Tower-based atmospheric CO2 mole fractions (raw, uncalibrated) measured as part of the National Carbon Program’s Mid Continent Intensive, U.S. Upper Midwest, 2007-2009

A total of nine communication tower-based CO2 sensors were located within a 500 x 800 km2 area within the Mid Continental Intensive (MCI) study region in the U.S. Upper Midwest from May 2007 through November 2009. With the goal of oversampling the MCI region, The Pennsylvania State University deployed instruments at five towers: Centerville (Iowa), Galesville (Wisconsin), Kewanee (Illinois), Mead (Nebraska), and Round Lake (Minnesota). Here we publish the raw, uncalibrated data from these five sites. The tower locations, sampling heights and calibration tanks are listed in the table below. These sites were instrumented with wavelength-scanned cavity ring-down spectroscopy (WS-CRDS) systems (Picarro, Inc., Santa Clara, California, model CADS). The air samples were dried with Nafion (Permapure, Inc) driers, calibrated prior to deployment and field calibrated daily. Richardson et al. [2012] document the quality assessment of the instruments during the MCI; 8 months of testing against a NOAA-ESRL NDIR system in West Branch, Iowa, yielded median daytime-only differences of 0.13 +- 0.63 ppm. In addition to these five sites, the National Oceanic and Atmospheric Administration (NOAA)’s Earth System Research Laboratory (ESRL) maintained measurements of CO2 mole fractions at two long-term tall towers in the study region (LEF, Park Falls, Wisconsin, and WBI, West Branch, Iowa), and two additional long-term sites measuring well-calibrated CO2 mole fraction (University of Minnesota KCMP, Rosemount, Minnesota, and Pennsylvania State University (PSU)/Oak Ridge National Laboratory (ORNL, Missouri Ozarks, Missouri) were located within the region. Miles et al. [2012] describes the spatial and temporal gradients observed via this dataset and Lauvaux et al. [2011] show the CO2 emissions derived from these data utilized in conjunction with an atmospheric transport model and a Bayesian inversion.