 |
| Titel |
Effects of biased CO2 flux measurements by open-path sensors on the interpretation of CO2 flux dynamics at contrasting ecosystems |
| VerfasserIn |
Manuel Helbig, Elyn Humphreys, Ivan Bogoev, William L. Quinton, Karoline Wischnewski, Oliver Sonnentag |
| Konferenz |
EGU General Assembly 2015
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 17 (2015) |
| Datensatznummer |
250101180
|
| Publikation (Nr.) |
 EGU/EGU2015-281.pdf |
|
|
|
|
|
| Zusammenfassung |
| Long-term measurements of net ecosystem exchange of CO2 (NEE) are conducted across a
global network of flux tower sites. These sites are characterised by varying climatic and
vegetation conditions, but also differ in the type of CO2/H2O gas analyser used to obtain
NEE.
Several studies have observed a systematic bias in measured NEE when comparing
open-path (OP) and closed-path (CP) sensors with consistently more negative daytime NEE
measurements when using OP sensors, both during the growing and non-growing season. A
surface heating correction has been proposed in the literature, but seems not to be universally
applicable. Systematic biases in NEE measurements are particularly problematic for
synthesis papers and inter-comparison studies between sites where the “true” NEE is small
compared to the potential instrument bias. For example, NEE estimates for boreal forest
sites derived from OP sensors show large, ecologically unreasonable winter CO2
uptake.
To better understand the causes and the magnitude of this potential bias, we conducted a
sensor inter-comparison study at the Mer Bleue peatland near Ottawa, ON, Canada. An eddy
covariance system with a CP (LI7000 & GILL R3-50) and an OP sensor (EC150 & CSAT3A)
was used. Measurements were made between September 2012 and January 2013 and covered
late summer, fall, and winter conditions. Flux calculations were made as consistently as
possible to minimise differences due to differing processing procedures (e.g. spectral
corrections).
The latent (LE, slope of orthogonal linear regression of LEOP on LECP: 1.02 ± 0.01 &
intercept: -0.2 ± 0.6 W m-2 and sensible heat fluxes (H, slope of HCSAT3A on HGILL:
0.96 ± 0.01 & intercept: 0.1 ± 0.03 W m-2) did not show any significant bias. However, a
significant bias was apparent in the NEE measurements (slope of NEEOP on NEECP: 1.36
± 0.02 & intercept: -0.1 ± 0.05). The differences between NEEOP and NEECP were
linearly related to the magnitude of HCSAT3A with a slope of -0.02 ± 0.001 μmol CO2 m-2
s-1 and an intercept of -0.1 ± 0.03 μmol CO2 m-2 s-1 (R2 = 0.82, p = 0.001) indicating a
consistent overestimation of CO2 uptake during the day and an overestimation of ecosystem
respiration during the night. Air temperatures did not have a significant effect on NEE
differences. Winter NEE measurements at two boreal forest, one boreal wetland, and one
tundra site show similar relationships with H further supporting the findings of this
study.
In contrast to OP sensors, CP sensors are less affected by high frequency air temperature
fluctuations and do not require a correction for air density fluctuations to obtain NEE. Our
results point toward a consistent bias in NEEOP that is likely related to the magnitude of H,
the main input to the WPL term. The additional findings from five contrasting ecosystems
suggest that the bias in NEEOP depends on the site-specific H regime, questioning the
accuracy of comparison studies across contrasting ecosystems. Since the absolute magnitude
of the bias seems to be directly related to the magnitude of H rather than to the
magnitude of NEE, the relative error is likely larger for sites with small NEE. These
findings are therefore particularly important for NEE studies at high latitude sites. |
|
|
|
|
|
|