Drylands Home | Human Health | Native Lands | Public Lands

Early-season invasive plants - measuring their temporal and spatial dynamics

Over large areas of the western United States, native vegetation has been displaced by early season invasive plants, including annual grasses, such as cheatgrass (Bromus tectorum) and red brome (Bromus rubens), and annual forbs, such as redstem filaree (Erodium cicutarium) and African mustard (Malcomia africana).

Erodium circutarium
Sheri Hagwood @ USDA-NRCS PLANTS Database

Bromus tectorum
Cassondra Skinner @ USDA-NRCS PLANTS Database

In semi-arid regions of the Colorado Plateau, the growth patterns of these early season invasives (ESI) are out of step with most of the native species.  A software system called DESI (Detection of Early Season Invasives) has been developed to detect these invaders using a time series of images acquired by satellites.  In these images, the invaders green up earlier in the spring than other plants.

Seasonal trends in greenness (dNDVI) for cheatgrass-invaded (solid line) and non-invaded sites (dashed line) in Canyonlands National Park in 2001 from MODIS data.

By using satellite images from Landsat sensors, the normalized difference vegetation index (NDVI) can be calculated from bands in the red and near-infrared wavelengths to identify the signal of ESI.  High NDVI values indicate the presence of abundant green vegetation.  Low NDVI values indicate a low fraction or absence of green plants on the surface.  To produce the map of potential cheatgrass infestation, NDVI calculated from Landsat data from mid-summer is subtracted from the early growing season (spring) NDVI:

dNDVI = NDVI_spring – NDVI_mid-summer

As a result, high dNDVI values indicate areas that are covered by plants that were green in spring and brown in mid-summer (that is, areas with a vegetation growth cycle like that of cheatgrass and other early season invasive plants).  The DESI software uses this image to create a map of ESI.  A map of ESI computed from satellite data collected on April 15 and July 4, 2001, over Canyonlands National Park in southeastern Utah, is shown below, where red indicates areas of potential infestation by invasives.

The accuracy of DESI has been evaluated using field data in Canyonlands National Park, Dinosaur National Monument, and Washington County, Utah.  Preliminary results from comparisons of DESI maps of invasives to field observations show accuracy is as high as 89% in detecting ESI covering as little as 5% of the area of a pixel (Landsat pixel size is nominally 30 meters).  Accurate detection at low levels of cover is important in semi-arid regions where plants are typically very sparse.

As the amount, timing and location of precipitation events on the Colorado Plateau can vary dramatically from year to year, the ability to track fluctuations in cover by invasive species using satellite images is useful for linking patterns of ESI growth and expansion to land use, climate change, and control measures.  The image below shows the variation in invasive cover over several years from 1993 to 2006.

DESI is being used to assist land managers with the National Park Service (NPS) and other federal and local agencies in understanding vegetation changes on public lands and combating the spread of invasives into sensitive ecosystems, including areas containing threatened and endangered species.

Partnerships:

Published Reports and Available Software:

Kokaly, R.F., 2011, Detecting cheatgrass on the Colorado Plateau using Landsat data: A tutorial for the DESI software: U.S. Geological Survey Open-File Report 2010–1327, 88 p.