As it was stated before, data gaps occur inevitably during long term ecosystem studies due to instrument malfunction or other factors (e.g. calibration periods, data storage problems, human factor, etc.). The profile method turned out to be inadequate for long term NEE calculation but useful for filling the data gaps.
Yearly net carbon dioxide exchange can be inferred from the monthly averages
of NEE (Moncrieff et al., 1996), or from the daily sums of CO
exchange
between the atmosphere and the biosphere. The latter method requires NEE information
for each day of year. To complete this task, a data gap filling procedure must
be developed (Falge et al., 2001) as an important step towards a defensible
NEE study.
In our case, the first guess for each day is the monthly average daily NEE cycle
calculated directly from the measured NEE (eddy flux at 82 m + rate of change
of CO storage below, i.e. 
). All further step modifies
this daily cycle where the improved estimate is available, leaving the rest
of the cycle intact.
The second estimate is the modelled NEE calculated from the NEE-PPFD function
during daytime (if PPFD is known), and from the NEE-
function during
nighttime (if is known).
The third estimate is the modified profile CO flux described in section
![[*]](http://nimbus.elte.hu/icons/latex2html/crossref.gif){kind=icon}
.
The fourth estimate is only applied in cases when 
is available,
but there is no 
available. If this is the case, the monthly average
daily cycle is used to estimate the current storage term.
The fifth -- and best -- estimate is the measured NEE (=).
It was described in section that the percentage of instationarity
is calculated for each hourly period to test the quality of the measurement
(Foken and Wichura, 1996). During the fifth step NEE values measured during
periods with instationarity exceeding 30% are rejected.
Many authors use the empirical light-response function and the NEE-temperature
function to fill data gaps (e.g. Valentini et al., 1996; Anthoni et al., 1999)
at their site. Our approach is to use the semi-empirical similarity theory for
this purpose, since it relies on a dynamic method compared to the empirical,
statistics-based environmental forces method. As it was seen in section ,
the nighttime NEE-
function exhibits large scatter (e.g. Greco and Baldocchi,
1996; Valentini et al., 1996), which means that estimates based on the empirical
NEE- function may provide inaccurate estimates.
Having all neccessary measured data, and having a methodology to fill measurement gaps, we are capable to present the long term NEE time series for the two measuring system.
|
Figure shows the temporal variation of the daily net CO
exchange during 1997, 1998 and 1999 as a function of time. Since direct flux
measurements started at the end of April 1997, monthly average daily cycles
were not available for January-April. The neccessary daily cycles are calculated
as the average of the 1998 and 1999 monthly values from January to April. Further
steps of the gap filling is performed as it was described earlier in this section.
The neccessay empirical environmental functions are presented in section .
Seasonality is evident in Fig. . CO uptake exceeded
25 g CO m
day
on some days during the growing
season in some cases. However, there were also days when the biosphere lost
carbon to the atmosphere. Wintertime NEE remains considerably larger than zero
indicating active respiration even during the coldest days.
Figure shows the cumulative carbon exchange for the 1997,
1998 and 1999 based on the presented data.
The calculations show that during 1997 CO NEE was -491 g CO m year
(-134 g C m year), during 1998, -537 g CO m year
(-146 g C m year) and during 1999, -337 g CO m year
(-92 g C m year). It means that during the 3 year period
of 1997-1999 the region sequestered 372 g C m.
The dynamics of these fluxes can be better understood by breaking them down
into subcomponents. NEE is defined as the sum of gross primary production (GPP)
and total ecosystem respiration (R
). R is defined as the
sum of autotrophic respiration (R
) and heterotrophic respiration
(R
), but this decomposition needs the measurement of at least one
of the two components. As it is not performed at our site because of the large
diversity of species, we only investigate GPP and R. Neither net
primary production (NPP=NEE-R) nor soil carbon flux estimate is possible
for the same reason. This is generally possible in sites with homogeneous vegetation
and appropriate measuring devices (e.g. using the cuvette or chamber method).
GPP and R is calculated using the NEE- relationship
determined in section . Daily respiration is calculated
from the nighttime NEE (which is actual respiration) and the modelled daytime
respiration (missing temperature data is estimated from the monthly average
daily course of temperature). GPP is calculated as NEE-R.
|
Figure shows the annual and interannual cycle of NEE, R
and GPP for 3 years between 1997 and 1999. Variability is present in the ecosystem
respiration, but the interannual variability of NEE is mainly caused by differences
in GPP (i.e. photosynthesys).
Maximum values of NEE are around 4 g C m day in 1997
and 1999, but reach 6 g C m day in 1998. Maximum rates
of GPP are less than 9 g C m day in 1997 but almost
reach 10 g C m day in 1998 and 1999. The maximum respiration
occurs in 1998.
Table summarizes the calculated NEE and its subcomponents
for each year from 1997 to 1999.
|
