PS 53-177 - A 28-year dataset to characterize the vegetation productivity of US rangelands

Wednesday, August 10, 2011
Exhibit Hall 3, Austin Convention Center
Ranjani Wasantha Kulawardhana1, Robert A. Washington-Allen2, Eric Schall3, Michael A. Austin4, Sorin C. Popescu1 and Matt C. Reeves5, (1)Department of Ecosystem Science and Management, Texas A & M University, College Station, TX, (2)Geography, University of Tennessee, Knoxville, TN, (3)Department of Oceanography, Texas A & M University,, College Station, TX, (4)Ecosystem Science and Management, Texas A & M University, College Station, TX, (5)Human Dimensions, USFS Rocky Mountain Research Station, Missoula, MT

The extent of global dryland degradation is unknown with estimates ranging from 10 – 80%. This uncertainty is due to monitoring phenomena such as livestock grazing and/or climatic events at inappropriate spatial and temporal scales. Time series analysis of ecological indicators of degradation [i.e., net primary productivity (NPP)] could help in assessing the extent of dryland degradation. Measures of vegetation productivity derived from historical archives of satellite data are now available at the appropriate spatial and temporal scales. Such data could be used to detect and separate the impact of anthropogenic activities and climatic events.

Two datasets that begin to meet the necessary space-and-time criteria for assessing rangeland degradation are the 8-km pixel resolution 25-year (1982 – 2006) normalized difference vegetation index (NDVI) data from the Global Inventory Modeling and Mapping Studies (GIMMS) project and 1-km resolution 10 year (2000 to 2009) NPP data derived from the Moderate Resolution Imaging Spectroradiometer (MODIS) data. The primary objective of this study is to standardize and calibrate the two satellite datasets to extend the NPP time series from 1982 to 2009.  We also wanted to address the recent controversy introduced by one national and two recent global studies of dryland degradation that used a single equation approach based on mean spatial variability between 8-km ∑NDVI and 1-km MODIS NPP to predict a longer NPP time series.


In this study ∑NDVI and NPP data over the data overlap period (2000 to 2006) were evaluated for their ability to predict NPP. Our approach incorporated both spatial and temporal variability by applying a pixel by pixel simple linear regression analysis on  a 7 year  time series.  We then distinguished the area grazed by livestock in the US, i.e., rangelands and determined that the overall trend of predicted MODIS NPP from 1982 to 2009 was increasing. This reflects an increase in carbon storage in drylands that is consistent with the results of warming experiments that show increased carbon gain with increased temperature and drought stress in both tall- and shortgrass prairie. However, this is also suggestive of increased woody encroachment in US rangelands at the national scale.

Copyright © . All rights reserved.
Banner photo by Flickr user greg westfall.