TY - JOUR
T1 - Adaptive selection of diurnal minimum variation
T2 - A statistical strategy to obtain representative atmospheric CO2 data and its application to European elevated mountain stations
AU - Yuan, Ye
AU - Ries, Ludwig
AU - Petermeier, Hannes
AU - Steinbacher, Martin
AU - Gómez-Pelaéz, Angel J.
AU - Leuenberger, Markus C.
AU - Schumacher, Marcus
AU - Trickl, Thomas
AU - Couret, Cedric
AU - Meinhardt, Frank
AU - Menzel, Annette
N1 - Publisher Copyright:
© Author(s) 2018.
PY - 2018/3/15
Y1 - 2018/3/15
N2 - Critical data selection is essential for determining representative baseline levels of atmospheric trace gases even at remote measurement sites. Different data selection techniques have been used around the world, which could potentially lead to reduced compatibility when comparing data from different stations. This paper presents a novel statistical data selection method named adaptive diurnal minimum variation selection (ADVS) based on CO2 diurnal patterns typically occurring at elevated mountain stations. Its capability and applicability were studied on records of atmospheric CO2 observations at six Global Atmosphere Watch stations in Europe, namely, Zugspitze-Schneefernerhaus (Germany), Sonnblick (Austria), Jungfraujoch (Switzerland), Izanã (Spain), Schauinsland (Germany), and Hohenpeissenberg (Germany). Three other frequently applied statistical data selection methods were included for comparison. Among the studied methods, our ADVS method resulted in a lower fraction of data selected as a baseline with lower maxima during winter and higher minima during summer in the selected data. The measured time series were analyzed for long-term trends and seasonality by a seasonal-trend decomposition technique. In contrast to unselected data, mean annual growth rates of all selected datasets were not significantly different among the sites, except for the data recorded at Schauinsland. However, clear differences were found in the annual amplitudes as well as the seasonal time structure. Based on a pairwise analysis of correlations between stations on the seasonal-trend decomposed components by statistical data selection, we conclude that the baseline identified by the ADVS method is a better representation of lower free tropospheric (LFT) conditions than baselines identified by the other methods.
AB - Critical data selection is essential for determining representative baseline levels of atmospheric trace gases even at remote measurement sites. Different data selection techniques have been used around the world, which could potentially lead to reduced compatibility when comparing data from different stations. This paper presents a novel statistical data selection method named adaptive diurnal minimum variation selection (ADVS) based on CO2 diurnal patterns typically occurring at elevated mountain stations. Its capability and applicability were studied on records of atmospheric CO2 observations at six Global Atmosphere Watch stations in Europe, namely, Zugspitze-Schneefernerhaus (Germany), Sonnblick (Austria), Jungfraujoch (Switzerland), Izanã (Spain), Schauinsland (Germany), and Hohenpeissenberg (Germany). Three other frequently applied statistical data selection methods were included for comparison. Among the studied methods, our ADVS method resulted in a lower fraction of data selected as a baseline with lower maxima during winter and higher minima during summer in the selected data. The measured time series were analyzed for long-term trends and seasonality by a seasonal-trend decomposition technique. In contrast to unselected data, mean annual growth rates of all selected datasets were not significantly different among the sites, except for the data recorded at Schauinsland. However, clear differences were found in the annual amplitudes as well as the seasonal time structure. Based on a pairwise analysis of correlations between stations on the seasonal-trend decomposed components by statistical data selection, we conclude that the baseline identified by the ADVS method is a better representation of lower free tropospheric (LFT) conditions than baselines identified by the other methods.
UR - http://www.scopus.com/inward/record.url?scp=85044141230&partnerID=8YFLogxK
U2 - 10.5194/amt-11-1501-2018
DO - 10.5194/amt-11-1501-2018
M3 - Article
AN - SCOPUS:85044141230
SN - 1867-1381
VL - 11
SP - 1501
EP - 1514
JO - Atmospheric Measurement Techniques
JF - Atmospheric Measurement Techniques
IS - 3
ER -