Scaling-up knowledge of growing-season net ecosystem exchange for long-term assessment of North Dakota grasslands under the Conservation Reserve Program

Abstract

Scaling-up knowledge of land-atmosphere net ecosystem exchange (NEE) from a single experimental site to numerous perennial grass fields in the Northern Great Plains (NGP) requires appropriate scaling protocols. We addressed this problem using synoptic data available from the Landsat sensor for 10 growing seasons (April 15 to September 30) over a North Dakota field-site, where we continuously measured CO₂ exchange using a Bowen Ratio Energy Balance (BREB) system. NEE observed during the growing season at our field-site from 1997 to 2006 vacillated with drought and deluge, with net carbon (C) losses to the atmosphere in 2006. We used stepwise linear regression with 10 years of Landsat and NEE data to construct and validate a model for estimating grassland growing-season NEE from field to landscape scales. Eighty-nine percent of the variability in NEE was explained by year, live biomass, carbon : nitrogen ratio, day of image acquisition, and annual precipitation. We then applied this model on 20 620 ha of North Dakota perennial grass fields enrolled in the Conservation Reserve Program (CRP), including 1272 fields east of the Missouri River and 165 fields west-river. Growing-season NEE for individual CRP fields was highly variable from 1997 to 2006, ranging from −366 to 692 g C m⁻² growing season⁻¹. Mean annual growing-season fluxes over 10 years for CRP fields located east-river and west-river were 317 g C m⁻² growing season⁻¹ and 239 g C m⁻² growing season⁻¹, respectively. Average cumulative growing-season NEE modeled for fields east- and west-river diverged from one another in 2002–2006, when west-river fields received < 70% of the long-term annual average precipitation during these years. Results indicate assessment of conservation practices on grassland CO₂ exchange during the growing season can be remotely estimated at field and landscape scales under variable environmental conditions and should be followed up with similar, spatially explicit investigations of NEE during the dormant season.