HR: 0800h
AN: B31B-0330    [Abstracts]
TI: De-trending for Climatic Variations to Reveal Stressed Ecosystems
AU: * Wylie, B K
EM: wylie@usgs.gov
AF: SAIC, contractor to U.S. Geological Survey (USGS) Center for Earth Resources, 47914 252nd St, Sioux Falls, sd 57198, United States
AU: Rover, J A
EM: jrover@usgs.gov
AF: SAIC, contractor to U.S. Geological Survey (USGS) Center for Earth Resources, 47914 252nd St, Sioux Falls, sd 57198, United States
AU: Zhang, L
EM: lizhang
AF: SAIC, contractor to U.S. Geological Survey (USGS) Center for Earth Resources, 47914 252nd St, Sioux Falls, sd 57198, United States
AU: Ji, L
EM: lij@usgs.gov
AF: SAIC, contractor to U.S. Geological Survey (USGS) Center for Earth Resources, 47914 252nd St, Sioux Falls, sd 57198, United States
AU: Bliss, N B
EM: bliss@usgs.gov
AF: SAIC, contractor to U.S. Geological Survey (USGS) Center for Earth Resources, 47914 252nd St, Sioux Falls, sd 57198, United States
AB: Short-term variation in climatic conditions results in dramatic fluxuations in ecosystem performance, particularly in water-limited ecosystems. These climatic-driven variations confound the effects of management, insect infestations, and changing soil conditions such as a thickening active soil layer in high latitudes. In this study, we presented a method to account for the influence of climatic variations on ecosystem performance, so that changes in underlying ecological condition are emphasized. We adopted a growing season integration of coarse resolution, remotely sensed, Normalized Difference Vegetation Index (NDVI) as a surrogate for actual ecosystem performance. Piecewise regression models were used to predict expected growing season performance. The models were developed using large samples of random pixels over multiple years from areas with the same land cover and within an ecoregion class. Independent variables in the models include seasonal (winter, spring, and summer) climate data (precipitation and temperature) and site potential or long-term historical performance. Model results were then used to construct annual maps representing the deviations, or performance anomalies, between expected ecosystem performance and actual ecosystem performance. Regression confidence limits were used to identify significant anomalies on the maps. Performance outside the confidence limits over multiple years indicates areas with consistent management, insect, fire, or ecosystem stress impacts. The trend in ecosystem anomalies over several years identifies areas where ecosystem stress is becoming more severe or less severe. This approach has been applied to sagebrush lands and boreal forests in the western United States using both 250 m resolution Moderate Resolution Imaging Spectroradiometer and 1 km resolution Advanced Very High Resolution Radiometer NDVI. Piecewise regression R² values have varied from 0.84 in the Yukon River Basin to 0.96 in Wyoming sagebrush regions, and from 0.86 to 0.95 for sagebrush and grasslands in a portion of southern Idaho. This approach holds promise for identifying stressed ecosystems that are vulnerable to changing to a new ecological state.
DE: 0428 Carbon cycling (4806)
DE: 0430 Computational methods and data processing
DE: 0439 Ecosystems, structure and dynamics (4815)
DE: 0475 Permafrost, cryosphere, and high-latitude processes (0702, 0716)
DE: 0480 Remote sensing
SC: Biogeosciences [B]
MN: 2007 Fall Meeting