Remote Sensing of Vegetation Recovery in Grasslands After the 1988 Fires in Yellowstone National Park

Abstract

Traditional methods for measurement of vegetative characteristics can be time-consuming and labor-intensive, especially across large areas. Yet such estimates are necessary to investigate the effects of large scale disturbances on ecosystem components and processes. Because foliage of plants differentially absorbs and reflects energy within the electromagnetic spectrum, one alternative for monitoring vegetation is to use remotely sensed spectral data (Tueller 1989). Spectral indices developed from field radiometric and Landsat data have been used successfully to quantify green leaf area, biomass, and total yields in relatively homogeneous fields for agronomic uses (Shibayama and Akiyama 1989), but have met with variable success in wildland situations (Pearson et aL 1976). Interference from soils (Hardinsky et al. 1984, Huete et al. 1985), weathered litter (Huete and Jackson 1987), and senesced vegetation (Sellers 1985) have diminished the relationship between green vegetation characteristics and various vegetation indices. In 1987, we found that a linear combination of Landsat Multi-spectral Scanner (MSS) band 7 and the ratio of MSS bands 6 to 4 explained 63% of the variation in green herbaceous phytomass (GHP) in sagebrush-grasslands on ungulate summer range in the northeastern portion of Yellowstone National Park (Merrill et al. 1993). The extensive fires that occurred in the Park in the summer of 1988 provided an opportunity to determine whether remote sensing could be used to estimate green phytomass in burned areas and to monitor grassland vegetation recovery in the Park after the fires. Remote sensing has previously been used to follow succession of seral stages in pine forests (Jakubauskas et al. 1990) after burning and to monitor plant cover in tundra (Hall et al. 1980) after wildfires. The objectives of our study were to: (1) develop a model for predicting GHP in sagebrush­ grassland communities using Landsat TM spectral information and field data on GHP for 2 years, (2) validate the model by comparing predictions made from the model to actual field data collected in a third year, and if successful (3) compare initial vegetation recovery in burned areas relative to unburned sagebrush-grassland.