How many measurements are needed to estimate accurate daily and annual soil respiration fluxes? Analysis using data from a temperate rainforest
Making accurate estimations of daily and annual <i>R</i><sub>s</sub> fluxes is key for understanding the carbon cycle process and projecting effects of climate change. In this study we used high-frequency sampling (24 measurements per day) of <i>R</i><sub>s&...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Copernicus Publications
2016-12-01
|
| Series: | Biogeosciences |
| Online Access: | http://www.biogeosciences.net/13/6599/2016/bg-13-6599-2016.pdf |
| Summary: | Making accurate estimations of daily and annual <i>R</i><sub>s</sub> fluxes is key
for understanding the carbon cycle process and projecting effects of climate
change. In this study we used high-frequency sampling (24 measurements per
day) of <i>R</i><sub>s</sub> in a temperate rainforest during 1 year, with the
objective of answering the questions of when and how often measurements
should be made to obtain accurate estimations of daily and annual
<i>R</i><sub>s</sub>. We randomly selected data to simulate samplings of 1, 2, 4 or
6 measurements per day (distributed either during the whole day or only
during daytime), combined with 4, 6, 12, 26 or 52 measurements per year. Based
on the comparison of partial-data series with the full-data series, we
estimated the performance of different partial sampling strategies based on
bias, precision and accuracy. In the case of annual <i>R</i><sub>s</sub>
estimation, we compared the performance of interpolation vs. using non-linear
modelling based on soil temperature. The results show that, under our study
conditions, sampling twice a day was enough to accurately estimate daily
<i>R</i><sub>s</sub> (RMSE < 10 % of average daily flux), even if both
measurements were done during daytime. The highest reduction in RMSE for the
estimation of annual <i>R</i><sub>s</sub> was achieved when increasing from four to
six
measurements per year, but reductions were still relevant when further
increasing the frequency of sampling. We found that increasing the number of
field campaigns was more effective than increasing the number of measurements
per day, provided a minimum of two measurements per day was used. Including
night-time measurements significantly reduced the bias and was relevant in
reducing the number of field campaigns when a lower level of acceptable error
(RMSE < 5 %) was established. Using non-linear modelling instead of
linear interpolation did improve the estimation of annual <i>R</i><sub>s</sub>, but
not as expected. In conclusion, given that most of the studies of
<i>R</i><sub>s</sub> use manual sampling techniques and apply only one measurement
per day, we suggest performing an intensive sampling at the beginning of the
study to determine minimum daily and annual frequencies of sampling. |
|---|---|
| ISSN: | 1726-4170 1726-4189 |