Vegetation indices, CO2 flux, and biomass for Northern Plains grasslands

Abstract

Native grasslands are a sink for atmospheric CO2 sequestration, but ways for extending site-specific CO2 flux measurements to a regional scale are lacking. Objectives of this study were to determine the utility of using canopy radiometric reflectance for estimating CO2 fluxes for semiarid grasslands. The relationship between the normalized difference vegetation index (NDVI) calculated from spectral reflectance data obtained with hand-held radiometers was compared to CO2 flux calculated from Bowen ratio/energy balance measurements. Carbon dioxide flux was measured during the plant growing season over a nongrazed prairie, grazed prairie, and a shrub dominated prairie site near Mandan, N.D. Measurements were also made of evapotranspiration (ET), green biomass, and green leaf area index (LAI). Correlation coefficients of NDVI with vegetation parameters of biomass and LAI for each site and year exceeded 0.84 in 1999, 0.74 in 2000, and 0.91 in 2001; with CO2 flux correlations exceeded 0.63 in 1999, 0.68 in 2000, and 0.69 in 2001; with ET correlations exceeded 0.91 in 1999, 0.92 in 2000, and 0.90 in 2001. Regression analysis over all years and sites produced a nonlinear relation between NDVI and both biomass (R2 = 0.83) and LAI (R2 = 0.77) and a linear relationship between NDVI and both CO2 flux (R2 = 0.51) and ET (R2 = 0.81). The relationships between NDVI and biomass, LAI, CO2 flux, and ET for the 3 grassland sites, which differed in management and vegetation, were generally quite similar suggesting that NDVI has potential for use in predicting canopy CO2 flux rates for semiarid grasslands in the Northern Great Plains.