Skip to main content
ARS Home » Plains Area » Mandan, North Dakota » Northern Great Plains Research Laboratory » Research » Publications at this Location » Publication #209023

Title: Determination of Rangeland Biomass Using Landsat from 1997 to 2003

Author
item BEERI, OFER - UNIV OF ND,GRAND FORKS,ND
item Phillips, Beckie
item FRANK, AL - COLLABORATOR,RETIRED-ARS

Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: 4/20/2007
Publication Date: 10/28/2007
Citation: Beeri, O., Phillips, B.L., Frank, A. 2007. Determination of Rangeland Biomass Using Landsat from 1997 to 2003. Meeting Abstract.

Interpretive Summary:

Technical Abstract: Sustainable range management calls for accurate estimations of grassland production over time under variable climactic conditions. The problem with spectral vegetation indices, such as the Normalized Differences Vegetation Index (NDVI), is the lack of quantitative data (e.g., kg ha-1) necessary for year to year comparisons. We aimed to determine the mass of live and senescent vegetation using Landsat data only over multiple years for a moderately grazed mixed-grass prairie pasture. We measured the mass of live and senescent vegetation every two weeks between April and September from 1997 to 2003 at the Northern Great Plains Research Laboratory by clipping to ground level all vegetation within 0.25 m-2 plots. Landsat imagery was acquired within one week of each sample collection and calibrated to ground reflectance. We developed different algorithms for live and senesce mass with data from three consecutive growing seasons, and evaluated accuracy with data for the following four seasons. We found that root-mean-squire-error (RMSE) for live mass using NDVI was 444 kg ha-1, while the RMSE for live mass using EVI was 374 kg ha-1. The lowest RMSE was recorded using the MSAVI index, which was 248 kg ha-1. Senescent material, on the other hand, was estimated using the red and the short-wave infrared bands, with an RMSE of 474 kg ha-1. Despite differences in precipitation and grassland productivity, the models depicted the amount of live and senesce biomass with reasonable accuracy at a 30-m pixel scale. These models are useful tools for estimating grassland production from archived data without ancillary information.