Approximation of multi-year time series of XCO2 concentrations using satellite observations and statistical interpolation methods

Abstract

Long-term time series of CO2 concentrations are required for studying e.g., the effects of rising CO2 concentrations on plants and ecosystems. Different methods for approximating column-averaged dry-air mole fractions of CO2 (XCO2) have been developed to substitute missing local CO2 measurements.This study presents a novel method (XCO2SAT+) for determining monthly XCO2 concentrations for selected sites, based entirely on observations from multiple satellites as well as statistical methods in the case of missing satellite data. The product currently covers a period of 16 years from 2003 to 2018. XCO2SAT+ results were validated against data of five European sites of the Total Carbon Column Observing Network (TCCON) and compared to two established XCO2 approximation methods. All three approximation methods were compared to CO2 measurements of German towers from the Integrated Carbon Observation System (ICOS).The mean absolute error (MAE) between the XCO2SAT+ method and TCCON site Karlsruhe, Germany ranges between 0.5 and 1.1 ppm (0.1–0.3%) for years 2014–2018. Mean differences between XCO2SAT+ and XCO2 approximation method CAMS-GCM shows a slightly better MAE of 0.1–0.3 ppm for CAMS-GCM at two German TCCON sites. Multi-year comparisons with ground-level ICOS measurements revealed a comparable seasonality of CO2 concentrations during the vegetation period, but underestimated high winter concentrations, a pattern also observed when comparing with CAMS-GCM. XCO2SAT+ is a suitable alternative and simple method for providing long-term time series of monthly averaged XCO2 concentrations for selected sites, expanding the methodological toolbox for in-situ CO2 research. The new method, which so far provides XCO2 concentrations, will thus serve as a first step towards providing further CO2 data sources, which could be used to further improve numerical models such as atmospheric models or crop growth models and to study CO2 impacts in agriculture and natural ecosystems.