Ecosystem Modelling and Decision Support

Science Center Objects

The Ecosystem Modelling and Decision Support Project seeks to understand how drivers of ecosystem change like wildfire, drought, and land use affected past spatial and temporal patterns of vegetation communities and wildlife. Research methods involve 1) analyzing field-collected information (e.g. long-term plot/transect data, repeat photography) on soils, vegetation, and/or wildlife with multitemporal satellite data, 2) and the development and use of spatial and temporal statistical models, landscape metrics, and ecological indicators to inform management actions.

Fire and Ecosystem Restoration in the US-Mexico Borderlands

The borderland region of the United States and northern Mexico is expected to undergo both extended periods of drought and longer wildfire seasons under forecasted global climate change; it is important to understand how these disturbances interact and affect recovery trajectories and potentially shift plant community composition. This research leverages satellite imagery to map and analyze recent fire patterns, to measure post-fire recovery relative to drought, and to assess how ecosystem management  may influence factors like burn severity and timing, the spread of exotic plants, and the cover of trees and shrubs. Two studies are highlighted here.

The first study focuses on reconstructing recent fire history in forests of the transboundary Madrean Archipelego of the southwestern United States and northern Mexico (Figure 1). Mountain ranges in the Madrean ecoregion have similar species assemblages and topographic characteristics, but have been managed in strikingly different ways since the mid-1800s. Differences in land use and forest management, particularly active wildfire suppression in the U.S. and lack thereof in Mexico, have led to contrasting fire regimes which may provide information to help guide forest restoration in the U.S. and inform Mexico’s fire management plans.

In the U.S., federal scientists developed the Monitoring Trends in Burn Severity (MTBS) national database of fire location and burn severity, however similar data for many remote areas outside of the U.S. are generally sparse. To help fill this data gap, we developed a semi-automated process to identify historical wildfire occurrence in Mexico using Landsat Thematic Mapper data time series from 1985-2011. We circumvented the need for a priori knowledge of fire occurrence by combining differenced Normalized Burn Ratio (dNBR) images covering sequential and overlapping seasonal blocks, identifying 83 large (> 1,000 acre) wildfires in northern Mexico during the period. This trans-boundary burn severity dataset will be used to assess differences in fire severity, timing and pattern relative to management and climate.

Map showing the location of the transboundary Madrean Archipelago Ecoregion and upland fires

Figure 1. Map showing the location of the transboundary Madrean Archipelago Ecoregion (top left), a previously unmapped fire in Mexico from 1988 (bottom left), and the location of all upland fires (>1,000 acres) mapped for the period 1984-2011 with Landsat TM imagery.(Credit: Miguel Villarreal, USGS. Public domain.)

 The second study uses satellite time-series to monitor changes in rangeland vegetation cover and greenness caused by prescribed fire and drought. Using a 25-year fire atlas of prescribed fire and wildfire locations, we paired sites with multiple fires with unburned control areas and compared satellite and field-based estimates of vegetation cover over time (Figure 2). To identify post-fire shifts in native, non-native and annual plant cover, we analyze vegetation cover and greenness estimated from Landsat Thematic Mapper satellite data (Figure 3).  The data highlighted anomalous greening during drought periods that was related to increased annual and non-native plant cover (Figure 4). Satellite and field-data suggest that aggressive application of prescribed fire effectively reduces woody plant cover as intended, but may also encourage the spread of non-native grasses and annual plant cover, especially during drought periods. These observations can inform management restoration activities, including optimal timing relative to season and climate, to produce desired ecosystem outcomes.

Photo of crew measuring biomass

Figure 2. A field crew measures plant species and cover in a non-native dominated grassland (top) and a degraded site dominated by annual plants, shrubs and bare ground (bottom).(Credit: Miguel Villarreal, USGS. Public domain.)

Normalized Difference Vegetation Index and Total Vegetation Fractional Cover

Figure 3. Combining satellite indices that measure plant greenness and vegetation cover can be used to identify vegetation changes that are not apparent using a single index alone. The Normalized Difference Vegetation Index (NDVI, top) is a measure of greenness based on the near-infrared wavelengths, while the Total Vegetation Fractional Cover (TVFC, bottom) is based on the shortwave-infrared bands and provides a good estimate of both green and senescent vegetation cover. These two graphs show index values at sites with no fire (pink), four fires (blue) and five fires (green and red) between 1985 and 2011. The TVFC shows a clear response to fire (i.e. drop in cover in 1986 followed by recovery). TVFC values begin to diverge considerably after fires occur during drought periods (post-2000).(Credit: Miguel Villarreal,, USGS. Public domain.)


Maps of vegetation change

Figure 4. Differencing NDVI and TVFC provides a new metric that highlights areas with low cover but high greenness (positive index values), which is indicative of annual and non-native plant cover.(Credit: Miguel Villarreal, USGS. Public domain.)




Villarreal, M.L., Cortés Montaño, C. and Prickett, J. 2015. Recent patterns of wildfire severity across a transboundary ecoregion of the United States-Mexico borderlands. 9th IALE World Congress, July 5-10, 2015, Portland, OR.

Prickett, J.K. and Villarreal, M.L. 2014Integrating Remote Sensing and GIS for retrospective wildfire mapping in remote areas. GeCo in the Rockies, September 22-26, 2014. Grand Junction, CO. 

Villarreal, M.L., Wallace, C.S.A., Norman, L.M. and Coe, M. 2014. Effects of climate, soils and species composition on multitemporal satellite estimates of desert grassland cover. Association of American Geographers Annual Meeting, April 8-12, 2014, Tampa, FL.

Villarreal, M.L., Wallace, C.S.A., Buckley, S., Norman, L.M. and Coe, M. 2013. Measuring the interacting effects of drought and fire on grassland cover and phenology with a 30 year time-series of Landsat data.  12th Biennial Conference of Science and Management on the Colorado Plateau. September 16-19, 2013, Northern Arizona University, Flagstaff, AZ.