, development, and water were masked out according to the NLCD 2016 Land Cover product. Data represents tree canopy cover in the following classes: 0-4% and >=5%

Cultivation risk

The probability of cultivation relative to climate (30 year normalization of mean annual precipitation, mean annual temperature, number of degree days greater than five degrees C, and duration of frost free period); soils (water holding capacity and other hydrological characteristics); and topography (compound topographic index, neighborhood roughness, slope, and surface relief). Various dataets were used, including regional climate data, USDA NRCS SSURGO, and USDA NASS Cropland Data Layer (CDL). Independent models were produced for each county; county level predicitons were merged for state coverage.

Lipsey, M.K., K.E. Doherty, D.E. Naugle, S. Fields, J.S. Evans, S.K. Davis and N. Koper. 2015. One step ahead of the plow: Using cropland conversion risk to guide grassland songbird conservation. Biological Conservation. 191:739–749.

Smith, J. T., J. S. Evans, B. H. Martin, S. Baruch-Mordo, J. M. Kiesecker, and D. E. Naugle. 2016. Reducing cultivation risk for at-risk species: Predicting outcomes of conservation easements for sage-grouse. Biological conservation 201:10–19.

Woody transitions

This product provides a rapid screening tool for identifying the leading edge of vegetation transitions in Great Plains rangelands and serves as an early warning for the loss of intact rangelands to woody expansion. Alternative ecological states are incapable of coexisting in the same space at a given point in time (Uden et al. 2019). For example, a savanna represents co-dominance of grasses and trees, meaning that grass-tree functional groups relatively coexist in the same space over time. In contrast, grass-tree functional groups fail to coexist in many regions of the world and represent fundamental alternative states in those instances. Transitions from one state to another are known to exhibit strong spatial order (Allen et al. 2016, Roberts et al. 2019); therefore, functional groups that do not coexist should covary negatively in space. This data product maps the geographic boundaries between grassy and woody states at two scales. When the geographic location of these boundaries change over time, rangeland resilience is being eroded (Uden et al. 2019) and the system is increasingly vulnerable to being lost to woody expansion. This is represented in the data product when a signal is present in a given year, persistent over multiple years, and non-stationarity in its geographic location over time.

For planning purposes, two scales were pulled from a multitude of scales in a multi-scale analysis (following Uden et al. 2019) and provided as woody transitions layers. The broad-scale layer is computed and visualized to represent the boundary of the Great Plains grasslands biome, with scale being a product of both moving window (i.e., kernel) dimension and pixel resolution (i.e., grain). The broad-scale layer was computed with a 139 x 139-pixel moving window algorithm at 240-meter resolution. A second, moderate-scale was computed and visualized for regional-scale planning and represents the general spatial boundaries between alternative grassy and woody ecological states at a finer scale. The moderate-scale layer was computed with an 81 x 81-pixel moving window algorithm at 30-meter resolution. Increasingly negative spatial covariance values between grass and tree functional groups represent increasingly severe segregation of grass/tree functional groups in space. Fill value = NA/NoData; corresponds to developed areas, croplands, or water, as classified in the 2011 National Land Cover Database (Homer et al. 2015), or wetlands and their 60-m buffers, as delineated in the National Wetlands Inventory (U.S. Fish and Wildlife Service 2019).

Input data: Perennial forbs/grasses vs. trees percent cover data. Sourced from Rangeland Analysis Platform (RAP) cover v2.0,

Data use and interpretation must follow guidelines set forth at, Uden et al. (2019), and Allred et al. (2020).

Reminders and limitations

Transition data are meant to be combined with RAP cover data and local expert knowledge to understand vegetation change and is not meant to replace those products.

When mapping spatial covariance data, upper and lower values over which the color ramp is stretched influence the sensitivity of the early warning signal of vegetation change and the display of transition severity. Such optimization of images may be useful for mapping variability over different ranges of transition severity in different locations; however, users should be advised that these adjustments also affect the potential for false positives/negatives in spatial transition detection.

Transition signals were computed for conservation planning at two scales and represent a subset of a more robust multi-scale analysis. Cross-scale considerations of transitions will elucidate patches of intact rangeland vegetation nested within broader regional patterns shown here. Product testing is underway on cross-scale products and outputs.

Only one functional group combination from RAP data is present in this data. Alternative functional group combinations may provide additional insights into vegetation transitions at various scales.

Allen, C.R., D.G. Angeler, G.S. Cumming, C. Folke, D. Twidwell, and D.R. Uden. 2016. Quantifying spatial resilience. Journal of Applied Ecology 53:625–635.

Allred, B.W., B.T. Bestelmeyer, C.S. Boyd, C. Brown, K.W. Davies, L.M. Ellsworth, T.A. Erickson, S.D. Fuhlendorf, T.V. Griffiths, V. Jansen, M.O. Jones, J. Karl, J.D. Maestas, J.J. Maynard, S.E. McCord, D.E. Naugle, H.D. Starns, D. Twidwell, and D.R. Uden. 2020. Improving Landsat predictions of rangeland fractional cover with multitask learning and uncertainty. bioRxiv:2020.06.10.142489.

Homer, C., J. Dewitz, L. Yang, S. Jin, P. Danielson, G. Xian, J. Coulston, N. Herold, J. Wickham, and K. Megown. 2015. Completion of the 2011 National Land Cover Database for the conterminous United States–representing a decade of land cover change information. Photogrammetric Engineering & Remote Sensing 81:345–354.

Roberts, C.P., C.R. Allen, D.G. Angeler, and D. Twidwell. 2019. Shifting avian spatial regimes in a changing climate. Nature Climate Change 9:562–568.

U.S. Fish and Wildlife Service. 2019. National Wetlands Inventory website. U.S. Department of the Interior, Fish and Wildlife Service, Washington, D.C., U.S.A. Available online at:

Uden, D. R., D. Twidwell, C. R. Allen, M. O. Jones, D. E. Naugle, J. D. Maestas, and B. W. Allred. 2019. Spatial Imaging and Screening for Regime Shifts. Frontiers in Ecology and Evolution 7:407.

Sagebrush conservation

Sagebrush data are available as GeoTIFFs from

Annual herbaceous cover

Estimated percent cover of herbaceous annuals at 30m resolution on rangelands across the sagebrush biome. Non-rangeland areas, such as forests, water, crops, and development are excluded. These data are a weighted average of three large-scale datasets, providing land managers with estimates of recent (2016-2018) annuals cover across western rangelands. This data layer was developed by a cheatgrass committee convened by the Western Governors Association-appointed Western Invasive Species Council as part of a new toolkit for invasive annual grass management across the western U.S. Detailed methods are available.

Maestas, J., Jones, M., Pastick, N.J., Rigge, M.B., Wylie, B.K., Garner, L., Crist, M., Homer, C., Boyte, S., and Witacre, B., 2020, Annual Herbaceous Cover across Rangelands of the Sagebrush Biome: U.S. Geological Survey data release,

Categorical tree cover

Annual categorical tree cover across the Sagebrush biome. Class categorization was performed with the rangeland cover product at 30m resolution. Agriculture, development, and water were masked out according to the NLCD 2016 Land Cover product. Data represents tree canopy cover in the following classes: 0-1%, 2-10%, 11-20%, and >=21%

Ecosystem resilience and resistance

This data provides a tool for rapid risk assessment across the range of sage-grouse using an index of sagebrush ecosystem resilience to disturbance and resistance to cheatgrass ("R&R"). Potential ecosystem R&R depends in part on the biophysical conditions an area is capable of supporting and soil temperature and moisture regimes can be used to depict this gradient (Chambers et al. 2014, 2016, 2017; Maestas et al. 2016). Soils data were derived from two primary sources: 1) completed and interim soil surveys available through the Soil Survey Geographic Database (SSURGO), and 2) the State Soils Geographic Database (STATSGO2) to fill gaps where SSURGO data were not available (Fig. 1). Using best available information and expert input, each soil temperature and moisture regime/moisture subclass was placed into one of three categories of relative R&R: high, moderate, and low (see: Chambers et al. 2014, 2016; Maestas et al. 2016). Soils with high water tables, wetlands, or frequent ponding that would not typically support sagebrush were not rated.

Chambers et al. 2014. Using resistance and resilience concepts to reduce impacts of annual grasses and altered fire regimes on the sagebrush ecosystem and sage-grouse– A strategic multi-scale approach. Fort Collins, CO, USA: U.S. Department of Agriculture, Forest Service, RMRS-GTR-326.

Chambers et al. 2016. Using resilience and resistance concepts to manage threats to sagebrush ecosystems, Gunnison sage-grouse, and greater sage-grouse in their eastern range: a strategic multi-scale approach. Gen. Tech. Rep. RMRS-GTR-356. Fort Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station.

Chambers et al. 2017. Using Resilience and Resistance Concepts to Manage Persistent Threats to Sagebrush Ecosystems and Greater Sage-grouse. Rangeland Ecology and Management. 70:149-164.

Maestas, J. D., and S. B. Campbell. 2014. Mapping Potential Ecosystem Resilience and Resistance across Sage-Grouse Range using Soil Temperature and Moisture Regimes. Fact Sheet. Sage Grouse Initiative.

Maestas et al. 2016. Tapping Soil Survey Information for Rapid Assessment of Sagebrush Ecosystem Resilience and Resistance. Rangelands 38:120-128.

RAP derivatives

Other select datasets derived from the Rangeland Analysis Platform are available from