Tom Loveland | Search USGS

USGS2009

Sample project: establishing a global forest monitoring capability using multi-resolution and multi-temporal remotely sensed data sets

Quantifying rates of forest-cover change is important for improved carbon accounting and climate change modeling, management of forestry and agricultural resources, and biodiversity monitoring. A practical solution to examining trends in forest cover change at global scale is to employ remotely sensed data. Satellite-based monitoring of forest cover can be implemented consistently across large regions at annual and inter-annual intervals. This research extends previous research on global forest-cover dynamics and land-cover change estimation to establish a robust, operational forest monitoring and assessment system. The approach integrates both MODIS and Landsat data to provide timely biome-scale forest change estimation. This is achieved by using annual MODIS change indicator maps to stratify biomes into low, medium and high change categories. Landsat image pairs can then be sampled within these strata and analyzed for estimating area of forest cleared.

The NASA LCLUC Program: an interdisciplinary appro↗

USGS1996

Spatial partitioning of environmental correlates of avian biodiversity in the conterminous United States

Classification and regression tree (CART) analysis was used to create hierarchically organized models of the distribution of bird species richness across the conterminous United States. Species richness data were taken from the Breeding Bird Survey and were related to climatic and land use data. We used a systematic spatial grid of approximately 12,500 hexagons, each approximately 640 square kilometres in area. Within each hexagon land use was characterized by the Loveland et al. land cover classification based on Advanced Very High Resolution Radiometer (AVHRR) data from NOAA polar orbiting meteorological satellites. These data were aggregated to yield fourteen land classes equivalent to an Anderson level II coverage; urban areas were added from the Digital Chart of the World. Each hexagon was characterized by climate data and landscape pattern metrics calculated from the land cover. A CART model then related the variation in species richness across the 1162 hexagons for which bird species richness data were available to the independent variables, yielding an R2-type goodness of fit metric of 47.5% deviance explained. The resulting model recognized eleven groups of hexagons, with species richness within each group determined by unique sequences of hierarchically constrained independent variables. Within the hierarchy, climate data accounted for more variability in the bird data, followed by land cover proportion, and then pattern metrics. The model was then used to predict species richness in all 12,500 hexagons of the conterminous United States yielding a map of the distribution of these eleven classes of bird species richness as determined by the environmental correlates. The potential for using this technique to interface biogeographic theory with the hierarchy theory of ecology is discussed. ?? 1996 Blackwell Science Ltd.

Biodiversity Letters↗