ESSDEarth System Science DataESSDEarth Syst. Sci. Data1866-3516Copernicus PublicationsGöttingen, Germany10.5194/essd-10-1327-2018Upscaled diurnal cycles of land–atmosphere fluxes: a new global half-hourly data productUpscaled diurnal cyclesBodesheimPaulpbodes@bgc-jena.mpg.deJungMartinGansFabianMahechaMiguel D.https://orcid.org/0000-0003-3031-613XReichsteinMarkusMax Planck Institute for Biogeochemistry, Jena, GermanyGerman Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, Leipzig, GermanyMichael Stifel Center Jena (MSCJ) for Data-Driven & Simulation Science, Jena, GermanyPaul Bodesheim (pbodes@bgc-jena.mpg.de)20July20181031327136519November201729January201829May201813June2018This work is licensed under the Creative Commons Attribution 4.0 International License. To view a copy of this licence, visit https://creativecommons.org/licenses/by/4.0/This article is available from https://essd.copernicus.org/articles/10/1327/2018/essd-10-1327-2018.htmlThe full text article is available as a PDF file from https://essd.copernicus.org/articles/10/1327/2018/essd-10-1327-2018.pdf
Interactions between the biosphere and the atmosphere can be well
characterized by fluxes between the two. In particular, carbon and energy
fluxes play a major role in understanding biogeochemical processes on an
ecosystem level or global scale. However, the fluxes can only be measured at
individual sites, e.g., by eddy covariance towers, and an upscaling of these
local observations is required to analyze global patterns. Previous work
focused on upscaling monthly, 8-day, or daily average values, and global maps
for each flux have been provided accordingly. In this paper, we raise the
upscaling of carbon and energy fluxes between land and atmosphere to the next
level by increasing the temporal resolution to subdaily timescales. We
provide continuous half-hourly fluxes for the period from 2001 to 2014 at
0.5∘ spatial resolution, which allows for analyzing diurnal cycles
globally. The data set contains four fluxes: gross primary production (GPP),
net ecosystem exchange (NEE), latent heat (LE), and sensible heat (H). We
propose two prediction approaches for the diurnal cycles based on large-scale
regression models and compare them in extensive cross-validation experiments
using different sets of predictor variables. We analyze the results for a set
of FLUXNET tower sites showing the suitability of our approaches for this
upscaling task. Finally, we have selected one approach to calculate the
global half-hourly data products based on predictor variables from remote
sensing and meteorology at daily resolution as well as half-hourly potential
radiation. In addition, we provide a derived product that only contains
monthly average diurnal cycles, which is a lightweight version in terms of
data storage that still allows studying the important characteristics of
diurnal patterns globally. We recommend to primarily use these monthly
average diurnal cycles, because they are less affected by the impacts of
day-to-day variation, observation noise, and short-term fluctuations on
subdaily timescales compared to the full half-hourly flux products. The
global half-hourly data products are available at 10.17871/BACI.224.
Introduction
Understanding the coupling of the atmosphere and the biosphere is key to
understanding Earth system dynamics and ultimately to predict future
trajectories based on dynamic and fully coupled Earth system models
. Observations of energy and carbon fluxes obtained by the
eddy covariance technique have revealed major insights into land–atmosphere
interactions see the overview by, but the
measurements are local by nature and it remains difficult to derive global
inferences. To overcome this limitation, continental to global scale products
of biosphere–atmosphere fluxes have been produced using machine learning
techniques that combine flux tower measurements, observations from remote
sensing, and climate data . These products
proved to be useful, for example, in terms of assessing large-scale patterns of
biosphere–atmosphere fluxes with climate data or to
provide cross-consistency checks for process-model simulations
. The general principle of this upscaling approach has
been to exploit relationships between climate or satellite-based driver
variables like temperature or leaf area index, and the targeted
biosphere–atmosphere flux . In the first (“training”)
step, a machine learning model of the flux data is established based on the
driver variables across a regional or global network of towers. In the second
(“production”
Note that an alternative notion would be to use the
term “prediction” here. However, in the climate community “prediction” is
typically used for future scenarios, while in machine learning the
application domain could be also at ungauged spatial locations
) step,
the model is applied to large spatial domains where only gridded estimates of
the drivers are available. Machine learning techniques are very effective
here since they are fully data-adaptive, do not require initial assumptions
on functional relationships, and can cope with nonlinear dependencies.
One of the first upscaling papers by deals with empirical
upscaling of monthly average values of gross primary production (GPP)
obtained from a biosphere model. They propose using a model tree ensemble
approach to perform the predictions and introduce both a new model tree
induction algorithm and a specific ensemble approach. Later,
estimated GPP for different biomes, focussing on global
median annual GPP derived using different prediction approaches. Covering a
larger number of variables, produced global flux products
at 0.5∘ spatial resolution for monthly average values of GPP,
terrestrial ecosystem respiration (TER), net ecosystem exchange (NEE), latent
heat (LE), and sensible heat (H). Their findings were confirmed by a
comprehensive cross-validation analysis using
FLUXNET
http://fluxnet.fluxdata.org/ (last access: 17 July
2018)
towers. In the latest study of , they investigate
the dependencies of changes in temperature and water availability on the
interannual variability in carbon fluxes both locally and globally using
their upscaled data products and process-based global land models.
There exist further upscaling approaches in the literature based on support
vector regression models that
estimate carbon fluxes on both regional and continental
scales. The work of
deals with estimating carbon fluxes for the
United States using data from MODIS and AmeriFlux. Only recently, a
systematic comparison of different regression algorithms for predicting
carbon and energy fluxes has been carried out by .
They were interested in the best prediction performances for estimating GPP,
TER, NEE, LE, and H, as well as net radiation at either 8-day or daily
temporal resolution. In their cross-validation analysis, they found that
prediction performance varies only slightly among different regression
algorithms from machine learning. However, they could show that accuracies
clearly differ between the individual fluxes, meaning that some fluxes are
harder to estimate than others, which is probably due to a lack of
information in the set of explanatory variables.
Upscaling flux tower measurements represents a “bottom-up” approach whereas
the “top-down” atmospheric CO2 inversions have been used for assessing
the net carbon exchange between the land and atmosphere. The atmospheric
inversions use measurements of CO2 in the atmosphere and prior
information, e.g., on anthropogenic emissions, together with wind fields and
a transport model to infer the land–atmosphere net CO2 flux. Due to the
relatively sparse atmospheric CO2 station network and uncertainties in
atmospheric transport, such inversion methods cannot precisely provide
located estimates but rather assessments for broad regions. The complementary
perspectives, strengths, and uncertainties in the “bottom-up” and
“top-down” approaches make it particularly appealing to bring these two
together. Therefore, we will compare our results for upscaling NEE with those
from an ensemble of atmospheric inversions by in
Sect. .
Today, global flux products feature, at best, a daily temporal resolution as
presented by . This is partly due to rapidly growing
computational issues in the training and production step scaling
quadratically with spatial resolution. In addition, consistent global
long-term products of driver data with hourly or higher temporal resolutions
are lacking or are not readily available. Upscaling half-hourly carbon and
energy fluxes raises previous upscaling approaches to the next level by
increasing the temporal resolution to subdaily timescales.
Furthermore, there is a need for a global data product of half-hourly fluxes.
Such a data product would allow for characterizing subdaily variations in the
diurnal cycles at places where no towers are currently installed. Please note
that we use the term diurnal cycle to name the full 24 h period of
48 half-hourly values per day. In the literature, this is sometimes rather
called a diel cycle, e.g., by
, which can be separated into a
diurnal pattern (daytime) and a nocturnal pattern (nighttime). However, in
the community of land–atmosphere exchange research the term diurnal cycle is
more common and we will use it throughout this paper. If we want to indicate
only those parts that correspond to light-driven processes like carbon
fixation in GPP, we explicitly call this daytime GPP (and use the term
nighttime to refer to the rest of the day). Furthermore, we use
diurnal courses as a synonym for diurnal cycles.
Characterizing typical subdaily flux patterns is critically needed for
certain satellite remote sensing applications. For example, the
interpretation of satellite retrievals of sun-induced fluorescence as proxy
for photosynthesis or integrated
atmospheric column carbon dioxide (XCO2) at certain overpass times
(usually around mid-day) requires consideration of strong diurnal variations
of biosphere–atmosphere carbon fluxes. Another research area where
half-hourly data products would be a crucial piece of information is
land–atmosphere feedback modeling studies. The derived products could allow
for checking the cross-consistency, since many processes governing
land–atmosphere interactions, e.g., related to the formation of heavy
rainfall or heat waves, in fact operate at subdaily timescales
.
In view of the need for global high-frequency flux data, we aim at increasing
the temporal resolution of data-driven carbon and energy flux products to
subdaily timescales by estimating half-hourly values at global
scale. We tackle the problem of
predicting diurnal cycles with half-hourly values globally for both carbon
and energy fluxes between biosphere and atmosphere by treating the upscaling
task as a large-scale regression problem. From the machine learning
perspective, the random forest regression framework serves as a basis for our
computations due to its good performance and suitable scaling properties with
respect to large data sets. We test two approaches for estimating half-hourly
GPP with random forest models and evaluate both of them using a
leave-one-site-out cross-validation strategy for a large set of FLUXNET
sites. We produce derived global products with 0.5∘ spatial and
half-hourly temporal resolution for GPP and NEE as well as for LE and H
covering the years 2001 to 2014. For the sake of clarity, some figures in
this paper only show the results obtained for GPP although similar plots can
easily be created for the other three fluxes that have been considered. Thus,
GPP serves as the running example throughout this paper. The choice of GPP
was somewhat arbitrary but it also constitutes the upscaled carbon flux that
received most attention in the past.
The following sections are organized as follows. First, we introduce the data
base that is used in our study by
describing both site-level and global forcing data
(Sect. ). Then, we explain the
methodological background (Sect. ) and
the algorithmic concept (Sect. ) of the
proposed upscaling approaches in detail. In Sect. ,
extensive evaluations and comparisons of the different upscaling strategies
are presented based on leave-one-site-out cross-validation, which validate the
proposed approach and the derived global products. Afterwards, we present the
empirical results at global scale in Sect. and highlight intrinsic
features of the new data sets. Finally, we discuss both our findings and
possible improvements for future applications (Sect. ).
The global data sets presented in this paper are freely available to any
interested user (see Sect. ).
Data sources
In this section, we shortly describe the two data sources we are using in our
studies. For learning the relationships between predictor variables and the
target fluxes as well as for the cross-validation experiments, we make use of
site-level data extracted at FLUXNET sites that are equipped with eddy
covariance towers (Sect. ). To perform global
upscaling of diurnal cycles, we require gridded data products of the
predictor variables at a global scale. The latter are described in
Sect. .
Site-level data
Fluxes at half-hourly resolution are currently only achieved by eddy
covariance instruments that provide local measurements and spatial extensions
are so far only possible by
deployment of those instruments on globally distributed towers. Based on
these in situ observations, we aim at predicting half-hourly fluxes globally
and therefore also rely on the data obtained by the eddy covariance method at
different sites. The eddy covariance method
has revolutionized the study of
land–atmosphere interactions by offering a means of continuously observing
net land–atmosphere fluxes of CO2, latent heat, and sensible heat
. By now, the flux towers are running for sufficient
time to enable studies about the interannual variability in land-surface
dynamics, but the temporal representativeness remains highly uneven
. In our studies, we rely on data from 222 FLUXNET eddy
covariance towers (see Appendix for a full list of
involved sites). All towers are typically equipped with a suite of comparable
micrometeorological devices; i.e., instrumentation and data outputs are
similar enough, such that training of machine learning methods on data from
multiple different sites is possible. Gross carbon fluxes can be derived
using different flux partitioning methods as described, for example, by
or , and here we rely on
the former method. In all our experiments, we only make use of measured
fluxes; i.e., no gap-filling has been applied and gaps in the
half-hourly flux data have simply been ignored (more information on the
distribution of data gaps can be found in
Appendix ).
As predictor variables, we use the ones selected by
Table 2 in the RS+METEO setup that they use for
estimating fluxes at daily resolution. For convenience, we have reproduced
this table and put it in Appendix . Besides the
plant functional type (PFT), the variables contain remote sensing data from
MODIS satellites and meteorological data either in situ measured at the flux
tower locations or from long-term time series of the ERA-Interim data set at
daily resolution. It should be noted that only the mean seasonal cycles (and
derived properties like amplitude, minimum, mean, and maximum) are taken into
account for the vegetation indices (normalized difference vegetation index –
NDVI, enhanced vegetation index – EVI) as well as for the normalized
difference water index (NDWI), the fraction of absorbed photosynthetically
active radiation (fAPAR), and the land surface temperature (LST). In
contrast, the actual values of air temperature, global radiation, potential
radiation, relative humidity, and of different water availability indices
have been used. For detailed descriptions, we refer to the corresponding
sections in the paper of
Sect. 2.1.3 and 2.1.4.
Global forcing data
In order to compute the global flux products at half-hourly resolution via
upscaling, we require the predictor variables mentioned in the previous
section at global scale, i.e.,
the variables of the RS+METEO setup from
Table in
Appendix , which has been reproduced from
Table 2 of . Concerning the remote sensing variables,
MODIS observations are used to compute mean values for each PFT and each day
aggregated to 0.5∘ spatial resolution. The distributions of each
PFT stem from the MODIS collection 5 global land cover product of
. Climatic data for the meteorological variables have
been obtained from
CRUNCEPv6
Data from
CRUNCEPv6 have been obtained via personal correspondence with Nicolas Viovy (email:
nicolas.viovy@lsce.ipsl.fr).
, which denotes a merged data product of monthly
observation-based climate variables at 0.5∘ spatial resolution from
the Climate Research Unit (CRU) at the University of East Anglia in Norwich,
UK and 6-hourly reanalysis data from the National Centers for Environmental
Prediction (NCEP) in Asheville and Silver Spring, USA.
Methodological background: random forest regression
Ensemble methods are powerful machine learning tools that combine the outputs
of many individual prediction models to obtain more accurate estimations for
a target variable. The random forest approach denotes a
typical example, which consists of a set of randomized decision trees.
Decision trees in general can be built for classification or regression
purposes and they are therefore also called classification trees or
regression trees. Multiple decision trees form a decision forest and learning
their decision rules typically involves some randomization, which leads to
the name randomized decision forest or short random forest. In the following,
the concepts of learning and testing randomized decision trees for regression
tasks are briefly summarized, because they denote the essential parts of
random forest regression. The reader who is familiar with the technical
details of random forest regression can skip this section and may directly
continue with the proposed upscaling approaches in
Sect. .
Besides the work of , detailed background
information about randomized decision trees and random forests can be found
in various machine learning textbooks, e.g., those
from chap. 3,
Sect. 14.3 and 14.4,
chap. 9 and 10, and
Sect. 16.2 and 16.4. Furthermore, many applications are
found in the area of computer vision and medical image analysis
. Specific use cases of random forests are also
land-cover classification , high-density biomass
estimations , and classification purposes in ecology
like the prediction of plant species presence among
others. Of course, previous work on the upscaling of fluxes at coarser
temporal resolutions also made use of random forests
or related model tree ensembles
. Hence, random forests and
tree-based machine learning techniques in general are applied in a broad
range of diverse research fields.
Randomized decision tree
Given a training set
X=x(i)∈IRD:i=1,2,…,N
of N samples with each sample x being a vector consisting of D
predictor variables x1,x2,…,xD and a corresponding real-valued
target variable y∈IR with observations
y1,y2,…,yN∈IR for the N training samples, the goal
is to find a set of rules that allow for predicting y based on x.
In the case of a decision tree, these rules are binary tests for individual
predictor variables with simple thresholds. A hierarchical tree structure is
built as shown in Fig. by selecting at each node i
a predictor variable di∈1,2,…,D and a threshold
ti∈IR. The estimate of a node i is the average value
y‾i of the observations computed from training samples that
reach this node. The first node of a decision tree, called root node,
contains all training samples, and hence the overall mean value
y‾1=1N∑n=1Nyn of observations yn from all
N training samples is an extremely coarse approximation that needs to be
refined depending on the constellation of the input variables x.
General structure of a decision tree for regression: binary splits
with thresholds for individual predictor variables will be used to navigate a
sample x to a leaf node that stores a continuous estimate for the
target variable.
Starting at node 1 in Fig. , the set of training
samples is partitioned into two subsets, represented by nodes 2 and 3, based
on the result of the binary test xd1≤t1. Both nodes, node 2 and
node 3, have associated predicted outputs y‾2 and
y‾3 that are computed as the average observation of samples that
reach the corresponding node. The split parameters d1 and t1 are
optimized such that the mean squared error for the training samples is
minimized given the respective predictions from node 2 or node 3. Such splits
are then computed for nodes 2 and 3 as well as for further derived nodes
until a stopping criterion is fulfilled. Typical stopping criteria are (i) a
split would create nodes with less than Nmin samples, (ii) the
variance of the observations from samples in a node is smaller than some
threshold σmin2, or (iii) a maximum depth dmax of
the tree is reached. The depth of a tree is defined as the largest distance
of a node to the root of the tree. Values of the parameters Nmin,
σmin2, and dmax can be changed to obtain either
smaller or larger trees, which allows for controlling the runtime of the
algorithm and the trade-off concerning generalization and overfitting.
It is usually the case that multiple stopping criteria are tested and if one
of them is fulfilled, the current node is not split but becomes a leaf node
that stores a final output prediction. Learning a decision tree therefore
consists of computing split parameters until only leaf nodes remain that are
not split any further (Fig. ). Hence, one
distinguishes between split nodes as the inner nodes of a tree and leaf
nodes, each of which contains the estimated output for any sample that
reaches this node.
To reduce overfitting to the training set, the learning process is carried
out in a stochastic manner by introducing several types of randomization.
Whenever split parameters need to be identified, only a random subset of the
D predictor variables is taken into account. Furthermore, only a fixed
amount of randomly chosen thresholds is tested. Both randomization techniques
also lead to reduced computation costs compared to exhaustive search. To
predict the output y* of a test sample x*, it is passed through
the tree according to the evaluation of the split functions at the inner
nodes starting at the root node. This is done until a leaf node ℓ is
reached, whose precomputed output y‾ℓ is assigned to
x. However, more accurate predictions can be achieved by considering
an ensemble of randomized decision trees.
Random forest as an ensemble of randomized decision trees
In his work about random forests, makes use of a
technique called bagging that he has introduced before .
Bagging is an acronym for bootstrap aggregating and
stands for aggregating predictions of individual models that have been
learned based on different sample sets built from the original training data
set. More precisely, individual sample sets are constructed by random
sampling with replacement from the original training set, which is commonly
referred to as bootstrapping. If the training set contains N samples, it is
possible that each of the sampled sets either contains also N samples
(which produces different sets with individual instances occurring several
times due to random sampling with replacement) or only a fraction ν of
the N samples. In both cases, the random subset selections introduced by
bagging additionally prevent overfitting to the training set. For bagging,
predictions from an ensemble of individual models are used and an ensemble of
randomized decision trees is called random forest, randomized decision
forest, or random decision forest (RDF). Each tree in the ensemble is learned separately and independent
from the other trees. Due to the involved randomization techniques during
learning of a single tree described before, different trees contain different
binary tests and provide different estimates for a single input sample
x. The individual predictions of each tree are then aggregated to
obtain a final result, which is typically carried out by simple averaging as
shown in Fig. .
Predicting the output y* of a sample x* with a
randomized decision forest is carried out by averaging the individual predictions obtained from the T
decision trees in the ensemble.
However, the number of trees Ntree is a hyperparameter whose value
needs to be chosen in advance but good assignments depend on various aspects.
Since pointed out that bagging leads to predictions
which are more stable compared to a single model, especially if the decision
function of the single model is highly instable with respect to the training
set, a larger number of trees is in favor of higher stability. On the other
hand, more trees are causing higher computational costs during both learning
and testing. In addition, a saturation effect for the prediction accuracy can
typically be observed for an increasing number of trees. Hence, accuracies
obtained by cross-validation for different numbers of trees can help to
identify this saturation and a proper value for Ntree.
Methods for upscaling diurnal cycles
The problem of upscaling diurnal cycles of carbon and energy fluxes can be
formulated as a large-scale regression task, i.e., estimating half-hourly
fluxes for every grid cell of the globe based on a set of predictor
variables. These predictor variables typically encode climate conditions or
Earth observations obtained from remote sensing at the corresponding spatial
positions. However, the temporal resolutions of variables can be different,
not only between the target flux (half-hourly) and a predictor variable
(e.g., daily) but also among different predictor variables (e.g., daily and
half-hourly). Therefore, two
prediction approaches for upscaling diurnal cycles are presented in
Sect. and ,
respectively, which account for this mismatch of temporal resolutions.
Although both approaches can be equipped with any regression algorithm, we
have decided to use the RDF as a nonlinear method, which has been summarized
in the previous section. The main reasons for this choice are the fast
learning and testing algorithms, because the upscaling tasks involve a huge
number of samples such that learning nonlinear kernel methods for regression
like Gaussian processes are impractical due to both
memory demand and computation time. Furthermore,
have shown in their cross-validation experiments that the accuracies for
estimating fluxes vary only slightly among different machine learning
methods.
Visualization of the first prediction approach: an individual RDF
regression model has been learned
for each half hour of a day by just using training data from the
corresponding half hour. Here, the predictions of half-hourly fluxes for a
single day are visualized. The predictor variables are passed to the
individual RDF regression models indicated by the arrows above the RDF
models. Each RDF model computes an output for the corresponding half hour,
which is shown by the arrows below the RDF models, such that the diurnal
cycle is estimated by a conjunction of 48 different predictions from 48
different regression models. Note that this approach allows for predicting
diurnal cycles only based on predictors with daily resolution (by ignoring
the vertical colored bars in the upper part). However, predictors with
half-hourly resolution can also be incorporated (by adding the corresponding
half-hourly values indicated by the vertical colored bars).
First prediction approach: an individual regression model for every half hour of the day
Recall from the beginning of Sect. that the
two main challenges for upscaling diurnal cycles to global
scale are the huge amount of data
which needs to be handled as well as the mismatch of temporal resolutions
between predictor variables and the target fluxes. The first approach for
predicting diurnal cycles has the advantage that it allows for using only
predictors of daily temporal resolution. This is very important, because
daily (average) values are often less noisy with respect to measurement noise
and the availability of daily values is much higher compared to half-hourly
values, especially when considering global products with values for every
grid cell. Furthermore, variables derived from remote sensing are often
limited to daily temporal resolution meaning that they are not more frequent
than once per 24 h. Therefore, the first prediction approach involves
learning an individual regression model for each half hour of the day and as
indicated at the beginning of Sect. , RDF
regression models are used for handling large-scale data. A schematic
overview for a single day and a diurnal cycle of GPP is shown in
Fig. .
Even if one uses only predictor variables of daily temporal resolution which
can be treated as constant for the whole day, different values of the target
flux for different half hours of the day can be estimated. The reason is that
the 48 different RDF models are learned with different values for the target
output variable y, although the same values for the predictor variables
x are used. For example, an RDF model that is learned for a half hour
during night only covers the rather small range of observations y that can
be observed at this time, while the range of observations around noon is
typically much larger, especially during the growing season. Hence, the 48
RDF models and their estimated outputs differ only because of different
observations y that are provided during learning together with the same set
of samples X. Of course, it is also possible to incorporate
predictor variables at half-hourly temporal resolution, which would directly
fit to the resolution of the target flux. Such predictor variables could
further enhance the distinction of individual half hours of a day and could
lead to more accurate estimations. However, they are optional and not
required for this prediction approach as indicated in
Fig. .
Visualization of the second prediction approach: a single RDF
regression model is able to predict the flux at every half hour of the day if
at least one predictor variable has a half-hourly temporal resolution (such
as the potential radiation Rpot). This is different from the first
prediction approach shown in Fig. ,
because the whole training data of all half hours are used to
learn a single RDF regression
model. The upper part of the figure shows the composition of the input data
(predictor variables) that are required for predicting the half-hourly values
of a single day. For each half hour, the corresponding values of the
predictors with half-hourly resolution are added to the predictors with
coarser temporal resolution (indicated by the vertical color bars). We then
have 48 distinct samples due to the distinction from the half-hourly
predictor variables and these samples are delivered to the single RDF
regression model indicated by the arrows in the upper part. For each sample,
the RDF model calculates an output value (lower part) that denotes the
estimate for the corresponding half hour within the diurnal cycle.
Second prediction approach: a single regression model suitable for all half hours of the day
In contrast to the first prediction approach, the second approach only uses a
single regression model that is able to estimate different values for
different half hours of the same day. It is then necessary that the
distinction between these half hours is somehow encoded in the predictor
variables, which is not the case if only predictors of daily resolution are
incorporated. Therefore, this approach requires at least one predictor
variable at half-hourly temporal resolution. Fortunately, the potential
radiation (Rpot) can be calculated globally at half-hourly
resolution, because it only depends on the time as well as the solar angle
that is defined given the spatial position via latitude and longitude. Thus,
the second approach with a single model, as visualized in
Fig. , is therefore also applicable for
upscaling diurnal cycles to global scale. Again, we make use of an RDF regression model due to its
large-scale capabilities.
In addition, besides the potential radiation, its first-order temporal
derivative can also be incorporated as an additional half-hourly predictor.
This allows for a distinction between
morning and afternoon via the sign of the derivative as well as for the
distinction between day and night. The latter is achieved because
Rpot is zero for many consecutive hours during night leading also
to a value of zero for its derivative in contrast to nonzero values of the
derivative in the daytime. For all our computations, we have always included
this derivative in the case when we also used Rpot. The nice property of this approach is
that information about the physical relationships between the predictor
variables and the fluxes can be shared among different half hours during
learning of the single regression model, which is not the case for individual
models as mentioned in the previous section. This second prediction approach
therefore seems to be more plausible from a physical perspective, because the
distinction between different half hours of the day is made based on the
data, e.g., (potential) radiation, and not enforced by learning independent
regression models for each half hour.
Although meteorological variables such as air temperature or vapor pressure
deficit (VPD) as well as incoming radiation are also potential candidates for
predictors that encode subdaily variations in the fluxes, they are currently
only available with a half-hourly resolution at individual sites, e.g., also
measured at eddy covariance towers. Due to the missing half-hourly
meteorological data products at a global scale, it is not possible to use these
information for the global upscaling. However, since we are interested in
whether such data products could be beneficial for upscaling diurnal cycles,
we use the corresponding site-level data in our cross-validation analysis to
get further insights. Hence, meteorological variables measured at the eddy
covariance towers of FLUXNET can still be used for validating the upscaling
approaches and evaluations of cross-validation experiments are presented in
the next section.
Assessing different upscaling strategies with leave-one-site-out cross-validation
The global products presented in this paper cover diurnal cycles of four
fluxes: GPP, NEE, LE, and H. For each of these fluxes, we have consistently
performed cross-validation experiments but the results presented in the
following only consider GPP as a running example. We have decided to apply
RDF models for regression due to its efficient training and testing
algorithms, even in the case of large-scale data, as well as its good performance
for upscaling daily mean values of GPP . Each RDF was
trained with 100 randomized decision trees, because we observed a saturation
effect for the prediction performance in preliminary experiments when
increasing the number of decision trees. Further parameters have been set to
its default values in Matlab's TreeBagger function, e.g., a minimum leaf size
of five samples, since we hardly observed any changes in the overall
performances when varying the parameter settings. Performances are measured
using the Nash–Sutcliffe modeling efficiency as well as
the root-mean-square error (RMSE) based on a leave-one-site-out
cross-validation scheme.
The Nash-Sutcliffe modeling efficiency, from now on simply called modeling
efficiency, has been introduced by Nash and Sutcliffe in the context of river
flow forecasting but it is often also used as an evaluation criterion in
other applications that involve the prediction of variables, especially in
related upscaling tasks . A perfect match
between observations and predictions would be reflected by a modeling
efficiency of 1, whereas always using the mean of all observations as an
estimate translates to a modeling efficiency of 0 and negative modeling
efficiencies indicate that the learned model performs worse than always
assigning the mean of the observations. The modeling efficiency is related to
the fraction of explained variance and to the coefficient of determination
(R2). While values close to 1 are preferred for the modeling efficiency,
RMSE is a non-negative measure with an optimal value of 0.
The motivation for the leave-one-site-out evaluation as a special case of
cross-validation is twofold. First, we want to evaluate regression models
that have been learned from as many observations as possible and based on
training sets that are most similar to the training set that will be used to
compute the global products, which will incorporate all the available data
from all FLUXNET sites. Second, we intend to mimic a realistic scenario most
similar to the upscaling task by predicting fluxes at locations where no
training data has been taken from. As a consequence, we predict fluxes at one
FLUXNET site using a regression model learned with all observations from all
the remaining FLUXNET sites. After doing this for each individual site, we
concatenate all site-specific predictions to form a long vector of
predictions that can be compared to the corresponding observations measured
at the corresponding sites. This allows for a general evaluation of the
prediction approaches in a site-independent manner.
Overview of experiments
We start with a short overview of the experiments that have been conducted in
order to clarify our ideas and motivations behind them. In
Sect. , we compare the two different
prediction approaches for upscaling diurnal cycles that have been introduced
in Sect. . Furthermore, we focus on comparing
different sets of predictor variables, e.g., the effect of meteorological
variables at half-hourly resolution on the prediction performances.
Evaluations of the prediction performance for monthly average diurnal cycles
derived from the half-hourly values are shown in
Sect. . These average diurnal
cycles per month nicely summarize the fluxes over a longer time period (one
month) by still keeping a half-hourly pattern that allows for monitoring
subdaily variations. In addition, averaging diurnal cycles for a specific
month removes noise in the individual half-hourly measurements and reduces
the effects of day-to-day variability, e.g., caused by cloud coverage, which
allows for comparing the main characteristics of the observations and the
predictions for the selected month. The evaluations of monthly average
diurnal cycles play an important role with regard to our provided data
products, since we also prepare derived products that only contain these
monthly average patterns. With two additional experiments presented in
Sect. , we want to demonstrate that the
quality of our achieved predictions is not inherently limited by the
presented upscaling approaches but rather by missing site-specific
information and latent driving forces that are not encoded in the set of
predictor variables that has been used. This is not a specific problem of
upscaling diurnal cycles of fluxes at half-hourly resolution but a general
challenge for all upscaling approaches that deal with carbon and energy
fluxes, also at coarser timescales.
Improved predictions by using half-hourly meteorological data
Prediction performances (a modeling efficiency;
b RMSE in
µmol m-2 s-1) for individual
half-hourly values of GPP depending on different sets of predictor variables
are shown. We compare our two proposed prediction approaches (individual RDF
models and single RDF model) and also include the results of a gap-filling
algorithm for comparison. Looking at the results of all sites as well as at
site-specific performances, we observe that meteorological predictor
variables at half-hourly resolution clearly improve the accuracies of the
estimations. Site acronyms: Canada Manitoba Black Spruce (CA-Man), Germany
Hainich (DE-Hai), France Puechabon (FR-Pue), Italy Castelporziano (IT-Cpz),
USA Mississippi Goodwin Creek (US-Goo), USA California Vaira Ranch (US-Var).
In the following, we compare the results of our presented prediction
approaches for half-hourly GPP depending on different sets of predictor
variables, which have been obtained by using the leave-one-site-out strategy
explained in the beginning of Sect. . As the core for
all sets of predictors, we include those variables that have been used for
upscaling daily mean GPP values by . In fact, we use
exactly the same set of predictors that
corresponds to the RS+METEO setup which has been
defined by
Table 2 and this table can also be found in
Appendix . Given the explanations of the first
prediction approach in Sect. with
individual regression models for each half hour, we can directly use these
predictor variables for estimating half-hourly GPP values. However, we also
added the potential incoming radiation Rpot at half-hourly
resolution to encode subdaily variations in the predictors as well as its
first temporal derivative to distinguish between morning and afternoon.
Furthermore, we have tested a third set of predictors by additionally
incorporating meteorological variables with half-hourly resolution measured
at FLUXNET tower sites. The added variables are air temperature, vapor
pressure deficit, and incoming global radiation. In a nutshell, the three
sets of predictor variables consist of (i) daily predictors, (ii) daily
predictors + half-hourly Rpot, and (iii) daily
predictors + half-hourly Rpot+ half-hourly meteorological
predictors, besides static predictors like PFT that are used in all three
cases. The second and third set are also used in the experiments for the
second prediction approach that includes only a single regression model,
because half-hourly information is encoded in some of the predictor
variables.
In Fig. , we have visualized the results for
all sites as well as for selected FLUXNET towers. Only for comparison to the
leave-one-site-out experiments, we also included the modeling efficiency of a
gap-filling algorithm as a potential upper bound
for our predictions. In fact, each measured value is also estimated by a
gap-filling algorithm that makes use of flux measurements at the same site
under similar climate conditions and hence provides only a theoretical
baseline, because it can not be applied for predicting fluxes at locations
without any observations. First, we focus on the leftmost group of bars in
Fig. , which shows the modeling efficiencies
for all sites. Looking at the results for the first prediction approach with
individual models for each half hour, including half-hourly Rpot
only slightly improves the average performance (0.67 compared to 0.66),
which is probably also caused by the stochastic nature of the RDF learning
algorithm. However, including the meteorological predictors at subdaily
temporal resolution leads to an increase in the performance to 0.70
modeling efficiency. A similar improvement can be observed for the second
prediction approach with a single model for all half hours of the day,
because the modeling efficiency increases from 0.67 to 0.71 when
including half-hourly meteorological data. This highlights that varying
subdaily meteorological conditions has a clear impact on predicting the
diurnal cycles of GPP fluxes.
On the one side, half-hourly Rpot has almost no influence on the
accuracy of the predictions but on the other side, it allows for applying the
second prediction approach with only a single regression model. Hence, it may
seem more natural from a physical perspective to distinguish individual half
hours of the day by the provided half-hourly Rpot rather than
enforcing the distinction by separately learned regression models. Comparing
both prediction strategies, they achieve similar performances when using the
same set of predictor variables. Since it is more convenient from a technical
perspective to only handle a single regression model instead of 48 different
models, the evaluations in the following sections will focus on the second
prediction approach with a single RDF model that is suitable for predicting
values at every half hour of the day. It is interesting to note that relative
performance differences between the two prediction approaches and among the
different sets of predictor variables look very similar when considering
single sites only. In all our cross-validation experiments, the best
prediction accuracies are always achieved by including half-hourly
meteorological variables in the set of predictors. However, absolute
performance values vary among sites. As shown in
Fig. , the accuracies at the sites CA-Man and
DE-Hai are between 0.80 and 0.90 modeling efficiency, whereas lower
performances (between 0.60 and 0.80) have been achieved for predicting
GPP at FR-Pue, IT-Cpz, and US-Goo. Moreover, the fluxes at US-Var seem to be
very difficult to estimate, since only modeling efficiencies between 0.20
and 0.30 were obtained.
Fingerprint plots of half-hourly GPP fluxes estimated for US-SO2 in
2004 with leave-one-site-out cross-validation which show that short-term
fluctuations on subdaily timescales are captured better when half-hourly
meteorological predictors have also been included (a) compared to
only using half-hourly Rpot(b). The difference of the
first two plots (c) also emphasizes this observation.
Some example sites with average diurnal cycles for different months
comparing two prediction approaches with the observations.
To further highlight the difference in the predictions when half-hourly
meteorology is encoded in the driver variables, we visualize all half-hourly
estimations over one year at a specific site using fingerprint plots. A
fingerprint in this context is a plot with 365 rows corresponding to 365 days
of a year and 48 columns corresponding to 48 half hours of each day such that
one fingerprint contains all half-hourly values of a whole year and shows
characteristic patterns for the selected site, e.g., length of the growing
season. In Fig. , the estimations of
half-hourly GPP with and without half-hourly meteorological predictors are
shown in two individual fingerprint plots and their difference is indicated
in a third plot. As expected, the predictions based on half-hourly
meteorology contain much more short-term fluctuations during single days,
whereas smoother estimations are obtained when only half-hourly
Rpot is used as a subdaily driver. This can also be observed from
the difference of the two fingerprints. Hence, half-hourly meteorological
predictor variables are required to better capture high-frequency changes in
the fluxes on subdaily timescales. In the following, we take a closer look at
average diurnal cycles per month.
Modeling efficiencies (and RMSE in
µmolm-2s-1) for the predictions
of all sites obtained from the leave-one-site-out experiments are summarized
and here we differentiate between comparing all individual half-hourly values
with the observations and only looking at monthly average diurnal cycles.
Prediction performances for monthly average diurnal cycles of GPP
are shown in the same way as the accuracies for all half-hourly values in
Fig. (a modeling efficiency;
b RMSE in µmolm-2s-1).
Analyzing average diurnal cycles per month
For visual inspection purposes, it is useful to look at average diurnal
cycles for individual months at specific sites. Example plots are shown in
Fig. . They show that our predictions are able to
produce the typical shapes of diurnal courses which are in line with
corresponding observations. For the depicted predictions, only observations
from other sites have been used to learn the regression models. It is
important to note that averaging diurnal cycles within a month reduces noise
in the observations as well as in the predictions of a single day, but also
smoothens high-frequency short-term fluctuations, e.g., due to (partial)
cloud coverage, and yet decreases the influence of day-to-day variations.
Hence, these mean diurnal courses are more stable and an evaluation of
averaged predictions with respect to averaged observations for all sites
leads to larger modeling efficiencies compared to those reported in the
previous section. An overview of modeling efficiencies when comparing all
half-hourly values versus only looking at the average diurnal cycles is given
in Table .
Average diurnal cycles of two sites showing the problems with
seasonal droughts. The error in the prediction of the half-hourly fluxes
increases during hot and dry summers for both sites, FR-Pue and IT-Cpz.
In fact, modeling efficiencies for monthly average diurnal cycles increase on
average across all sites to a range between 0.78 and 0.80 depending on
the set of predictor variables with the best results being accomplished again
by incorporating half-hourly meteorological data. This can also be observed
from Fig. , which is organized in the
same way as Fig. but contains the achieved
modeling efficiencies for comparing monthly average diurnal cycles of
observations and predictions. For the monthly mean diurnal courses, the
difference between only using half-hourly Rpot or also including
half-hourly meteorology is not so large anymore compared to the evaluations
for all half-hourly values. This holds for both the overall accuracies for
all sites as well as for single selected sites. As previously mentioned, the
reason is that averaging fluxes within a month reduces the effect of
short-term fluctuations on subdaily timescales. Therefore, if one is only
interested in monthly average diurnal cycles, the results obtained by using
daily predictors and half-hourly Rpot are only slightly worse
compared to including half-hourly meteorology and for some sites, the
prediction performances are even on the same level of accuracy. This is
important to know, since we also provide a derived product from our global
half-hourly fluxes that contains the monthly average diurnal cycles globally
at the same spatial resolution (Sect. ).
However, the average diurnal cycles can also be used to identify potential
problems of the predictions. In Fig. , mean
diurnal courses of several months at the sites FR-Pue and IT-Cpz are shown.
It can be observed for both sites that the averaged observations are lower in
the summer months compared to the corresponding predictions. In other words,
the regression models overestimate GPP during these months. We believe that
this is caused by the fact that our current prediction models are not able to
cope with seasonal droughts, which is not a specific problem of diurnal
upscaling but a challenge that every upscaling approach for carbon and energy
fluxes needs to tackle. Although the observations show decreased productivity
due to drought stress in summer, the regression models still estimate large
amplitudes of the diurnal cycles, i.e., a larger productivity. One reason for
this behavior could be the insufficient characterization of water
availability that is present in the set of predictor variables. Currently, we
plan to investigate this issue in further research. In the following, we show
that our current sets of predictor variables are lacking some site-specific
information, probably not only with respect to water storage capacities.
Are we missing (site-specific) information in the predictors?
Comparison between leave-one-site-out (a) and
leave-one-month-out (b) at IT-Cpz. It can be observed that
site-specific training in the leave-one-month-out setup reduces the
prediction errors during seasonal droughts. Thus, the drought effects only
lead to problems when training across sites and predicting fluxes in the
leave-one-site-out setup or for the upscaling when fluxes are estimated at
locations where no towers exist.
In order to gain any insights into whether site-specific information is
currently not well represented in the predictors, we have conducted two
auxiliary experiments. During the first experiment, we additionally estimate
GPP fluxes at each site in a leave-one-month-out setup and compare the
resulting predictions with those of the leave-one-site-out setup. For the
leave-one-month-out estimations, we learn and test regression models for each
month at each site separately. Furthermore, each regression model for each
month is only learned with data from the same
site but measured in different months (and years). Hence, the regression
models are highly site-specific, since only correspondences between predictor
variables and GPP fluxes at a single site are used and predictions are made
at the same site but in a different time period. As a result, we have
observed improved flux estimations, which is shown exemplarily in
Fig. for IT-Cpz. It can be clearly seen that
the gaps between averaged observations and averaged predictions are getting
smaller and mostly almost disappear; i.e., the predictions match the
observations much better in the leave-one-month-out setup. In terms of
modeling efficiency, the performances increase to a range between 0.75 and
0.79 when comparing all individual half-hourly predictions from the
leave-one-month-out setup at all sites with the observations (best
performance with leave-one-site-out is 0.70). Regarding the comparison of
averaged predictions and averaged observations within each month as presented
in the previous section, the leave-one-month-out setup leads to modeling
efficiencies between 0.87 and 0.89. This is clearly larger than the
results of the leave-one-site-out-experiment (best performance: 0.80).
Table allows for a direct comparison of the
results from the leave-one-site-out and the leave-one-month-out experiments
using both prediction approaches with the best set of predictor variables,
i.e., daily predictor variables, half-hourly Rpot, and half-hourly
meteorological variables.
Improvements in the initial estimations (a) at FR-Pue can
be observed when using daily GPP as an additional daily predictor if it would
be available a priori (b). Hence, the problems with seasonal drought
effects would be greatly reduced in the leave-one-site-out setup for every
half hour of the diurnal cycle in the case when an accurate estimate of the daily
average value is given.
This table also contains the prediction performances obtained from a second
experiment, in which we have used the daily GPP as an additional daily
predictor for our regression models in the leave-one-site-out setup. Of
course, this is only possible in the cross-validation analysis where we
actually have the daily averages of GPP, but the following evaluation reveals
interesting insights. Using the daily average GPP basically incorporates
information about the amplitudes of the diurnal cycles, hence drought effects
of reduced productivity can directly be observed in this additional predictor
variable. First of all, it can be seen in
Fig. that using the daily GPP as an
additional predictor clearly improves the predictions at FR-Pue during summer
months. Especially the decreased productivity in July and August 2005 can be
nicely predicted by the regression models. Since the daily GPP as an
additional predictor constrains the size of the peak in a diurnal cycle, the
predictions become much more powerful and the characteristic shapes of the
diurnal cycles can be produced. The modeling efficiencies are even larger
than those obtained with the leave-one-month-out setup. They are in the range
of 0.83 to 0.87 for all half-hourly values at all sites, which is
comparable with the performance of the gap-filling algorithm that has been
included as an additional reference in Fig.
as well as in Table . The gap-filling also achieves a 0.87
modeling efficiency; i.e., the upper performance limit shown as a green bar
in the leftmost group in Fig. can be
obtained by including the daily GPP as an additional predictor. Regarding
monthly averaged diurnal cycles, a modeling efficiency of up to 0.94 is
obtained by the regression models that use daily GPP as an additional
predictor, while gap-filling
reaches 0.93. This is also summarized in Table .
Comparing modeling efficiencies (and RMSE in
µmolm-2s-1) of the two
auxiliary experiments (leave-one-month-out setup and including the daily GPP
as an additional predictor in the leave-one-site-out setup) to the best
performances obtained with the leave-one-site-out experiments by using daily
predictor variables, half-hourly potential radiation, and half-hourly
meteorological variables.
From this experiment, we can conclude that the problems for predicting
diurnal cycles of GPP are mainly caused by the lack of estimating the daily
mean GPP properly. If the daily mean is given, predictions of half-hourly
values are much more accurate. Hence, the main problems for the upscaling of
half-hourly fluxes are not related to producing the right shapes of the
diurnal courses, but turn out to be problems of estimating the correct
amplitudes. These are then the same problems as for upscaling daily average
values (or fluxes at coarser timescales) and are not introduced by the step
of going to a larger temporal resolution in terms of half hours.
Key insights from the cross-validation experiments
In this section, we want to shortly summarize the main findings from our
cross-validation experiments. First, we have seen that it does not really
matter which of the two proposed prediction approaches we are using, since
prediction performances hardly differed between the single model approach and the
individual model approach. We
prefer to use the single model approach, because it seems to be more
plausible from a physical perspective to make distinctions between half hours
of a day by the information encoded in the predictor variables and
half-hourly Rpot can always be used for this purpose. Second,
including half-hourly meteorological information in the predictors clearly
helps to improve the prediction performances for fluxes on the half-hourly
timescale. However, for monthly average diurnal cycles the differences are
not so prominent anymore and estimations based on half-hourly Rpot
as the only predictor at half-hourly resolution may be sufficient for
analyzing the monthly patterns. Third, we have shown that the main problem
for upscaling half-hourly fluxes is not the fact that we increase the
temporal resolution, since we are able to reproduce the characteristic
subdaily patterns. Moreover, we are lacking additional information in the
predictors that encode site-specific characteristics as well as certain
special conditions like seasonal droughts. This currently prevents us from
obtaining the correct day-to-day variability and also, in the end, the
correct interannual variability.
However, these are also problems that need to be tackled when an upscaling of
carbon and energy fluxes at coarser timescales is
considered . In the following section, we summarize
the results from our cross-validation experiments for all the four fluxes
(GPP, NEE, LE, H) with the setup that has been used to compute the global
half-hourly data products.
The selected approach for computing the global products
While the previous sections validate the presented prediction approaches and
point to potential problems in the estimation of half-hourly fluxes, we also
decided to produce the first global products of half-hourly GPP and
NEE,
as well as LE and H that will be described in the next section. So far, the
analyses have shown that best predictions are obtained by incorporating
meteorological variables at half-hourly resolution, but such data products
are not available at a global scale. Therefore, we have computed the global
products only based on the daily predictors of the RS+METEO setup from
Table 2, which can also be found in
Appendix , as well as using half-hourly values
of Rpot and its first temporal derivative. The used data sources
have been described in Sect. and the set of daily
predictors varies between carbon and energy fluxes as indicated within the
aforementioned table.
Furthermore, we have decided to use the second prediction approach
(Sect. ) by learning one single regression
model that is suitable for all half hours of the day. For us, it seems more
natural from a physical perspective to distinguish between different half
hours of a day by (potential) radiation as an additional variable rather than
enforcing the distinction with individual models for each half hour as it is
done in the first prediction
approach (Sect. ).
Prediction performances in terms of modeling efficiency (and RMSE in
µmolm-2s-1) are estimated from the leave-one-site-out
cross-validation experiments with the setup that has been used to compute the
global half-hourly products for the four fluxes.
GPPNEELEHModeling efficiencies related to all individual half-hourly values0.67(3.94)0.61(3.66)0.72(37.96)0.77(45.18)Modeling efficiencies related to monthly average diurnal cycles0.78(2.95)0.76(2.66)0.83(26.12)0.86(29.40)
In Table , we report the corresponding prediction
performances from the leave-one-site-out cross-validation experiments for
this setup, i.e., for the selected set of predictor variables and the single
regression model approach. The modeling efficiencies for both individual
half-hourly values and monthly average diurnal cycles are stated. Comparing
these values, we observe that the accuracies for predicting energy fluxes are
higher compared to those for the carbon fluxes. Half-hourly values of the
sensible heat flux can be best estimated by achieving a modeling efficiency
of 0.77 across all sites. On the other hand, net ecosystem exchange has
only been predicted with a modeling efficiency of 0.61. This performance is
lower compared to the one for gross primary production (0.67), probably due
to missing information in the predictor variables for the respiration
component of NEE. For all four fluxes, the modeling efficiencies are
higher when comparing monthly average diurnal cycles of observations and
predictions. The main reasons, as also mentioned in
Sect. , are the reduction of noise
and the smoothing of short-term fluctuations at subdaily timescales due to
the averaging. In the following section, we present the global half-hourly
flux products that have been calculated with the upscaling approach and the
setup of this section.
Global flux products with half-hourly resolution
For each of the four fluxes (GPP, NEE, LE, H), we have learned a single
regression model for all half hours based on all available half-hourly values
of the corresponding flux at the 222 FLUXNET sites listed in
Appendix , i.e., one regression model for GPP, one for
NEE, one for LE, and one for H. These models are then used to compute
half-hourly fluxes globally with 0.5∘ spatial resolution and
continuously from 2001 (1 January) to 2014 (31 December) using global forcing
data described in Sect. . As mentioned in the
previous section, we have used the daily predictors of the RS+METEO setup
from Table 2 as well as half-hourly values of
Rpot and its first temporal derivative. Note that this table has
been reproduced in Appendix for faster
reference. Furthermore, it should be noted that the global products have been
initially calculated such that they are tiled by PFT. The final flux for each
point in space and time has then been determined as a weighted sum depending
on the percentage of each PFT to be present in the corresponding grid cell.
When looking at annual sums of the half-hourly data products, we observe that
these sums are pretty constant among the different years for the individual
fluxes. On average, we get 125.94PgC for GPP and
-21.42PgC for NEE as well as 182.22ZJ for LE and
144.79ZJ for H.
In addition to the provided half-hourly data, we also offer derived products
containing the monthly average diurnal cycles of the four fluxes for the 14
years that are covered by the half-hourly product. For the potential user of
the data, it will be much more convenient to directly obtain the monthly
average diurnal cycles compared to downloading the much larger half-hourly
data product and computing the monthly averages afterwards. Furthermore, the
monthly average diurnal cycles are more robust, which has also been shown by
larger modeling efficiencies in the experimental evaluations, e.g., as listed
in Table . Since only daily
predictor variables and half-hourly Rpot are used to estimate the
half-hourly fluxes, short-term fluctuations on subdaily timescales due to cloud
cover and other effects can not be captured by the current version of the
product. Therefore, also day-to-day variations may not be represented
accordingly. However, the averaging to create monthly mean diurnal cycles
reduces the impact of these factors and additionally smoothens errors due to
observation noise. As a consequence, we recommend to primarily use the
monthly average diurnal cycles because of larger robustness and stability. In
the following, we show some characteristics of the computed global flux
products at half-hourly resolution, which can only be calculated due to the
subdaily timescale.
The global maps show estimated values for half-hourly
GPP (a), NEE (b), LE (c), and H (d) on
14 June 2014 at 13:00 UTC. In addition, fingerprints for selected grid cells
are used to visualize half-hourly values for each day of the year. The dot in
each fingerprint marks the value that is shown in the global map. Note that
the fingerprints display different extensions of the growing season in
different regions and the global maps allow for distinguishing between
daytime (e.g., in Europe and Africa) and nighttime (e.g., in East Asia,
Australia, and in the western parts of North America).
Maximum diurnal amplitudes of GPP within a month are shown for June
2014 (a) and December 2014 (b). Differences between summer
and winter for both the Northern Hemisphere and the Southern Hemisphere as well as
(almost) constant productivity in tropical regions can be observed from both
maps. Note the logarithmic color scale.
Global maps and fingerprints
Cutouts of the global products are visualized in
Fig. , where we have selected
14 June 2014 at 13:00 UTC in the time domain. Global maps of GPP and NEE are
shown in the top row of Fig. and one
can nicely distinguish daytime from nighttime for individual regions around
the world. Furthermore, selected locations are highlighted and all
half-hourly values of the year 2014 for these grid cells are summarized in
fingerprint plots, which allow for identifying different characteristics at
the corresponding places due to the different patterns in these plots. The
fingerprints provide a nice overview of the half-hourly fluxes over a whole
year and different lengths of the growing season as well as varying lengths
of the day (time between sunrise and sunset) can directly be observed.
Corresponding maps for LE and H with fingerprint plots for the same locations
are shown in the bottom row of Fig. .
Larger values of LE compared to H at this specific point in time can be
observed in western, central, and eastern Europe as well as in the tropical
regions of Africa, whereas it is vice versa on the Iberian
Peninsula as well as in the
northern and southern regions of Africa.
Maximum diurnal amplitudes within a month
Besides the fingerprint plots summarizing a whole year of half-hourly values
for a specific location, it is also possible to compute diurnal amplitudes
for each grid cell from the global products. We again picked GPP acting as an
example for all the fluxes and determined maximum diurnal amplitudes within
each month. In Fig. , the maximum diurnal
amplitudes of GPP are visualized for June and December 2014 with a
logarithmic color scale. These months have been chosen to indicate
differences between summer and winter. The biosphere at the Northern
Hemisphere is quite active in June showing large amplitudes, whereas maximum
amplitudes are close to zero at most of the grid cells of the Northern
Hemisphere in December. In the tropics, amplitudes of GPP do not vary much
between June and December with values around
30 µmolm-2s-1. As expected, the behavior in the
Southern Hemisphere is opposite to the Northern Hemisphere; i.e., the
productivity in the Southern Hemisphere is higher in December compared to
June.
Maximum half-hourly fluxes
Furthermore, we have been interested in the maximum flux at each spatial
position. These statistics have been calculated among all the years 2001 to
2014 to produce a single map of maximum half-hourly values for each flux. The
results are shown in Fig. . Those maximum
values denote the capabilities of each flux at each grid cell. For GPP, the
hot spots with maximum capacities are in the corn belt of the USA, in eastern
China, and in the tropical regions. Largest values of NEE are obtained
in the tropics as well, especially in the Amazon. Regarding the energy
fluxes, it is not so easy to identify single hot spot regions since large
values of LE or H are widespread. However, distinct spatial patterns can be
observed in all maps of the maximum fluxes.
Maximum half-hourly values of GPP (a), NEE (b),
LE (c), and H (d) during the years 2001 to 2014 are shown
for each grid cell.
Comparison with ensemble of atmospheric inversions
Finally, we want to compare our global product of NEE with an ensemble of
atmospheric CO2 inversions that contains CarbonTracker
, CarbonTracker Europe , the Jena
CarboScope inversion scheme (s99_V3.6) from , and
MACC-II of that has also been used by
. We consider temporal aggregations with a monthly
resolution and have computed the mean seasonal cycle (MSC) from the years
2001 to 2010 for each grid cell. In the spatial domain, we have then
performed an averaging step according to the 11 TransCom land regions
: North American Boreal, North American Temperate, South
American Tropical, South American Temperate, Northern Africa, Southern
Africa, Eurasian Boreal, Eurasian Temperate, Tropical Asia, Australia, and
Europe.
Comparison of our upscaled global NEE fluxes (red lines) with an
ensemble of atmospheric CO2 inversions (blue lines) considering the mean seasonal cycle (MSC) for
11 TransCom regions estimated from the years 2001 to 2010. For the ensemble
of atmospheric inversions, the average MSC among the ensemble members is
shown together with shaded regions spanned by minimum and maximum values
within the ensemble.
The upper panel in Fig. contains the MSC of
our NEE product for the 11 TransCom land regions as well as the average MSC
from the ensemble of atmospheric inversions. For the latter, the shaded
regions are spanned by minimum and maximum values per month. Since it is
known that upscaling methods tend to overestimate the carbon uptake of the
biosphere (too large carbon sink) compared to atmospheric inversions, we have
also subtracted the mean value from all curves in the upper panel and show the
differences from the corresponding mean values in the lower half in order to
see whether the patterns in the MSC are matching between the upscaled product
and the inversions. One can nicely see from the plots in the lower half that
this is the case for most of the regions, except for the tropics where
seasonality is also small. In the upper panel, we also observe the largest
discrepancies between the upscaled NEE product and the results from the
atmospheric inversions in the tropical regions. However, this is a known
problem for the upscaling approaches that rely on flux tower measurements
and not a specific problem of our
product. The reason is not yet clarified in the community but one issue is
related to the fact that there are only very few flux tower sites in tropical
regions, which can also be observed from Fig. .
For some regions, such as Southern Africa and South American Temperate, mismatches
between the inversions and the flux tower upscaling might also be due to
contributions of fire emissions which are “seen” by the atmosphere but not
in our approach. Overall, the patterns of the MSC in most of the regions are
very similar to the results of the atmospheric inversions. This is remarkable
given that the two approaches and data sources are entirely independent.
Thus, our upscaling product has the potential to provide further constraints
for the atmospheric inversion methods with the benefit of high resolution in
both space and time.
Distributions of correlation coefficients for the mean seasonal
cycles in the 11 TransCom regions comparing either individual ensemble
members from the set of atmospheric inversions or each ensemble member with
our upscaled NEE product.
We use the atmospheric inversions as an independent benchmark here, even though
a number of uncertainties also apply to those. To put the agreement of the
upscaling with the inversions into context of agreement among inversions, we
display the values of pairwise correlation coefficients for the MSC of the
different inversion methods together with the correlation coefficients
between the MSC of each inversion approach and our upscaling product for each
TransCom land region in Fig. . Overall, the
agreement of our upscaling approach with the inversions is comparable with
the agreement among different inversions suggesting promise for a synergistic
joint usage of both approaches. There is very large consistency among
inversions and with the upscaling approach for the regions North American
Boreal, North American Temperate, Eurasian Boreal, and Europe. Interestingly,
the upscaling approach is more consistent with the inversions than among the
inversions for the Eurasian Temperate region. We observe that the seasonality
of NEE in the tropical regions is not well constrained by each approach.
Data availability and usage notes
The calculated global half-hourly flux products are publicly available for
free at https://doi.org/10.17871/BACI.224 under
the creative commons license CC BY 4.0
Please check
https://creativecommons.org/licenses/by/4.0/ as well as
https://creativecommons.org/licenses/by/4.0/legalcode
. More precisely,
gridded products at 0.5∘ spatial resolution and half-hourly
temporal resolution are provided covering GPP, NEE, LE, and H for the years
2001 to 2014. In addition, a derived product of monthly average diurnal
cycles globally for these four fluxes and the given range of years at the
same spatial resolution has been prepared for download as well. It is much
more convenient for the user to just download the lightweight data of the
monthly averages instead of getting the half-hourly data of much larger file
size and then performing the averaging on the local machine. As mentioned in
the paper, the monthly average diurnal cycles are primarily recommended for
usage, since this derived product turned out to be more robust.
Please note that all data files of GPP can contain slightly negative values,
which seems to be implausible at first glance. However, these negative values
mainly occur during nighttime and are the result of an artifact in the flux
partitioning method at site level carried out for the FLUXNET eddy covariance tower network, where observed NEE is
separated into GPP and ecosystem respiration. The negative values from the
sites are part of the training set for the proposed upscaling approach, and
therefore the machine learning model can produce negative GPP values for
similar environmental conditions as well. Since our data products are obtained by an entirely
data-driven machine learning approach, the observational
error at site level (that also causes negative nighttime GPP at the sites)
propagates to global scale.
Hence, dealing with negative GPP observations is not only a problem in our
global data products but also occurs when working with site-level data.
Neglecting negative values at the sites during model learning or manually setting them to zero would lead
to biased regression models and setting negative estimations to zero would
cause biased predictions. We therefore decided to keep negative values both
in the training set and in our provided global data products. If these
negative values are causing trouble within any application that builds on our
data products, they can easily be set to zero by the user as an appropriate
post-processing step. However, the user should keep in mind that this leads
to an overall bias within the data product.
The data products are stored as individual files for each variable and each
year that has been considered. We have chosen the platform-independent
NetCDF
https://www.unidata.ucar.edu/software/netcdf/
file
format and software packages exist in many scientific programming
languages (including Matlab, python, and R) for easy data access.
Conclusions and future work
In this paper, we have shown how to perform an upscaling of half-hourly
carbon and energy fluxes from local in situ measurements to global
scale. We have introduced two
general prediction approaches to estimate half-hourly values mainly from
predictor variables at coarser temporal resolution. Since the problem has
been formulated as a large-scale regression task, we have been working with
random forest regression, although other regression algorithms could be
applied as well. Our prediction approaches have been validated by a set of
cross-validation experiments employing a leave-one-site-out strategy for the
FLUXNET towers that provide the observations. As a result of our analyses, we
have presented global flux products at half-hourly temporal resolution for
the years 2001 to 2014 covering four important variables: gross primary
production, net ecosystem exchange, latent heat, and sensible heat. Detailed
descriptions of the experimental setup for the cross-validation as well as
for the computations that have led to the global products were given as well.
Concerning the global data products, we have also shown derived statistics
like maximum diurnal amplitudes of a month as well as maximum half-hourly
fluxes at each spatial position. These properties can only be computed from
data products with subdaily temporal resolution showing the benefits of our
contributions.
In future work, we aim at improving the prediction performance of half-hourly
fluxes in various ways. First, we plan to add additional sources of
information to the drivers by extending the set of predictor variables to
cover further relevant aspects for the individual fluxes like water
availability or soil properties. This would allow for tackling difficult
scenarios like seasonal droughts, where the current approaches have shown
larger errors in the prediction. Second, we also want to incorporate the
history of the predictor variables in order to account for lagged effects. So
far, samples are treated independently in the prediction but their temporal
context due to the time series characteristics may provide additional
knowledge that can be exploited for the estimation of fluxes. Third, subdaily
meteorology could be included in the calculations of the global products by
incorporating the new generation of meteorological reanalysis data of ERA5 at an hourly
timescale that will be released in the near future or by exploiting
observations from geostationary satellites. Of course, the global products
will be updated if these additional ideas lead to better prediction
performances. Another import aspect of future work is providing uncertainties
for the flux estimations, which could be done by quantile regression
approaches .
FLUXNET tower locations
Since our machine learning models for the upscaling tasks depend on in situ
measurements from FLUXNET towers that are required to create the training
set, it is necessary to look at the spatial distribution of these towers to
judge on the adequacy of the global data products. In
Fig. , we display the FLUXNET tower locations of the
222 sites that have been used in this paper by superimposing them on the
maximum half-hourly GPP map of Fig. . One
can clearly observe a bias in the distribution of sites, since most of the
towers are located in Europe and North America. Hence, regions with similar
climatic conditions are well represented by our global data products compared
to ecosystems that are also less captured by the site network, e.g., the
tropical regions. Please note that this bias is not a specific problem for
the upscaling approaches and global data products of this paper, because it
affects all upscaling studies and corresponding data products
. Thus, it does not
only influence the diurnal cycle of fluxes, which is the key contribution of
this paper, but rather remains a more general issue. We also refer to
previous studies about the representativeness of FLUXNET, see for example the
work of , who make use of climate-space mappings to
illustrate the impact of an imbalanced distribution of sites on carbon
uptake.
The superimposed FLUXNET tower locations of the sites used in this
paper clearly show the biased distribution of underlying in situ measurements
due to better spatial coverage of regions in Europe and North America
compared to the rest of the world.
Gaps in the flux data
In Sect. , we mentioned that we have ignored gaps in
the half-hourly flux data. From a machine learning perspective, we can anyway
only do the training on the data we have, independent of their distribution
in time. Furthermore, we have decided to not include gap-filled data in order
to prevent the machine learning model to adapt too much to the gap-filling
method. However, we have found that within all the site days that we have
taken into account, there are roughly 35 % gaps and
Fig. shows their distribution among the half
hours.
The distribution of gaps in the flux data of 129 456 site days that
we have used from 222 FLUXNET towers clearly shows a nighttime dominance of
the gaps. In total, there are roughly 35 % of gaps in this half-hourly
flux data.
One can clearly observe a nighttime dominance of the gaps. For GPP, this is
not a big problem, because it is assumed to be zero anyhow. Considering NEE,
the absolute fluxes are also smaller during night compared to daytime
observations. The nighttime dominance of gaps arises from less turbulence
during these hours and this is an inherent problem of the measurement devices
that we cannot resolve. However, it should be noted that such a biased
distribution of gaps does not directly lead to a model bias, as it would be
the case, for example, for linear methods. Since we have picked random
forests as a nonlinear machine learning technique, our derived models are
less biased for imbalanced data because the final estimations in the leaf
nodes of the decision trees are made locally in predictor space by
considering mean values from samples that fall into the respective leaf node.
Hence, they are independent from samples that are far away in predictor space
but could potentially have higher or lower density.
In addition, we have also carried out preliminary experiments where we have
only used site days with no gaps, i.e., where all 48 half-hourly values have
been available. This has then reduced the overall number of training samples
massively and has clearly reduced prediction performance, most likely due to
worse generalization abilities because the reduced training data did not
capture all environmental conditions sufficiently well.
Predictor variables
In Table , we have reproduced Table 2 of
that lists the predictor variables with at most
daily temporal resolution used in our upscaling study. Note that we only used
the variables from the RS+METEO setup.
This table has been reproduced from Table 2
of and the original caption is the following one.
“Selected predictors for both setups for CO2 fluxes (GPP, TER and NEE) and
energy fluxes (H, LE and Rn). List of acronyms: Enhanced Vegetation Index
(EVI), fraction of absorbed photosynthetically active radiation (fAPAR), leaf
area index (LAI), daytime land surface temperature (LSTDay) and
nighttime land surface temperature (LSTNight), middle infrared
reflectance (band 7; MIR1), Normalized Difference Vegetation Index (NDVI),
Normalized Difference Water Index (NDWI), plant functional type (PFT),
incoming global radiation (Rg), top of atmosphere potential
radiation (Rpot), Index of Water Availability (IWA), relative
humidity (Rh), Water Availability Index lower (WAIL), and upper
(WAIU), and mean seasonal cycle (MSC). The product between A and B
(A × B) is shown as (A, B).”
SetupType of variabilityCO2 fluxesEnergy fluxesRSSpatialPFTPFTAmplitude of MSC of EVIMaximum of MSC of (fAPAR, Rg)Amplitude of MSC of MIR1Minimum of MSC of RgMaximum of MSC of LSTDaySpatial and seasonalMSC of LAIMSC of (EVI, LSTDay)RpotSpatial, seasonal, and interannualNDWIRgLSTDayLSTDayLSTNightAnomalies of LSTNight(NDVI, Rg)Anomalies of (EVI, LSTDay)RS+METEOSpatialPFTPFTAmplitude of MSC of NDVIMaximum of MSC of WAIUAmplitude of MSC of band 4BRDF reflectance2Mean of MSC of band 6 BRDFreflectance2Minimum of MSC of NDWIMax of MSC of (fPAR, Rg)Amplitude of MSC of WAILSpatial and seasonalMSC of LSTNightRpotMSC of (fPAR, LSTDay)MSC of NDWIMSC of (EVI, Rpot)MSC of LSTNightMSC of (EVI, Rg)Spatial, seasonal, and interannualTairRain(Rg, MSC of NDVI)RgWAILRh(MSC of NDVI, Rg, IWA)
1 Derived from the MOD13 product. 2 Derived from the MCD43
product.
FLUXNET sites
In this study, we made use of data from 222 FLUXNET sites that are equipped
with eddy covariance towers. We would like to thank all the data providers of
these sites for their hard work by collecting, filtering, and processing the
raw data as well as for sharing the data with the community. In
Table , we list the used FLUXNET sites together with
corresponding references. Note that a map of tower locations is shown in
Fig. .
This is a list of 222 FLUXNET sites from
which we have used data in our
study.
Site IDNameCountryLatLongYearsReferenceAT-NeuNeustift/Stubai ValleyAustria47.1211.322002–2006AU-FogFogg DamAustralia-12.54131.312006–2007AU-HowHoward SpringsAustralia-12.49131.152001–2006AU-TumTumbarumbaAustralia-35.66148.152001–2006AU-WacWallaby CreekAustralia-37.43145.192005–2007BE-BraBrasschaat (De Inslag Forest)Belgium51.314.521997–2006BE-JalJalhayBelgium50.566.072006–2006BE-LonLonzeeBelgium50.554.742004–2006BE-VieVielsalmBelgium50.316.001996–2006BR-BanEcotone Bananal IslandBrazil-9.82-50.162003–2006BR-CaxCaxiuana Forest – AlmeirimBrazil-1.72-51.461999–2003BR-Ji2Rond. – Rebio Jaru Ji Parana – Tower ABrazil-10.08-61.932000–2002BR-Ma2Manaus – ZF2 K34Brazil-2.61-60.211999–2006BR-Sa1Santarem – Km67 – Primary ForestBrazil-2.86-54.962002–2004BR-Sa2Santarem – Km77 – PastureBrazil-3.01-54.542001–2002BR-Sa3Santarem – Km83 – Logged ForestBrazil-3.02-54.972000–2003BR-Sp1Sao Paulo CerradoBrazil-21.62-47.652001–2002BW-GhgGhanzi Grass SiteBotswana-21.5121.742003–2003BW-GhmGhanzi Mixed SiteBotswana-21.2021.752003–2003BW-Ma1Maun – Mopane WoodlandBotswana-19.9223.561999–2001CA-Ca1BC – Campbell River – Mature Forest SiteCanada49.87-125.331997–2005CA-Ca2BC – Campbell River – Clearcut SiteCanada49.87-125.292000–2005CA-Ca3BC – Campbell River – Young Plantation SiteCanada49.53-124.902001–2005CA-GroON – Groundhog River – Boreal Mixed WoodCanada48.22-82.162003–2005CA-LetAB – Lethbridge – Mixed Grass PrairieCanada49.71-112.941998–2005CA-ManMB – Manitoba – Northern Old Black Spruce – BOREAS Northern Study AreaCanada55.88-98.481994–2003CA-MerON – Eastern Peatland – Mer BleueCanada45.41-75.521998–2005CA-NS1UCI – 1850 burn siteCanada55.88-98.482002–2005CA-NS2UCI – 1930 burn siteCanada55.91-98.522001–2005CA-NS3UCI – 1964 burn siteCanada55.91-98.382001–2005CA-NS4UCI – 1964 burn site wetCanada55.91-98.382002–2004CA-NS5UCI – 1981 burn siteCanada55.86-98.492001–2005CA-NS6UCI – 1989 burn siteCanada55.92-98.962001–2005CA-NS7UCI – 1998 burn siteCanada56.64-99.952002–2005CA-OasSK – SSA Old AspenCanada53.63-106.201997–2005CA-ObsSK – SSA Old Black SpruceCanada53.99-105.121999–2005CA-OjpSK – SSA Old Jack PineCanada53.92-104.691999–2005CA-QcuQC – Boreal Cutover SiteCanada49.27-74.042001–2006CA-QfoQC – Mature Boreal Forest SiteCanada49.69-74.342003–2006
Continued.
Site IDNameCountryLatLongYearsReferenceCA-SF1SK – Fire 1977Canada54.49-105.822003–2005CA-SF2SK – Fire 1989Canada54.25-105.882003–2005CA-SF3SK – Fire 1998Canada54.09-106.002003–2005CA-SJ1SK – 1994 Harv. Jack PineCanada53.91-104.662001–2005CA-SJ2SK – 2002 Harvested Jack PineCanada53.94-104.652003–2005CA-SJ3SK – SSA 1975 Harv. Yng Jack PineCanada53.88-104.642004–2005CA-TP1ON – Turkey Point Seedling White PineCanada42.66-80.562004–2005CA-TP2ON – Turkey Point Young White PineCanada42.77-80.462003–2005CA-TP3ON – Turkey Point Middle-aged White PineCanada42.71-80.352003–2005CA-TP4ON – Turkey Point Mature White PineCanada42.71-80.362003–2005CA-WP1AB – Western Peatland – LaBiche River – Black Spruce/Larch FenCanada54.95-112.472003–2005CA-WP2AB – Western Peatland – Sphagnum moss – Poor FenCanada55.54-112.332004–2004CG-TchTchizalamouRep. of Congo-4.2911.662006–2006CH-Oe1Oensingen1 grassSwitzerland47.297.732002–2006CH-Oe2Oensingen2 cropSwitzerland47.297.732005–2005CN-BedBeijing DaxingChina39.53116.252005–2006CN-ChaChangbaishanChina42.40128.102003–2003CN-Do1Dongtan 1China31.52121.962005–2005CN-Do2Dongtan 2China31.58121.902005–2005CN-Do3Dongtan 3China31.52121.972005–2005CN-Du1Duolun_croplandChina42.05116.672005–2006CN-Du2Duolun_grasslandChina42.05116.282006–2006CN-Ku1Kubuqi_populus forestChina40.54108.692005–2006CN-Ku2Kubuqi_shrublandChina40.38108.552006–2006CN-Xi1Xilinhot fenced steppe (X06)China43.55116.682006–2006CN-Xi2Xilinhot grassland site (X03)China43.55116.672006–2006CZ-BK1Bily Kriz – Beskidy MountainsCzech Rep.49.5018.542000–2006CZ-BK2Bily Kriz – grasslandCzech Rep.49.4918.542004–2006CZ-wetCZECHWET – TrebonCzech Rep.49.0314.772006–2006DE-GebGebeseeGermany51.1010.912004–2006DE-GriGrillenburg – grass stationGermany50.9513.512005–2006DE-HaiHainichGermany51.0810.452000–2006DE-HarHartheimGermany47.937.602005–2006DE-KliKlingenberg – croplandGermany50.8913.522004–2006DE-MehMehrstedt 1Germany51.2810.662003–2006DE-ThaAnchor Station Tharandt – old spruceGermany50.9613.571996–2006DE-WetWetzsteinGermany50.4511.462002–2006DK-FouFoulumDenmark56.489.592005–2005DK-LvaLille Valby (Rimi)Denmark55.6812.082005–2006DK-RisRisbyholmDenmark55.5312.102004–2005()DK-SorSoroeDenmark55.4911.651996–2006ES-ES1El SalerSpain39.35-0.321999–2006ES-ES2El Saler – SuecaSpain39.28-0.322004–2006ES-LMaLas Majadas del TietarSpain39.94-5.772004–2006ES-VDAVall d'AlinyaSpain42.151.452004–2006FI-HyyHyytialaFinland61.8524.291996–2006FI-KaaKaamanen wetlandFinland69.1427.302000–2006FI-SiiSiikaneva fenFinland61.8324.192004–2005FI-SodSodankylaFinland67.3626.642000–2006FR-AurAuradeFrance43.551.112005–2005FR-FonFontainebleauFrance48.482.782005–2006
Continued.
Site IDNameCountryLatLongYearsReferenceFR-GriGrignon (after 5 June 2005)France48.841.952005–2006FR-HesHesse Forest – SarrebourgFrance48.677.061997–2006FR-LBrLe Bray (after 28 June 1998)France44.72-0.771996–2006FR-LamLamasquereFrance43.491.242005–2005FR-Lq1LaqueuilleFrance45.642.742004–2006FR-Lq2Laqueuille extensiveFrance45.642.742004–2006FR-PuePuechabonFrance43.743.602000–2006GF-GuyGuyafluxFrench Guyana5.28-52.932004–2006HU-BugBugacpusztaHungary46.6919.602002–2006HU-MatMatraHungary47.8519.732004–2006ID-PagPalangkarayaIndonesia-2.35114.042002–2003IE-DriDripseyIreland51.99-8.752003–2005IL-YatYatirIsrael31.3435.052001–2006IT-AmpAmpleroItaly41.9013.612002–2006IT-BCiBorgo CioffiItaly40.5214.962004–2006IT-BonBonisItaly39.4816.532006–2006IT-ColCollelongo- Selva PianaItaly41.8513.591996–2006IT-CpzCastelporzianoItaly41.7112.381997–2006IT-LMaLa MandriaItaly45.587.152003–2006IT-LavLavarone (after March 2002)Italy45.9611.282000–2006 ()IT-MBoMonte BondoneItaly46.0211.052003–2006()IT-MalMalga ArpacoItaly46.1211.702003–2006IT-NoeSardinia/Arca di NoeItaly40.618.152004–2006IT-NonNonantolaItaly44.6911.092001–2006IT-PT1Zerbolo – Parco Ticino – CanarazzoItaly45.209.062002–2004IT-PiaIsland of PianosaItaly42.5810.082002–2005IT-RenRenon/Ritten (Bolzano)Italy46.5911.431999–2006IT-Ro1Roccarespampani 1Italy42.4111.932000–2006IT-Ro2Roccarespampani 2Italy42.3911.922002–2006IT-SRoSan RossoreItaly43.7310.281999–2006JP-MasMase paddy flux site – Tsukuba – Japan (MSE)Japan36.05140.032002–2003JP-TakTakayamaJapan36.15137.421999–2004JP-TomTomakomai National ForestJapan42.74141.512001–2003KR-HnmHaenamKorea34.55126.572004–2006KR-Kw1Gwangneung Coniferous ForestKorea37.75127.162004–2007NL-Ca1Cabauwthe Netherlands51.974.932003–2006NL-HaaHaastrechtthe Netherlands52.004.812003–2004NL-HorHorstermeerthe Netherlands52.035.072004–2006NL-LanLangerakthe Netherlands51.954.902005–2006NL-LooLoobosthe Netherlands52.175.741996–2006NL-LutLutjewadthe Netherlands53.406.362006–2006NL-MolMolenwegthe Netherlands51.654.642005–2006PT-EspEspirraPortugal38.64-8.602002–2006PT-Mi1Mitra (Evora)Portugal38.54-8.002003–2005PT-Mi2Mitra IV TojalPortugal38.48-8.022004–2006RU-CheCherskiiRussia68.61161.342002–2005RU-CokChokurdakhRussia70.62147.882003–2005RU-FyoFyodorovskoye wet spruce standRussia56.4632.921998–2006RU-Ha1Ubs Nur – Hakasija – grasslandRussia54.7390.002002–2004 ()RU-Ha2Ubs Nur – Hakasija – recovering grasslandRussia54.7789.962002–2003 ()RU-Ha3Ubs Nur – Hakasija – Site 3Russia54.7089.082004–2004 ()RU-ZotZotinoRussia60.8089.352002–2004
Continued.
Site IDNameCountryLatLongYearsReferenceSE-AbiAbiskoSweden68.3618.792005–2005SE-DegDegeroSweden64.1819.552001–2005SE-NorNorundaSweden60.0917.481996–2005SE-Sk1Skyttorp1 youngSweden60.1317.922005–2005SE-Sk2SkyttorpSweden60.1317.842004–2005SE-St1Stordalen Forest- Mountain BirchSweden68.3719.052006–2006SK-TatTatraSlovak Rep.49.1220.162005–2005UK-AMoAuchencorth Moss – ScotlandUK55.79-3.242005–2005UK-EBuEaster Bush- ScotlandUK55.87-3.212004–2006UK-ESaEast SaltounUK55.91-2.862003–2005UK-GriGriffin – Aberfeldy – ScotlandUK56.61-3.801997–2006UK-HamHampshireUK51.12-0.862004–2005UK-HerHertfordshireUK51.78-0.482006–2006UK-TadTadham MoorUK51.21-2.832001–2001US-ARMOK – ARM Southern Great Plains site – LamontUSA36.61-97.492003–2006US-AudAZ – Audubon Research RanchUSA31.59-110.512002–2006US-BarNH – Bartlett Experimental ForestUSA44.06-71.292004–2005US-BkgSD – BrookingsUSA44.35-96.842004–2006US-BloCA – Blodgett ForestUSA38.90-120.631997–2006US-Bn1AK – Bonanza Creek; 1920 Burn site near Delta JunctionUSA63.92-145.382003–2003US-Bn2AK – Bonanza Creek; 1987 Burn site near Delta JunctionUSA63.92-145.382003–2003US-Bn3AK – Bonanza Creek; 1999 Burn site near Delta JunctionUSA63.92-145.742003–2003US-Bo1IL – BondvilleUSA40.01-88.291996–2007US-Bo2IL – Bondville (companion site)USA40.01-88.292004–2006US-BrwAK – BarrowUSA71.32-156.631998–2002US-CaVWV – Canaan ValleyUSA39.06-79.422004–2005US-Dk1NC – Duke Forest – open fieldUSA35.97-79.092001–2005US-Dk2NC – Duke Forest – hardwoodsUSA35.97-79.102003–2005US-Dk3NC – Duke Forest – loblolly pineUSA35.98-79.092001–2005US-FPeMT – Fort PeckUSA48.31-105.102000–2006US-FR2TX – Freeman Ranch – Mesquite JuniperUSA29.95-98.002004–2006US-FufAZ – Flagstaff – Unmanaged ForestUSA35.09-111.762005–2006US-FwfAZ – Flagstaff – WildfireUSA35.45-111.772005–2006US-GooMS – Goodwin CreekUSA34.25-89.872002–2006US-Ha1MA – Harvard Forest EMS Tower (HFR1)USA42.54-72.171991–2006US-Ha2MA – Harvard Forest Hemlock SiteUSA42.54-72.182004–2004US-Ho1ME – Howland Forest (main tower)USA45.20-68.741996–2004US-Ho2ME – Howland Forest (west tower)USA45.21-68.751999–2004US-IB1IL – Fermi National Accelerator Lab –Batavia (Agricultural site)USA41.86-88.222005–2007US-IB2IL – Fermi National Accelerator Lab –Batavia (Prairie site)USA41.84-88.242004–2007US-IvoAK – IvotukUSA68.49-155.752003–2006US-KS1FL – Kennedy Space Center (slash pine)USA28.46-80.672002–2002US-KS2FL – Kennedy Space Center (scrub oak)USA28.61-80.672000–2006US-LPHMA – Little Prospect HillUSA42.54-72.182002–2005US-LosWI – Lost CreekUSA46.08-89.982001–2005US-MMSIN – Morgan Monroe State ForestUSA39.32-86.411999–2005US-MOzMO – Missouri Ozark SiteUSA38.74-92.202004–2006US-Me1OR – Metolius – Eyerly burnUSA44.58-121.502004–2005US-Me2OR – Metolius – intermediate aged ponderosa pineUSA44.45-121.562003–2005US-Me3OR – Metolius – second young aged pineUSA44.32-121.612004–2005US-Me4OR – Metolius – old aged ponderosa pineUSA44.50-121.621996–2000
Continued.
Site IDNameCountryLatLongYearsReferenceUS-NC1NC – NC_ClearcutUSA35.81-76.712005–2006US-NC2NC – NC_Loblolly PlantationUSA35.80-76.672005–2006US-NR1CO – Niwot Ridge Forest (LTER NWT1)USA40.03-105.551999–2003US-Ne1NE – Mead – irrigated continuous maize siteUSA41.17-96.482001–2005US-Ne2NE – Mead – irrigated maize-soybean rotation siteUSA41.16-96.472001–2005US-Ne3NE – Mead – rainfed maize-soybean rotation siteUSA41.18-96.442001–2005US-PFaWI – Park Falls/WLEFUSA45.95-90.271996–2003US-SO2CA – Sky Oaks – Old StandUSA33.37-116.621997–2006US-SO3CA – Sky Oaks – Young StandUSA33.38-116.621997–2006US-SO4CA – Sky Oaks – New StandUSA33.38-116.642004–2006US-SP1FL – Slashpine – Austin Cary – 65yrs nat regenUSA29.74-82.222000–2005US-SP2FL – Slashpine – Mize-clearcut – 3yrs regenUSA29.76-82.241998–2004US-SP3FL – Slashpine – Donaldson-mid-rot – 12yrsUSA29.75-82.161999–2004US-SRMAZ – Santa Rita MesquiteUSA31.82-110.872004–2006US-SyvMI – Sylvania Wilderness AreaUSA46.24-89.352002–2006US-TonCA – Tonzi RanchUSA38.43-120.972001–2006US-UMBMI – Univ. of Mich. Biological StationUSA45.56-84.711999–2003US-VarCA – Vaira Ranch- IoneUSA38.41-120.952001–2006US-WCrWI – Willow CreekUSA45.81-90.081999–2006US-Wi0WI – Young red pine (YRP)USA46.62-91.082002–2002US-Wi1WI – Intermediate hardwood (IHW)USA46.73-91.232003–2003US-Wi4WI – Mature red pine (MRP)USA46.74-91.172002–2005US-Wi5WI – Mixed young jack pine (MYJP)USA46.65-91.092004–2004US-Wi6WI – Pine barrens #1 (PB1)USA46.62-91.302002–2002US-Wi8WI – Young hardwood clearcut (YHW)USA46.72-91.252002–2002US-WkgAZ – Walnut Gulch Kendall GrasslandsUSA31.74-109.942004–2006US-WrcWA – Wind River Crane SiteUSA45.82-121.951998–2006VU-CocCocoFluxVanuatu-15.44167.192001–2004ZA-KruSkukuza – Kruger National ParkSouth Africa-25.0231.502001–2003
PB and MJ designed the experiments and PB carried them out, which
also involved the incorporation of ideas and suggestions from MM and MR. FG
contributed to the technical implementation. Evaluating the experiments as
well as preparing the presentation of results was done by PB, with
additional input from MJ, MM, and MR. PB wrote the manuscript with
contributions from all co-authors.
The authors declare that they have no conflict of
interest.
Acknowledgements
The work presented in this paper is part of the project “Detecting
changes in essential ecosystem and biodiversity properties – towards a
Biosphere Atmosphere Change Index: BACI”. This project has received funding
from the European Union's Horizon 2020 Research and Innovation
programme under grant agreement no. 640176. Furthermore, this work used eddy
covariance data acquired by the FLUXNET community and in particular by the
following networks: AmeriFlux (US Department of Energy, Biological and
Environmental Research, Terrestrial Carbon Program, DE-FG02-04ER63917 and
DE-FG02-04ER63911), AfriFlux, AsiaFlux, CarboAfrica, CarboEuropeIP,
CarboItaly, CarboMont, ChinaFlux, Fluxnet-Canada (supported by CFCAS, NSERC,
BIOCAP, Environment Canada, and NRCan), GreenGrass, KoFlux, LBA, NECC,
OzFlux, TCOS-Siberia, and USCCC. We acknowledge the financial support to the
eddy covariance data harmonization provided by CarboEuropeIP, FAO-GTOS-TCO,
iLEAPS, Max Planck Institute for Biogeochemistry, National Science
Foundation, University of Tuscia, Université Laval and Environment
Canada, and the US Department of Energy, as well as the database development and technical
support from the Berkeley Water Center, Lawrence Berkeley National Laboratory,
Microsoft Research eScience, Oak Ridge National Laboratory, University of
California – Berkeley, and the University of Virginia. Edited by: Vinayak Sinha Reviewed by: Ronald
Prinn and one anonymous referee
ReferencesAcosta, M., Pavelka, M., Pokorný, R., Janouš, D., and
Marek, M. V.: Seasonal Variation in CO2 Efflux of Stems and Branches
of Norway Spruce Trees, Ann. Bot., 101, 469–477, 10.1093/aob/mcm304,
2008.Allard, V., Soussana, J.-F., Falcimagne, R., Berbigier, P.,
Bonnefond, J.-M., Ceschia, E., D'hour, P., Hénault, C.,
Laville, P., Martin, C., and Pinarès-Patino, C.: The role of
grazing management for the net biome productivity and greenhouse gas budget
(CO2, N2O and CH4) of semi-natural grassland, Agr. Ecosyst.
Environ., 121, 47–58, 10.1016/j.agee.2006.12.004, 2007.Allard, V., Ourcival, J. M., Rambal, S., Joffre, R., and Rocheteau,
A.: Seasonal and annual variation of carbon exchange in an evergreen
Mediterranean forest in southern France, Glob. Change Biol., 14,
714–725, 10.1111/j.1365-2486.2008.01539.x, 2008.Allison, V. J., Yermakov, Z., Miller, R. M., Jastrow, J. D., and
Matamala, R.: Using landscape and depth gradients to decouple the impact of
correlated environmental variables on soil microbial community composition,
Soil Biol. Biochem., 39, 505–516, 10.1016/j.soilbio.2006.08.021, 2007.Amiro, B. D., Barr, A. G., Black, T. A., Iwashita, H., Kljun, N.,
McCaughey, J. H., Morgenstern, K., Murayama, S., Nesic, Z.,
Orchansky, A. L., and Saigusa, N.: Carbon, energy and water fluxes at
mature and disturbed forest sites, Saskatchewan, Canada, Agr. Forest
Meteorol., 136, 237–251, 10.1016/j.agrformet.2004.11.012, 2006.Ammann, C., Flechard, C. R., Leifeld, J., Neftel, A., and Fuhrer,
J.: The carbon budget of newly established temperate grassland depends on
management intensity, Agr. Ecosyst. Environ., 121, 5–20,
10.1016/j.agee.2006.12.002, 2007.Anthoni, P. M., Knohl, A., Rebmann, C., Freibauer, A., Mund, M.,
Ziegler, W., Kolle, O., and Schulze, E.-D.: Forest and agricultural
land-use-dependent CO2 exchange in Thuringia, Germany, Glob. Change
Biol., 10, 2005–2019, 10.1111/j.1365-2486.2004.00863.x, 2004.Arain, M. A. and Restrepo-Coupe, N.: Net ecosystem production in a
temperate pine plantation in southeastern Canada, Agr. Forest Meteorol., 128,
223–241, 10.1016/j.agrformet.2004.10.003, 2005.Arneth, A., Veenendaal, E. M., Best, C., Timmermans, W., Kolle, O.,
Montagnani, L., and Shibistova, O.: Water use strategies and
ecosystem-atmosphere exchange of CO2 in two highly seasonal environments,
Biogeosciences, 3, 421–437, 10.5194/bg-3-421-2006, 2006.Aubinet, M., Chermanne, B., Vandenhaute, M., Longdoz, B., Yernaux,
M., and Laitat, E.: Long term carbon dioxide exchange above a mixed forest
in the Belgian Ardennes, Agr. Forest Meteorol., 108, 293–315,
10.1016/S0168-1923(01)00244-1, 2001.Aubinet, M., Vesala, T., and Papale, D. (Eds.): Eddy Covariance: A
Practical Guide to Measurement and Data Analysis, 1st edn., Springer
Atmospheric Sciences, Springer, 10.1007/978-94-007-2351-1, 2012.Aurela, M., Laurila, T., and Tuovinen, J.-P.: The timing of snow melt
controls the annual CO2 balance in a subarctic fen, Geophys. Res. Lett.,
31, 1–4, 10.1029/2004GL020315, 2004.
Balddocchi, D.: Measuring fluxes of trace gases and energy between ecosystems
and the atmosphere – the state and future of the eddy covariance method,
Glob. Change Biol., 20, 3600–3609, 2014.Baldocchi, D. D., Hincks, B. B., and Meyers, T. P.: Measuring
Biosphere-Atmosphere Exchanges of Biologically Related Gases with
Micrometeorological Methods, Ecology, 69, 1331–1340, 10.2307/1941631,
1988.Balzarolo, M., Anderson, K., Nichol, C., Rossini, M., Vescovo, L.,
Arriga, N., Wohlfahrt, G., Calvet, J.-C., Carrara, A., Cerasoli,
S., Cogliati, S., Daumard, F., Eklundh, L., Elbers, J. A.,
Evrendilek, F., Handcock, R. N., Kaduk, J., Klumpp, K., Longdoz,
B., Matteucci, G., Meroni, M., Montagnani, L., Ourcival, J.-M.,
Sánchez-Cañete, E. P., Pontailler, J.-Y., Juszczak, R.,
Scholes, B., and Martín, M. P.: Ground-Based Optical Measurements at
European Flux Sites: A Review of Methods, Instruments and Current
Controversies, Sensors, 11, 7954–7981, 10.3390/s110807954, 2011.Beer, C., Reichstein, M., Tomelleri, E., Ciais, P., Jung, M.,
Carvalhais, N., Rödenbeck, C., Arain, M. A., Baldocchi, D.,
Bonan, G. B., Bondeau, A., Cescatti, A., Lasslop, G., Lindroth, A.,
Lomas, M., Luyssaert, S., Margolis, H., Oleson, K. W., Roupsard,
O., Veenendaal, E., Viovy, N., Williams, C., Woodward, F. I., and
Papale, D.: Terrestrial Gross Carbon Dioxide Uptake: Global Distribution
and Covariation with Climate, Science, 329, 834–838,
10.1126/science.1184984, 2010.Belelli Marchesini, L., Papale, D., Reichstein, M., Vuichard, N., Tchebakova,
N., and Valentini, R.: Carbon balance assessment of a natural steppe of
southern Siberia by multiple constraint approach, Biogeosciences, 4,
581–595, 10.5194/bg-4-581-2007, 2007.Berbigier, P., Bonnefond, J.-M., and Mellmann, P.: CO2 and water
vapour fluxes for 2 years above Euroflux forest site, Agr. Forest
Meteorol., 108, 183–197, 10.1016/S0168-1923(01)00240-4, 2001.Bergeron, O., Margolis, H. A., Black, T. A., Coursolle, C., Dunn,
A. L., Barr, A. G., and Wofsy, S. C.: Comparison of carbon dioxide fluxes
over three boreal black spruce forests in Canada, Glob. Change Biol., 13,
89–107, 10.1111/j.1365-2486.2006.01281.x, 2007.Beringer, J., Hutley, L. B., Tapper, N. J., and Cernusak, L. A.:
Savanna fires and their impact on net ecosystem productivity in North
Australia, Glob. Change Biol., 13, 990–1004,
10.1111/j.1365-2486.2007.01334.x, 2007.Beringer, J., Livesley, S. J., Randle, J., and Hutley, L. B.: Carbon
dioxide fluxes dominate the greenhouse gas exchanges of a seasonal wetland in
the wet-dry tropics of northern Australia, Agr. Forest Meteorol., 182,
239–247, 10.1016/j.agrformet.2013.06.008, 2013.Bernhofer, C., Aubinet, M., Clement, R., Grelle, A., Grünwald,
T., Ibrom, A., Jarvis, P., Rebmann, C., Schulze, E.-D., and
Tenhunen, J. D.: Spruce forests (Norway and Sitka spruce, including
Douglas fir): Carbon and water fluxes and Balances, ecological and
ecophysiological determinants, in: Fluxes of carbon, water and energy of
European forests, edited by: Valentini, R., Springer, 99–123,
10.1007/978-3-662-05171-9_6, 2003.Béziat, P., Ceschia, E., and Dedieu, G.: Carbon balance of a three
crop succession over two cropland sites in South West France, Agr.
Forest Meteorol., 149, 1628–1645, 10.1016/j.agrformet.2009.05.004,
2009.
Bishop, C. M.: Pattern Recognition and Machine Learning, Springer,
ISBN: 978-0-387-31073-2, 2006.Bodesheim, P., Jung, M., Gans, F., Mahecha, M. D., and Reichstein, M.:
Upscaled diurnal cycles of carbon and energy fluxes, Max Planck Institute for
Biogeochemistry, 10.17871/BACI.224, 2017.Bonal, D., Bosc, A., Ponton, S., Goret, J.-Y., Burban, B., Gross,
P., Bonnefond, J.-M., Elbers, J., Longdoz, B., Epron, D., Guehl,
J.-M., and Granier, A.: Impact of severe dry season on net ecosystem
exchange in the Neotropical rainforest of French Guiana, Glob. Change
Biol., 14, 1917–1933, 10.1111/j.1365-2486.2008.01610.x, 2008.Bonan, G. B.: Forests and climate change: forcings, feedbacks, and the
climate benefits of forests, Science, 320, 1444–1449,
10.1126/science.1155121, 2008.Bonan, G. B., Lawrence, P. J., Oleson, K. W., Levis, S., Jung, M.,
Reichstein, M., Lawrence, D. M., and Swenson, S. C.: Improving canopy
processes in the Community Land Model version 4 (CLM4) using global flux
fields empirically inferred from FLUXNET data, J. Geophys. Res.-Biogeo.,
116, G02014, 10.1029/2010JG001593, 2011.Borma, L. S., Rocha, H. R. D., Cabral, O. M., Randow, C. v.,
Collicchio, E., Kurzatkowski, D., Brugger, P. J., Freitas, H.,
Tannus, R., Oliveira, L., Renno, C. D., and Artaxo, P.: Atmosphere
and hydrological controls of the evapotranspiration over a floodplain forest
in the Bananal Island region, Amazonia, J. Geophys. Res.-Biogeo., 114,
G01003, 10.1029/2007JG000641, 2009.Bracho, R., Powell, T. L., Dore, S., Li, J., Hinkle, C. R., and
Drake, B. G.: Environmental and biological controls on water and energy
exchange in Florida scrub oak and pine flatwoods ecosystems, J. Geophys.
Res.-Biogeo., 113, 1–13, 10.1029/2007JG000469, 2008.Bracho, R., Starr, G., Gholz, H. L., Martin, T. A., Cropper, W. P.,
and Loescher, H. W.: Controls on carbon dynamics by ecosystem structure and
climate for southeastern U.S. slash pine plantations, Ecol. Monogr., 82,
101–128, 10.1890/11-0587.1, 2012.Breiman, L.: Bagging Predictors, Mach. Learn., 24, 123–140,
10.1007/BF00058655, 1996.Breiman, L.: Random Forests, Mach. Learn., 45, 5–32,
10.1023/A:1010933404324, 2001.Broquet, G., Chevallier, F., Bréon, F.-M., Kadygrov, N., Alemanno, M.,
Apadula, F., Hammer, S., Haszpra, L., Meinhardt, F., Morguí, J. A.,
Necki, J., Piacentino, S., Ramonet, M., Schmidt, M., Thompson, R. L.,
Vermeulen, A. T., Yver, C., and Ciais, P.: Regional inversion of CO2
ecosystem fluxes from atmospheric measurements: reliability of the
uncertainty estimates, Atmos. Chem. Phys., 13, 9039–9056,
10.5194/acp-13-9039-2013, 2013.Carvalhais, N., Reichstein, M., Seixas, J., James Collatz, G.,
Santos Pereira, J., Berbigier, P., Carrara, A., Granier, A.,
Montagnani, L., Papale, D., Rambal, S., Sanz, M. J., and Valentini,
R.: Implications of the carbon cycle steady state assumption for
biogeochemical modeling performance and inverse parameter retrieval, Global
Biogeochem. Cy., 22, 1–16, 10.1029/2007GB003033, 2008.Cescatti, A. and Marcolla, B.: Drag coefficient and turbulence intensity
in conifer canopies, Agr. Forest Meteorol., 121, 197–206,
10.1016/j.agrformet.2003.08.028, 2004.Chen, S., Chen, J., Lin, G., Zhang, W., Miao, H., Wei, L.,
Huang, J., and Han, X.: Energy balance and partition in Inner
Mongolia steppe ecosystems with different land use types, Agr. Forest
Meteorol., 149, 1800–1809, 10.1016/j.agrformet.2009.06.009, 2009.Chevallier, F., Ciais, P., Conway, T. J., Aalto, T., Anderson,
B. E., Bousquet, P., Brunke, E. G., Ciattaglia, L., Esaki, Y.,
Fröhlich, M., Gomez, A., Gomez-Pelaez, A. J., Haszpra, L.,
Krummel, P. B., Langenfelds, R. L., Leuenberger, M., Machida, T.,
Maignan, F., Matsueda, H., Morgui, J. A., Mukai, H., Nakazawa, T.,
Peylin, P., Ramonet, M., Rivier, L., Sawa, Y., Schmidt, M.,
Steele, L. P., Vay, S. A., Vermeulen, A. T., Wofsy, S., and Worthy,
D.: CO2 surface fluxes at grid point scale estimated from a global 21
year reanalysis of atmospheric measurements, J. Geophys. Res.-Atmos., 115,
1–17, 10.1029/2010JD013887, 2010.Chiesi, M., Maselli, F., Bindi, M., Fibbi, L., Cherubini, P.,
Arlotta, E., Tirone, G., Matteucci, G., and Seufert, G.: Modelling
carbon budget of Mediterranean forests using ground and remote sensing
measurements, Agr. Forest Meteorol., 135, 22–34,
10.1016/j.agrformet.2005.09.011, 2005.Christensen, T. R., Johansson, T., Olsrud, M., Ström, L.,
Lindroth, A., Mastepanov, M., Malmer, N., Friborg, T., Crill, P.,
and Callaghan, T. V.: A catchment-scale carbon and greenhouse gas budget of
a subarctic landscape, Philos. T. Roy. Soc. A, 365, 1643–1656,
10.1098/rsta.2007.2035, 2007.Christensen, T. R., Jackowicz-Korczyński, M., Aurela, M., Crill,
P., Heliasz, M., Mastepanov, M., and Friborg, T.: Monitoring the
Multi-Year Carbon Balance of a Subarctic Palsa Mire with Micrometeorological
Techniques, AMBIO, 41, 207–217, 10.1007/s13280-012-0302-5, 2012.Chu, H., Baldocchi, D. D., John, R., Wolf, S., and Reichstein, M.: Fluxes all
of the time? A primer on the temporal representativeness of FLUXNET, J.
Geophys. Res.-Biogeo., 122, 289–307, 10.1002/2016JG003576, 2017.Cook, B. D., Davis, K. J., Wang, W., Desai, A., Berger, B. W.,
Teclaw, R. M., Martin, J. G., Bolstad, P. V., Bakwin, P. S., Yi,
C., and Heilman, W.: Carbon exchange and venting anomalies in an upland
deciduous forest in northern Wisconsin, USA, Agr. Forest Meteorol., 126,
271–295, 10.1016/j.agrformet.2004.06.008, 2004.Corradi, C., Kolle, O., Walter, K., Zimov, S. A., and Schulze,
E.-D.: Carbon dioxide and methane exchange of a north-east Siberian tussock
tundra, Glob. Change Biol. 11, 1910–1925,
10.1111/j.1365-2486.2005.01023.x, 2005.Criminisi, A. and Shotton, J. (Eds.): Decision Forests for Computer
Vision and Medical Image Analysis, Springer, 10.1007/978-1-4471-4929-3,
2013.Cutler, D. R., Edwards, T. C., Beard, K. H., Cutler, A., Hess,
K. T., Gibson, J., and Lawler, J. J.: Random forests for classification
in ecology, Ecology, 88, 2783–2792, 10.1890/07-0539.1, 2007.Cyronak, T., Santos, I. R., McMahon, A., and Eyre, B. D.: Carbon
cycling hysteresis in permeable carbonate sands over a diel cycle:
Implications for ocean acidification, Limnol. Oceanogr., 58, 131–143,
10.4319/lo.2013.58.1.0131, 2012.Davidson, S. J., Santos, M. J., Sloan, V. L., Watts, J. D.,
Phoenix, G. K., Oechel, W. C., and Zona, D.: Mapping Arctic Tundra
Vegetation Communities Using Field Spectroscopy and Multispectral Satellite
Data in North Alaska, USA, Remote Sens., 8, 1–24,
10.3390/rs8120978, 2016.Delpierre, N., Berveiller, D., Granda, E., and Dufrêne, E.: Wood
phenology, not carbon input, controls the interannual variability of wood
growth in a temperate oak forest, New Phytol., 210, 459–470,
10.1111/nph.13771, 2016.Desai, A. R., Bolstad, P. V., Cook, B. D., Davis, K. J., and Carey,
E. V.: Comparing net ecosystem exchange of carbon dioxide between an
old-growth and mature forest in the upper Midwest, USA, Agr. Forest
Meteorol., 128, 33–55, 10.1016/j.agrformet.2004.09.005, 2005.Desai, A. R., Noormets, A., Bolstad, P. V., Chen, J., Cook, B. D.,
Davis, K. J., Euskirchen, E. S., Gough, C., Martin, J. G.,
Ricciuto, D. M., Schmid, H. P., Tang, J., and Wang, W.: Influence of
vegetation and seasonal forcing on carbon dioxide fluxes across the Upper
Midwest, USA: Implications for regional scaling, Agr. Forest Meteorol.,
148, 288–308, 10.1016/j.agrformet.2007.08.001, 2008.Desai, A. R., Xu, K., Tian, H., Weishampel, P., Thom, J.,
Baumann, D., Andrews, A. E., Cook, B. D., King, J. Y., and Kolka,
R.: Landscape-level terrestrial methane flux observed from a very tall tower,
Agr. Forest Meteorol., 201, 61–75, 10.1016/j.agrformet.2014.10.017,
2015.Dirmeyer, P. A., Cash, B. A., Kinter III, J. L., Stan, C., Jung,
T., Marx, L., Towers, P., Wedi, N., Adams, J. M., Altshuler, E. L.,
Huang, B., Jin, E. K., and Manganello, J.: Evidence for Enhanced
Land-Atmosphere Feedback in a Warming Climate, J. Hydrometeor., 13, 981–995,
10.1175/JHM-D-11-0104.1, 2012.Dolman, A. J., Moors, E. J., and Elbers, J. A.: The carbon uptake of a
mid latitude pine forest growing on sandy soil, Agr. Forest Meteorol., 111,
157–170, 10.1016/S0168-1923(02)00024-2, 2002.Don, A., Rebmann, C., Kolle, O., Scherer-Lorenzen, M., and Schulze,
E.-D.: Impact of afforestation-associated management changes on the carbon
balance of grassland, Glob. Change Biol., 15, 1990–2002,
10.1111/j.1365-2486.2009.01873.x, 2009.Dore, S., Kolb, T. E., Montes-Helu, M., Sullivan, B. W., Winslow,
W. D., Hart, S. C., Kaye, J. P., Koch, G. W., and Hungate, B. A.:
Long-term impact of a stand-replacing fire on ecosystem CO2 exchange of
a ponderosa pine forest, Glob. Change Biol., 14, 1801–1820,
10.1111/j.1365-2486.2008.01613.x, 2008.Drewer, J., Lohila, A., Aurela, M., Laurila, T., Minkkinen, K.,
Penttilä, T., Dinsmore, K. J., McKenzie, R. M., Helfter, C.,
Flechard, C., Sutton, M. A., and Skiba, U. M.: Comparison of greenhouse
gas fluxes and nitrogen budgets from an ombotrophic bog in Scotland and a
minerotrophic sedge fen in Finland, Eur. J. Soil Sci., 61, 640–650,
10.1111/j.1365-2389.2010.01267.x, 2010.Dunn, A. L., Barford, C. C., Wofsy, S. C., Goulden, M. L., and
Daube, B. C.: A long-term record of carbon exchange in a boreal black
spruce forest: means, responses to interannual variability, and decadal
trends, Glob. Change Biol., 13, 577–590,
10.1111/j.1365-2486.2006.01221.x, 2007.Falk, M., Wharton, S., Schroeder, M., Ustin, S., and Paw U, K. T.:
Flux partitioning in an old-growth forest: seasonal and interannual dynamics,
Tree Physiol., 28, 509–520, 10.1093/treephys/28.4.509, 2008.Famulari, D., Fowler, D., Hargreaves, K., Milford, C., Nemitz, E.,
Sutton, M. A., and Weston, K.: Measuring Eddy Covariance Fluxes of
Ammonia Using Tunable Diode Laser Absorption Spectroscopy, Water Air Soil
Poll., 4, 151–158, 10.1007/s11267-004-3025-1, 2004.Fischer, M. L., Billesbach, D. P., Berry, J. A., Riley, W. J., and
Torn, M. S.: Spatiotemporal Variations in Growing Season Exchanges of
CO2, H2O, and Sensible Heat in Agricultural Fields of the Southern
Great Plains, Earth Interact., 11, 1–21, 10.1175/EI231.1, 2007.Flanagan, L. B., Wever, L. A., and Carlson, P. J.: Seasonal and
interannual variation in carbon dioxide exchange and carbon balance in a
northern temperate grassland, Glob. Change Biol., 8, 599–615,
10.1046/j.1365-2486.2002.00491.x, 2002.Friedl, M. A., Sulla-Menashe, D., Tan, B., Schneider, A.,
Ramankutty, N., Sibley, A., and Huang, X.: MODIS Collection 5 global
land cover: Algorithm refinements and characterization of new datasets,
Remote Sens. Environ., 114, 168–182, 10.1016/j.rse.2009.08.016, 2010.Garbulsky, M., Peñuelas, J., Papale, D., and Filella, I.: Remote
estimation of carbon dioxide uptake by a Mediterranean forest, Glob. Change
Biol., 14, 2860–2867, 10.1111/j.1365-2486.2008.01684.x, 2008.Giasson, M.-A., Coursolle, C., and Margolis, H. A.: Ecosystem-level
CO2 fluxes from a boreal cutover in eastern Canada before and after
scarification, Agr. Forest Meteorol., 140, 23–40,
10.1016/j.agrformet.2006.08.001, 2006.Gielen, B., Verbeeck, H., Neirynck, J., Sampson, D. A., Vermeiren, F., and
Janssens, I. A.: Decadal water balance of a temperate Scots pine forest
(Pinus sylvestris L.) based on measurements and modelling,
Biogeosciences, 7, 1247–1261, 10.5194/bg-7-1247-2010, 2010.Gilmanov, T. G., Soussana, J. F., Aires, L., Allard, V., Ammann,
C., Balzarolo, M., Barcza, Z., Bernhofer, C., Campbell, C. L.,
Cernusca, A., Cescatti, A., Clifton-Brown, J., Dirks, B. O. M.,
Dore, S., Eugster, W., Fuhrer, J., Gimeno, C., Gruenwald, T.,
Haszpra, L., Hensen, A., Ibrom, A., Jacobs, A. F. G., Jones, M. B.,
Lanigan, G., Laurila, T., Lohila, A., Manca, G., Marcolla, B.,
Nagy, Z., Pilegaard, K., Pinter, K., Pio, C., Raschi, A.,
Rogiers, N., Sanz, M. J., Stefani, P., Sutton, M., Tuba, Z.,
Valentini, R., Williams, M. L., and Wohlfahrt, G.: Partitioning
European grassland net ecosystem CO2 exchange into gross primary
productivity and ecosystem respiration using light response function
analysis, Agr. Ecosyst. Environ., 121, 93–120,
10.1016/j.agee.2006.12.008, 2007.Gislason, P. O., Benediktsson, J. A., and Sveinsson, J. R.: Random
Forests for land cover classification, Pattern Recogn. Lett., 27, 294–300,
10.1016/j.patrec.2005.08.011, 2006.Glenn, A. J., Flanagan, L. B., Syed, K. H., and Carlson, P. J.:
Comparison of net ecosystem CO2 exchange in two peatlands in western
Canada with contrasting dominant vegetation, Sphagnum and Carex, Agr.
Forest Meteorol., 140, 115–135, 10.1016/j.agrformet.2006.03.020, 2006.Goldstein, A. H., Hultman, N. E., Fracheboud, J. M., Bauer, M. R.,
Panek, J. A., Xu, M., Qi, Y., Guenther, A. B., and Baugh, W.:
Effects of climate variability on the carbon dioxide, water, and sensible
heat fluxes above a ponderosa pine plantation in the Sierra Nevada
(CA), Agr. Forest Meteorol., 101, 113–129,
10.1016/S0168-1923(99)00168-9, 2000.Gomez-Casanovas, N., Matamala, R., Cook, D. R., and Gonzalez-Meler,
M. A.: Net ecosystem exchange modifies the relationship between the
autotrophic and heterotrophic components of soil respiration with abiotic
factors in prairie grasslands, Glob. Change Biol., 18, 2532–2545,
10.1111/j.1365-2486.2012.02721.x, 2012.Gorsel, E. v., Leuning, R., Cleugh, H. A., Keith, H., and Suni, T.:
Nocturnal carbon efflux: reconciliation of eddy covariance and chamber
measurements using an alternative to the u*-threshold filtering technique,
Tellus B, 59, 397–403, 10.1111/j.1600-0889.2007.00252.x, 2007.Gough, C. M., Vogel, C. S., Schmid, H. P., Su, H.-B., and Curtis,
P. S.: Multi-year convergence of biometric and meteorological estimates of
forest carbon storage, Agr. Forest Meteorol., 148, 158–170,
10.1016/j.agrformet.2007.08.004, 2008.Goulden, M. L., Winston, G. C., McMillan, A. M. S., Litvak, M. E.,
Read, E. L., Rocha, A. V., and Elliot, J. R.: An eddy covariance
mesonet to measure the effect of forest age on land–atmosphere exchange,
Glob. Change Biol., 12, 2146–2162, 10.1111/j.1365-2486.2006.01251.x,
2006.Granier, A., Ceschia, E., Damesin, C., Dufrêne, E., Epron, D.,
Gross, P., Lebaube, S., Dantec, V. L., Goff, N. L., Lemoine, D.,
Lucot, E., Ottorini, J. M., Pontailler, J. Y., and Saugier, B.: The
carbon balance of a young Beech forest, Funct. Ecol., 14, 312–325,
10.1046/j.1365-2435.2000.00434.x, 2000.Grünzweig, J. M., Lin, T., Rotenberg, E., Schwartz, A., and
Yakir, D.: Carbon sequestration in arid-land forest, Glob. Change Biol., 9,
791–799, 10.1046/j.1365-2486.2003.00612.x, 2003.Gu, L., Meyers, T., Pallardy, S. G., Hanson, P. J., Yang, B.,
Heuer, M., Hosman, K. P., Riggs, J. S., Sluss, D., and
Wullschleger, S. D.: Direct and indirect effects of atmospheric conditions
and soil moisture on surface energy partitioning revealed by a prolonged
drought at a temperate forest site, J. Geophys. Res.-Atmos., 111, 1–13,
10.1029/2006JD007161, 2006.Guan, D.-X., Wu, J.-B., Zhao, X.-S., Han, S.-J., Yu, G.-R., Sun,
X.-M., and Jin, C.-J.: CO2 fluxes over an old, temperate mixed forest
in northeastern China, Agr. Forest Meteorol., 137, 138–149,
10.1016/j.agrformet.2006.02.003, 2006.Guanter, L., Zhang, Y., Jung, M., Joiner, J., Voigt, M., Berry, J. A.,
Frankenberg, C., Huete, A. R., Zarco-Tejada, P., Lee, J.-E., Moran, M. S.,
Ponce-Campos, G., Beer, C., Camps-Valls, G., Buchmann, N., Gianelle, D.,
Klumpp, K., Cescatti, A., Baker, J. M., and Griffis, T. J.: Global and
time-resolved monitoring of crop photosynthesis with chlorophyll
fluorescence, P. Natl. Acad. Sci. USA, 111, 1327–1333,
10.1073/pnas.1320008111,2014.Gurney, K. R., Law, R. M., Denning, A. S., Rayner, P. J., Baker,
D., Bousquet, P., Bruhwiler, L., Chen, Y.-H., Ciais, P., Fan, S.,
Fung, I. Y., Gloor, M., Heimann, M., Higuchi, K., John, J., Maki,
T., Maksyutov, S., Masarie, K., Peylin, P., Prather, M., Pak,
B. C., Randerson, J., Sarmiento, J., Taguchi, S., Takahashi, T., and
Yuen, C.-W.: Towards robust regional estimates of CO2 sources and
sinks using atmospheric transport models, Nature, 415, 626–630,
10.1038/415626a, 2002.Haapanala, S., Rinne, J., Pystynen, K.-H., Hellén, H., Hakola, H., and
Riutta, T.: Measurements of hydrocarbon emissions from a boreal fen using the
REA technique, Biogeosciences, 3, 103–112,
10.5194/bg-3-103-2006, 2006.Hadley, J. L., Kuzeja, P. S., Daley, M. J., Phillips, N. G.,
Mulcahy, T., and Singh, S.: Water use and carbon exchange of red oak- and
eastern hemlock-dominated forests in the northeastern USA: implications for
ecosystem-level effects of hemlock woolly adelgid, Tree Physiol., 28,
615–627, 10.1093/treephys/28.4.615, 2008.Hadley, J. L., O'Keefe, J., Munger, J. W., Hollinger, D. Y., and
Richardson, A. D.: Phenology of Forest-Atmosphere Carbon Exchange for
Deciduous and Coniferous Forests in Southern and Northern New England,
in: Phenology of Ecosystem Processes: Applications in Global Change Research,
edited by: Noormets, A., Springer, 119–141,
10.1007/978-1-4419-0026-5_5, 2009.Halsey, K. H., Milligan, A. J., and Behrenfeld, M. J.: Physiological
optimization underlies growth rate-independent chlorophyll-specific gross and
net primary production, Photosynth. Res., 103, 125–137,
10.1007/s11120-009-9526-z, 2010.Harding, R. J. and Lloyd, C. R.: Evaporation and energy balance of a wet
grassland at Tadham Moor on the Somerset Levels, Hydrol. Proc., 22,
2346–2357, 10.1002/hyp.6829, 2008.
Hastie, T., Tibshirani, R., and Friedman, J.: The Elements of
Statistical Learning: Data Mining, Inference, and Prediction, 2nd edn.,
Springer, 2009.Heilman, J. L., Litvak, M. E., McInnes, K. J., Kjelgaard, J. F.,
Kamps, R. H., and Schwinning, S.: Water-storage capacity controls energy
partitioning and water use in karst ecosystems on the Edwards Plateau,
Texas, Ecohydrology, 7, 127–138, 10.1002/eco.1327, 2014.Hendriks, D. M. D., van Huissteden, J., Dolman, A. J., and van der Molen, M.
K.: The full greenhouse gas balance of an abandoned peat meadow,
Biogeosciences, 4, 411–424, 10.5194/bg-4-411-2007, 2007.Hirano, T., Hirata, R., Fujinuma, Y., Saigusa, N., Yamamoto, S.,
Harazono, Y., Takada, M., Inukai, K., and Inoue, G.: CO2 and
water vapor exchange of a larch forest in northern Japan, Tellus B, 55,
244–257, 10.1034/j.1600-0889.2003.00063.x, 2003.Hirano, T., Segah, H., Harada, T., Limin, S., June, T., Hirata,
R., and Osaki, M.: Carbon dioxide balance of a tropical peat swamp forest
in Kalimantan, Indonesia, Glob. Change Biol., 13, 412–425,
10.1111/j.1365-2486.2006.01301.x, 2007.Hollinger, D. Y., Aber, J., Dail, B., Davidson, E. A., Goltz,
S. M., Hughes, H., Leclerc, M. Y., Lee, J. T., Richardson, A. D.,
Rodrigues, C., Scott, N. A., Achuatavarier, D., and Walsh, J.:
Spatial and temporal variability in forest–atmosphere CO2 exchange,
Glob. Change Biol., 10, 1689–1706, 10.1111/j.1365-2486.2004.00847.x,
2004.Hong, J., Kim, J., Lee, D., and Lim, J.-H.: Estimation of the storage
and advection effects on H2O and CO2 exchanges in a hilly KoFlux
forest catchment, Water Resour. Res., 44, 1–10, 10.1029/2007WR006408,
2008.Houborg, R. M. and Soegaard, H.: Regional simulation of ecosystem
CO2 and water vapor exchange for agricultural land using NOAA AVHRR
and Terra MODIS satellite data. Application to Zealand, Denmark,
Remote Sens. Environ., 93, 150–167, 10.1016/j.rse.2004.07.001, 2004.Humphreys, E. R., Black, T. A., Morgenstern, K., Cai, T., Drewitt,
G. B., Nesic, Z., and Trofymow, J. A.: Carbon dioxide fluxes in coastal
Douglas-fir stands at different stages of development after clearcut
harvesting, Agr. Forest Meteorol., 140, 6–22,
10.1016/j.agrformet.2006.03.018, 2006.Ichii, K., Ueyama, M., Kondo, M., Saigusa, N., Kim, J., Alberto,
M. C., Ardö, J., Euskirchen, E. S., Kang, M., Hirano, T.,
Joiner, J., Kobayashi, H., Marchesini, L. B., Merbold, L., Miyata,
A., Saitoh, T. M., Takagi, K., Varlagin, A., Bret-Harte, M. S.,
Kitamura, K., Kosugi, Y., Kotani, A., Kumar, K., Li, S. G.,
Machimura, T., Matsuura, Y., Mizoguchi, Y., Ohta, T., Mukherjee,
S., Yanagi, Y., Yasuda, Y., Zhang, Y., and Zhao, F.: New data-driven
estimation of terrestrial CO2 fluxes in Asia using a standardized
database of eddy covariance measurements, remote sensing data, and support
vector regression, J. Geophys. Res.-Biogeo., 122, 767–795,
10.1002/2016JG003640, 2017.Irvine, J., Law, B. E., and Hibbard, K. A.: Postfire carbon pools and
fluxes in semiarid ponderosa pine in Central Oregon, Glob. Change Biol.,
13, 1748–1760, 10.1111/j.1365-2486.2007.01368.x, 2007.Jacobs, C. M. J., Jacobs, A. F. G., Bosveld, F. C., Hendriks, D. M. D.,
Hensen, A., Kroon, P. S., Moors, E. J., Nol, L., Schrier-Uijl, A., and
Veenendaal, E. M.: Variability of annual CO2 exchange from Dutch
grasslands, Biogeosciences, 4, 803–816,
10.5194/bg-4-803-2007, 2007.Jenkins, J. P., Richardson, A. D., Braswell, B. H., Ollinger, S. V.,
Hollinger, D. Y., and Smith, M.-L.: Refining light-use efficiency
calculations for a deciduous forest canopy using simultaneous tower-based
carbon flux and radiometric measurements, Agr. Forest Meteorol., 143, 64–79,
10.1016/j.agrformet.2006.11.008, 2007.Jung, M., Reichstein, M., and Bondeau, A.: Towards global empirical upscaling
of FLUXNET eddy covariance observations: validation of a model tree ensemble
approach using a biosphere model, Biogeosciences, 6, 2001–2013,
10.5194/bg-6-2001-2009, 2009.Jung, M., Reichstein, M., Ciais, P., Seneviratne, S. I., Sheffield, J.,
Goulden, M. L., Bonan, G., Cescatti, A., Chen, J., de Jeu, R., Dolman,
A. J., Eugster, W., Gerten, D., Gianelle, D., Gobron, N., Heinke, J.,
Kimball, J., Law, B. E., Montagnani, L., Mu, Q., Mueller, B., Oleson, K.,
Papale, D., Richardson, A. D., Roupsard, O., Running, S., Tomelleri, E.,
Niovy, N., Weber, U., Williams, C., Wood, E., Zaehle, S., and Zhang, K.:
Recent decline in the global land evapotranspiration trend due to limited
moisture supply, Nature, 467, 951–954, 10.1038/nature09396, 2010.Jung, M., Reichstein, M., Margolis, H. A., Cescatti, A.,
Richardson, A. D., Arain, M. A., Arneth, A., Bernhofer, C., Bonal,
D., Chen, J., Gianelle, D., Gobron, N., Kiely, G., Kutsch, W.,
Lasslop, G., Law, B. E., Lindroth, A., Merbold, L., Montagnani, L.,
Moors, E. J., Papale, D., Sottocornola, M., Vaccari, F., and
Williams, C.: Global patterns of land-atmosphere fluxes of carbon dioxide,
latent heat, and sensible heat derived from eddy covariance, satellite, and
meteorological observations, J. Geophys. Res.-Biogeo., 116, G00J07,
10.1029/2010JG001566, 2011.Jung, M., Reichstein, M., Schwalm, C. R., Huntingford, C., Sitch,
S., Ahlström , A., Arneth, A., Camps-Valls, G., Ciais, P.,
Friedlingstein, P., Gans, F., Ichii, K., Jain, A. K., Kato, E.,
Papale, D., Poulter, B., Raduly, B., Rödenbeck, C., Tramontana,
G., Viovy, N., Wang, Y.-P., Weber, U., Zaehle, S., and Zeng, N.:
Compensatory water effects link yearly global land CO2 sink changes to
temperature, Nature, 541, 516–520, 10.1038/nature20780, 2017.Kilinc, M., Beringer, J., Hutley, L. B., Haverd, V., and Tapper,
N. J.: An analysis of the surface energy budget above the world's tallest
angiosperm forest, Agr. Forest Meteorol., 166, 23–31,
10.1016/j.agrformet.2012.05.014, 2012.Knohl, A., Schulze, E.-D., Kolle, O., and Buchmann, N.: Large carbon
uptake by an unmanaged 250-year-old deciduous forest in Central Germany,
Agr. Forest Meteorol., 118, 151–167, 10.1016/S0168-1923(03)00115-1,
2003.Krishnan, P., Meyers, T. P., Scott, R. L., Kennedy, L., and Heuer,
M.: Energy exchange and evapotranspiration over two temperate semi-arid
grasslands in North America, Agr. Forest Meteorol., 153, 31–44,
10.1016/j.agrformet.2011.09.017, 2012.Kurbatova, J., Arneth, A., Vygodskaya, N. N., Kolle, O., Varlargin,
A. V., Milyukova, I. M., Tchebakova, N. M., Schulze, E.-D., and
Lloyd, J.: Comparative ecosystem-atmosphere exchange of energy and mass in
a European Russian and a central Siberian bog I, Interseasonal and
interannual variability of energy and latent heat fluxes during the snowfree
period, Tellus B, 54, 497–513, 10.1034/j.1600-0889.2002.01354.x, 2002.Kurbatova, J., Li, C., Varlagin, A., Xiao, X., and Vygodskaya, N.: Modeling
carbon dynamics in two adjacent spruce forests with different soil conditions
in Russia, Biogeosciences, 5, 969–980,
10.5194/bg-5-969-2008, 2008.Kutsch, W. L., Aubinet, M., Buchmann, N., Smith, P., Osborne, B.,
Eugster, W., Wattenbach, M., Schrumpf, M., Schulze, E. D.,
Tomelleri, E., Ceschia, E., Bernhofer, C., Beziat, P., Carrara, A.,
Tommasi, P. D., Grünwald, T., Jones, M., Magliulo, V., Marloie,
O., Moureaux, C., Olioso, A., Sanz, M. J., Saunders, M., Sogaard,
H., and Ziegler, W.: The net biome production of full crop rotations in
Europe, Agr. Ecosyst. Environ., 139, 336–345,
10.1016/j.agee.2010.07.016, 2010.Kwon, H.-J., Oechel, W. C., Zulueta, R. C., and Hastings, S. J.:
Effects of climate variability on carbon sequestration among adjacent wet
sedge tundra and moist tussock tundra ecosystems, J. Geophys. Res.-Biogeo.,
111, 1–18, 10.1029/2005JG000036, 2006.Lafleur, P. M., Roulet, N. T., Bubier, J. L., Frolking, S., and
Moore, T. R.: Interannual variability in the peatland-atmosphere carbon
dioxide exchange at an ombrotrophic bog, Global Biogeochem. Cy., 17, 1–14,
10.1029/2002GB001983, 2003.Lagergren, F., Lindroth, A., Dellwik, E., Ibrom, A., Lankreijer,
H., Launiainen, S., Mölder, M., Kolari, P., Pilegaard, K., and
Vesala, T.: Biophysical controls on CO2 fluxes of three Northern
forests based on long-term eddy covariance data, Tellus B, 60, 143–152,
10.1111/j.1600-0889.2006.00324.x, 2008.Lasslop, G., Reichstein, M., Papale, D., Richardson, A., Arneth, A., Barr,
A., Stoy, P., and Wohlfahrt, G.: Separation of net ecosystem exchange into
assimilation and respiration using a light response curve approach: critical
issues and global evaluation, Glob. Change Biol., 16, 187–208,
10.1111/j.1365-2486.2009.02041.x, 2010.Launiainen, S., Katul, G. G., Kolari, P., Lindroth, A., Lohila, A.,
Aurela, M., Varlagin, A., Grelle, A., and Vesala, T.: Do the energy
fluxes and surface conductance of boreal coniferous forests in Europe scale
with leaf area?, Glob. Change Biol., 22, 4096–4113, 10.1111/gcb.13497,
2016.Law, B. E., Thornton, P. E., Irvine, J., Anthoni, P. M., and Tuyl,
S. V.: Carbon storage and fluxes in ponderosa pine forests at different
developmental stages, Glob. Change Biol., 7, 755–777,
10.1046/j.1354-1013.2001.00439.x, 2001.
Lee, H. C., Hong, J., Cho, C.-H., Choi, B.-C., Oh, S.-N., and
Kim, J.: Surface Exchange of Energy and Carbon Dioxide between the
Atmosphere and a Farmland in Haenam, Korea, Korean J. Agr. Forest
Meteorol., 5, 61–69, 2003.Lipson, D. A., Wilson, R. F., and Oechel, W. C.: Effects of elevated
atmospheric CO2 on soil microbial biomass, activity, and diversity in a
chaparral ecosystem, Appl. Environ. Microb., 71, 8573–8580,
10.1128/AEM.71.12.8573-8580.2005, 2005.Liu, C., Zhang, Z., Sun, G., Zha, T., Zhu, J., Shen, L., Chen,
J., Fang, X., and Chen, J.: Quantifying evapotranspiration and
biophysical regulations of a poplar plantation assessed by eddy covariance
and sap-flow methods, J. Plant Ecol., 33, 706–718,
10.3773/j.issn.1005-264x.2009.04.009, 2009.Liu, H., Randerson, J. T., Lindfors, J., and Chapin, F. S.: Changes
in the surface energy budget after fire in boreal ecosystems of interior
Alaska: An annual perspective, J. Geophys. Res.-Atmos., 110, 1–12,
10.1029/2004JD005158, 2005.Loubet, B., Laville, P., Lehuger, S., Larmanou, E., Fléchard,
C., Mascher, N., Genermont, S., Roche, R., Ferrara, R. M., Stella,
P., Personne, E., Durand, B., Decuq, C., Flura, D., Masson, S.,
Fanucci, O., Rampon, J.-N., Siemens, J., Kindler, R., Gabrielle,
B., Schrumpf, M., and Cellier, P.: Carbon, nitrogen and Greenhouse
gases budgets over a four years crop rotation in northern France, Plant
Soil, 343, 109–137, 10.1007/s11104-011-0751-9, 2011.Ma, S., Baldocchi, D. D., Xu, L., and Hehn, T.: Inter-annual
variability in carbon dioxide exchange of an oak/grass savanna and open
grassland in California, Agr. Forest Meteorol., 147, 157–171,
10.1016/j.agrformet.2007.07.008, 2007.Marcolla, B. and Cescatti, A.: Experimental analysis of flux footprint
for varying stability conditions in an alpine meadow, Agr. Forest Meteorol.,
135, 291–301, 10.1016/j.agrformet.2005.12.007, 2005.Marcolla, B., Cescatti, A., Montagnani, L., Manca, G.,
Kerschbaumer, G., and Minerbi, S.: Importance of advection in the
atmospheric CO2 exchanges of an alpine forest, Agr. Forest Meteorol.,
130, 193–206, 10.1016/j.agrformet.2005.03.006, 2005.Marek, M. V., Janouš, D., Taufarová, K., Havránková,
K., Pavelka, M., Kaplan, V., and Marková, I.: Carbon exchange
between ecosystems and atmosphere in the Czech Republic is affected by
climate factors, Environ. Pollut., 159, 1035–1039,
10.1016/j.envpol.2010.11.025, 2011.Matese, A., Alberti, G., Gioli, B., Toscano, P., Vaccari, F., and
Zaldei, A.: Compact_Eddy: A compact, low consumption remotely
controlled eddy covariance logging system, Comput. Electron. Agr., 64,
343–346, 10.1016/j.compag.2008.07.002, 2008.McCaughey, J. H., Pejam, M. R., Arain, M. A., and Cameron, D. A.:
Carbon dioxide and energy fluxes from a boreal mixedwood forest ecosystem in
Ontario, Canada, Agr. Forest Meteorol., 140, 79–96,
10.1016/j.agrformet.2006.08.010, 2006.
Meinshausen, N.: Quantile regression forests, J. Mach. Learn. Res., 7,
983–999, 2006.Merbold, L., Ardö, J., Arneth, A., Scholes, R. J., Nouvellon, Y., de
Grandcourt, A., Archibald, S., Bonnefond, J. M., Boulain, N., Brueggemann,
N., Bruemmer, C., Cappelaere, B., Ceschia, E., El-Khidir, H. A. M., El-Tahir,
B. A., Falk, U., Lloyd, J., Kergoat, L., Le Dantec, V., Mougin, E., Muchinda,
M., Mukelabai, M. M., Ramier, D., Roupsard, O., Timouk, F., Veenendaal, E.
M., and Kutsch, W. L.: Precipitation as driver of carbon fluxes in 11 African
ecosystems, Biogeosciences, 6, 1027–1041,
10.5194/bg-6-1027-2009, 2009.Meyers, T. P. and Hollinger, S. E.: An assessment of storage terms in the
surface energy balance of maize and soybean, Agr. Forest Meteorol., 125,
105–115, 10.1016/j.agrformet.2004.03.001, 2004.Migliavacca, M., Meroni, M., Manca, G., Matteucci, G., Montagnani,
L., Grassi, G., Zenone, T., Teobaldelli, M., Goded, I., Colombo,
R., and Seufert, G.: Seasonal and interannual patterns of carbon and water
fluxes of a poplar plantation under peculiar eco-climatic conditions, Agr.
Forest Meteorol., 149, 1460–1476, 10.1016/j.agrformet.2009.04.003,
2009.
Mitchell, T. M.: Machine Learning, McGraw–Hill, 1997.Mkhabela, M. S., Amiro, B. D., Barr, A. G., Black, T. A.,
Hawthorne, I., Kidston, J., McCaughey, J. H., Orchansky, A. L.,
Nesic, Z., Sass, A., Shashkov, A., and Zha, T.: Comparison of carbon
dynamics and water use efficiency following fire and harvesting in Canadian
boreal forests, Agr. Forest Meteorol., 149, 783–794,
10.1016/j.agrformet.2008.10.025, 2009.Monson, R. K., Turnipseed, A. A., Sparks, J. P., Harley, P. C.,
Scott-Denton, L. E., Sparks, K., and Huxman, T. E.: Carbon
sequestration in a high-elevation, subalpine forest, Glob. Change Biol., 8,
459–478, 10.1046/j.1365-2486.2002.00480.x, 2002.Moors, E. J., Jacobs, C., Jans, W., Supit, I., Kutsch, W. L.,
Bernhofer, C., Béziat, P., Buchmann, N., Carrara, A., Ceschia,
E., Elbers, J., Eugster, W., Kruijt, B., Loubet, B., Magliulo, E.,
Moureaux, C., Olioso, A., Saunders, M., and Soegaard, H.: Variability
in carbon exchange of European croplands, Agr. Ecosyst. Environ., 139,
325–335, 10.1016/j.agee.2010.04.013, 2010.Moureaux, C., Debacq, A., Bodson, B., Heinesch, B., and Aubinet,
M.: Annual net ecosystem carbon exchange by a sugar beet crop, Agr. Forest
Meteorol., 139, 25–39, 10.1016/j.agrformet.2006.05.009, 2006.
Murphy, K. P.: Machine Learning: A Probabilistic Perspective, The MIT
Press, 2012.Mutanga, O., Adam, E., and Cho, M. A.: High density biomass estimation
for wetland vegetation using WorldView-2 imagery and random forest
regression algorithm, Int. J. Appl. Earth Obs., 18, 399–406,
10.1016/j.jag.2012.03.012, 2012.Nagy, Z., Pintér, K., Czóbel, S., Balogh, J., Horváth,
L., Fóti, S., Barcza, Z., Weidinger, T., Csintalan, Z., Dinh,
N. Q., Grosz, B., and Tuba, Z.: The carbon budget of semi-arid grassland
in a wet and a dry year in Hungary, Agr. Ecosyst. Environ., 121, 21–29,
10.1016/j.agee.2006.12.003, 2007.Nash, J. E. and Sutcliffe, J. V.: River flow forecasting through
conceptual models part I – A discussion of principles, J. Hydrol., 10,
282–290, 10.1016/0022-1694(70)90255-6, 1970.Noormets, A., McNulty, S. G., Domec, J.-C., Gavazzi, M., Sun, G.,
and King, J. S.: The role of harvest residue in rotation cycle carbon
balance in loblolly pine plantations. Respiration partitioning approach,
Glob. Change Biol., 18, 3186–3201, 10.1111/j.1365-2486.2012.02776.x,
2012.Owen, K. E., Tenhunen, J., Reichstein, M., Wang, Q., Falge, E.,
Geyer, R., Xiao, X., PaulStoy, Ammann, C., Arain, A., Aubinet,
M., Aurela, M., Bernhofer, C., Chojnicki, B. H., Granier, A.,
Gruenwald, T., Hadley, J., Heinesch, B., Hollinger, D., Knohl, A.,
Kutsch, W., Lohila, A., Meyers, T., Moors, E., Moureaux, C.,
Pilegaard, K., Saigusa, N., Verma, S., Vesala, T., and Vogel, C.:
Linking flux network measurements to continental scale simulations: ecosystem
carbon dioxide exchange capacity under non-water-stressed conditions, Glob.
Change Biol., 13, 734–760, 10.1111/j.1365-2486.2007.01326.x, 2007.Papale, D., Black, T. A., Carvalhais, N., Cescatti, A., Chen, J., Jung, M.,
Kiely, G., Lasslop, G., Mahecha, M. D., Margolis, H., Merbold, L.,
Montagnani, L., Moors, E., Olesen, J. E., Reichstein, M., Tramontana, G., van
Gorsel, E., Wohlfahrt, G., and Ráduly, B.: Effect of spatial sampling
from European flux towers for estimating carbon and water fluxes with
artificial neural networks, J. Geophys. Res.-Biogeo., 120, 1941–1957,
10.1002/2015JG002997, 2015.Peichl, M., Brodeur, J. J., Khomik, M., and Arain, M. A.: Biometric
and eddy-covariance based estimates of carbon fluxes in an age-sequence of
temperate pine forests, Agr. Forest Meteorol., 150, 952–965,
10.1016/j.agrformet.2010.03.002, 2010.Peichl, M., Leahy, P., and Kiely, G.: Six-year stable annual uptake of
carbon dioxide in intensively managed humid temperate grassland, Ecosystems,
14, 112–126, 10.1007/s10021-010-9398-2, 2011.Pereira, J. S., Mateus, J. A., Aires, L. M., Pita, G., Pio, C., David, J. S.,
Andrade, V., Banza, J., David, T. S., Paço, T. A., and Rodrigues, A.: Net
ecosystem carbon exchange in three contrasting Mediterranean ecosystems –
the effect of drought, Biogeosciences, 4, 791–802,
10.5194/bg-4-791-2007, 2007.Perez-Priego, O., El-Madany, T. S., Migliavacca, M., Kowalski, A. S.,
Jung, M., Carrara, A., Kolle, O., Martín, M. P.,
Pacheco-Labrador, J., Moreno, G., and Reichstein, M.: Evaluation of
eddy covariance latent heat fluxes with independent lysimeter and sapflow
estimates in a Mediterranean savannah ecosystem, Agr. Forest Meteorol.,
236, 87–99, 10.1016/j.agrformet.2017.01.009, 2017.Peters, W., Jacobson, A. R., Sweeney, C., Andrews, A. E., Conway,
T. J., Masarie, K., Miller, J. B., Bruhwiler, L. M. P., Petron, G.,
Hirsch, A. I., Worthy, D. E. J., Werf, G. R. v. d., Randerson, J. T.,
Wennberg, P. O., Krol, M. C., and Tans, P. P.: An atmospheric
perspective on North American carbon dioxide exchange: CarbonTracker,
P. Natl. Acad. Sci. USA, 104, 18925–18930, 10.1073/pnas.0708986104,
2007.Peters, W., Krol, M. C., Werf, G. R. v. d., Houweling, S., Jones,
C. D., Hughes, J., Schaefer, K., Masarie, K. A., Jacobson, A. R.,
Miller, J. B., Cho, C. H., Ramonet, M., Schmidt, M., Ciattaglia,
L., Apadula, F., Heltai, D., Meinhardt, F., Sarra, A. G. D.,
Piacentino, S., Sferlazzo, D., Aalto, T., Hatakka, J., Ström,
J., Haszpra, L., Meijer, H. A. J., Laan, S. v. d., Neubert, R. E. M.,
Jordan, A., Rodo, X., Morgui, J.-A., Vermeulen, A. T., Popa, E.,
Rozanski, K., Zimnoch, M., Manning, A. C., Leuenberger, M.,
Uglietti, C., Dolman, A. J., Ciais, P., Heimann, M., and Tans,
P. P.: Seven years of recent European net terrestrial carbon dioxide
exchange constrained by atmospheric observations, Glob. Change Biol., 16,
1317–1337, 10.1111/j.1365-2486.2009.02078.x, 2010.Peylin, P., Law, R. M., Gurney, K. R., Chevallier, F., Jacobson, A. R., Maki,
T., Niwa, Y., Patra, P. K., Peters, W., Rayner, P. J., Rödenbeck, C., van
der Laan-Luijkx, I. T., and Zhang, X.: Global atmospheric carbon budget:
results from an ensemble of atmospheric CO2 inversions, Biogeosciences,
10, 6699-6720, 10.5194/bg-10-6699-2013, 2013Pilegaard, K., Hummelshøj, P., Jensen, N. O., and Chen, Z.: Two
years of continuous CO2 eddy-flux measurements over a Danish beech
forest, Agr. Forest Meteorol., 107, 29–41,
10.1016/S0168-1923(00)00227-6, 2001.Powell, T. L., Bracho, R., Li, J., Dore, S., Hinkle, C. R., and
Drake, B. G.: Environmental controls over net ecosystem carbon exchange of
scrub oak in central Florida, Agr. Forest Meteorol., 141, 19–34,
10.1016/j.agrformet.2006.09.002, 2006.Powell, T. L., Gholz, H. L., Clark, K. L., Starr, G., Cropper Jr.,
W. P., and Martin, T. A.: Carbon exchange of a mature, naturally
regenerated pine forest in north Florida, Glob. Change Biol., 14,
2523–2538, 10.1111/j.1365-2486.2008.01675.x, 2008.
Rasmussen, C. E. and Williams, C. K. I.: Gaussian Processes for Machine
Learning, The MIT Press, 2006.Reichstein, M., Rey, A., Freibauer, A., Tenhunen, J., Valentini,
R., Banza, J., Casals, P., Cheng, Y., Grünzweig, J. M., Irvine,
J., Joffre, R., Law, B. E., Loustau, D., Miglietta, F., Oechel, W.,
Ourcival, J.-M., Pereira, J. S., Peressotti, A., Ponti, F., Qi, Y.,
Rambal, S., Rayment, M., Romanya, J., Rossi, F., Tedeschi, V.,
Tirone, G., Xu, M., and Yakir, D.: Modeling temporal and large-scale
spatial variability of soil respiration from soil water availability,
temperature and vegetation productivity indices, Global Biogeochem. Cy., 17,
1–15, 10.1029/2003GB002035, 2003.Reichstein, M., Falge, E., Baldocchi, D., Papale, D., Aubinet, M., Berbigier,
P., Bernhofer, C., Buchmann, N., Gilmanov, T., Granier, A., Grünwald, T.,
Havránková, K., Ilvesniemi, H., Janous, D., Knohl, A., Laurila, T.,
Lohila, A., Loustau, D., Matteucci, G., Meyers, T., Miglietta, F., Ourcival,
J.-M., Pumpanen, J., Rambal, S., Rotenberg, E., Sanz, M., Seufert, G.,
Tenhunen, J., Vaccari, F., Vesala, T., Yakir, D., and Valentini, R.: On the
separation of net ecosystem exchange into assimilation and ecosystem
respiration: review and improved algorithm, Glob. Change Biol., 11,
1424–1439, 10.1111/j.1365-2486.2005.001002.x, 2005.Reichstein, M., Bahn, M., Mahecha, M. D., Kattge, J., and
Baldocchi, D. D.: Linking plant and ecosystem functional biogeography, P.
Natl. Acad. Sci. USA, 111, 13697–13702, 10.1073/pnas.1216065111, 2014.Restrepo-Coupe, N., Rocha, H. R. d., Hutyra, L. R., Araujo, A. C. d.,
Borma, L. S., Christoffersen, B., Cabral, O. M. R., Camargo, P.
B. d., Cardoso, F. L., Costa, A. C. L. d., Fitzjarrald, D. R.,
Goulden, M. L., Kruijt, B., Maia, J. M., Malhi, Y. S., Manzi,
A. O., Miller, S. D., Nobre, A. D., Randow, C. v., Sa, L. D. A.,
Sakai, R. K., Tota, J., Wofsy, S. C., Zanchi, F. B., and Saleska,
S. R.: What drives the seasonality of photosynthesis across the Amazon
basin? A cross-site analysis of eddy flux tower measurements from the
Brasil flux network, Agr. Forest Meteorol., 182–183, 128–144,
10.1016/j.agrformet.2013.04.031, 2013.Rey, A., Pegoraro, E., Tedeschi, V., Parri, I. D., Jarvis, P. G.,
and Valentini, R.: Annual variation in soil respiration and its components
in a coppice oak forest in Central Italy, Glob. Change Biol., 8,
851–866, 10.1046/j.1365-2486.2002.00521.x, 2002.Rödenbeck, C.: Estimating CO2 sources and sinks from atmospheric
mixing ratio measurements using a global inversion of atmospheric transport,
Tech. Rep. 6, Max Planck Institute for Biogeochemistry, Jena, available at:
http://www.bgc-jena.mpg.de/~christian.roedenbeck/download/2005-Roedenbeck-TechReport6.pdf
(last access: 17 July 2018), 2005.Rodrigues, A., Pita, G., Mateus, J., Kurz-Besson, C., Casquilho,
M., Cerasoli, S., Gomes, A., and Pereira, J.: Eight years of continuous
carbon fluxes measurements in a Portuguese eucalypt stand under two main
events: Drought and felling, Agr. Forest Meteorol., 151, 493–507,
10.1016/j.agrformet.2010.12.007, 2011.Roupsard, O., Bonnefond, J.-M., Irvine, M., Berbigier, P.,
Nouvellon, Y., Dauzat, J., Taga, S., Hamel, O., Jourdan, C.,
Saint-André, L., Mialet-Serra, I., Labouisse, J.-P., Epron, D.,
Joffre, R., Braconnier, S., Rouzière, A., Navarro, M., and
Bouillet, J.-P.: Partitioning energy and evapo-transpiration above and
below a tropical palm canopy, Agr. Forest Meteorol., 139, 252–268,
10.1016/j.agrformet.2006.07.006, 2006.Sagerfors, J., Lindroth, A., Grelle, A., Klemedtsson, L., Weslien,
P., and Nilsson, M.: Annual CO2 exchange between a nutrient-poor,
minerotrophic, boreal mire and the atmosphere, J. Geophys. Res.-Biogeo., 113,
1–15, 10.1029/2006JG000306, 2008.Saito, M., Miyata, A., Nagai, H., and Yamada, T.: Seasonal variation
of carbon dioxide exchange in rice paddy field in Japan, Agr. Forest
Meteorol., 135, 93–109, 10.1016/j.agrformet.2005.10.007, 2005.Schindler, D., Türk, M., and Mayer, H.: CO2 fluxes of a
Scots pine forest growing in the warm and dry southern upper Rhine plain,
SW Germany, Eur. J. Forest Res., 125, 201–212,
10.1007/s10342-005-0107-1, 2006.Schmid, H. P., Grimmond, C. S. B., Cropley, F., Offerle, B., and
Su, H.-B.: Measurements of CO2 and energy fluxes over a mixed hardwood
forest in the mid-western United States, Agr. Forest Meteorol., 103,
357–374, 10.1016/S0168-1923(00)00140-4, 2000.Scholes, R., Gureja, N., Giannecchinni, M., Dovie, D., Wilson, B.,
Davidson, N., Piggott, K., McLoughlin, C., Velde, K. v. d.,
Freeman, A., Bradley, S., Smart, R., and Ndala, S.: The environment
and vegetation of the flux measurement site near Skukuza, Kruger
National Park, Koedoe, 44, 73–83, 10.4102/koedoe.v44i1.187, 2001.Scott, R. L., Jenerette, G. D., Potts, D. L., and Huxman, T. E.:
Effects of seasonal drought on net carbon dioxide exchange from a
woody-plant-encroached semiarid grassland, J. Geophys. Res.-Biogeo., 114,
1–13, 10.1029/2008JG000900, 10.1029/2008JG000900, 2009.Scott, R. L., Hamerlynck, E. P., Jenerette, G. D., Moran, M. S., and
Barron-Gafford, G. A.: Carbon dioxide exchange in a semidesert grassland
through drought-induced vegetation change, J. Geophys. Res.-Biogeo., 115,
1–12, 10.1029/2010JG001348, 2010.Smallman, T. L., Moncrieff, J. B., and Williams, M.: WRFv3.2-SPAv2:
development and validation of a coupled ecosystem–atmosphere model, scaling
from surface fluxes of CO2 and energy to atmospheric profiles, Geosci.
Model Dev., 6, 1079–1093, 10.5194/gmd-6-1079-2013, 2013Soegaard, H., Jensen, N. O., Boegh, E., Hasager, C. B., Schelde,
K., and Thomsen, A.: Carbon dioxide exchange over agricultural landscape
using eddy correlation and footprint modelling, Agr. Forest Meteorol., 114,
153–173, 10.1016/S0168-1923(02)00177-6, 2003.Soussana, J. F., Allard, V., Pilegaard, K., Ambus, P., Amman, C.,
Campbell, C., Ceschia, E., Clifton-Brown, J., Czobel, S.,
Domingues, R., Flechard, C., Fuhrer, J., Hensen, A., Horvath, L.,
Jones, M., Kasper, G., Martin, C., Nagy, Z., Neftel, A., Raschi,
A., Baronti, S., Rees, R. M., Skiba, U., Stefani, P., Manca, G.,
Sutton, M., Tuba, Z., and Valentini, R.: Full accounting of the
greenhouse gas (CO2, N2O, CH4) budget of nine European grassland
sites, Agr. Ecosyst. Environ., 121, 121–134,
10.1016/j.agee.2006.12.022, 2007.Spano, D., Snyder, R. L., Sirca, C., and Duce, P.: ECOWAT–A
model for ecosystem evapotranspiration estimation, Agr. Forest Meteorol.,
149, 1584–1596, 10.1016/j.agrformet.2009.04.011, 2009.Stoy, P. C., Katul, G. G., Siqueira, M. B. S., Juang, J.-Y.,
Novick, K. A., McCarthy, H. R., Oishi, A. C., Uebelherr, J. M., and
an Ram Oren, H.-S. K.: Separating the effects of climate and vegetation on
evapotranspiration along a successional chronosequence in the southeastern
US, Glob. Change Biol., 12, 2115–2135,
10.1111/j.1365-2486.2006.01244.x, 2006.Sturm, P., Leuenberger, M., Moncrieff, J., and Ramonet, M.:
Atmospheric O2, CO2 and δ13C measurements from aircraft
sampling over Griffin Forest, Perthshire, UK, Rapid Commun. Mass Sp.,
19, 2399–2406, 10.1002/rcm.2071, 2005.Sulman, B. N., Desai, A. R., Cook, B. D., Saliendra, N., and Mackay, D. S.:
Contrasting carbon dioxide fluxes between a drying shrub wetland in Northern
Wisconsin, USA, and nearby forests, Biogeosciences, 6, 1115–1126,
10.5194/bg-6-1115-2009, 2009.Sun, Y., Fu, R., Dickinson, R., Joiner, J., Frankenberg, C., Gu, L., Xia, Y.,
and Fernando, N.: Drought onset mechanisms revealed by satellite
solar-induced chlorophyll fluorescence: Insights from two contrasting extreme
events, J. Geophys. Res.-Biogeo., 120, 2427–2440,
10.1002/2015JG003150, 2015.
Suni, T., Rinne, J., Reissell, A., Altimir, N., Keronen, P.,
Rannik, Ü., Maso, M. D., Kulmala, M., and Vesala, T.: Long-term
measurements of surface fluxes above a Scots pine forest in Hyytiala,
southern Finland, 1996–2001, Boreal Environ. Res., 8, 287–302, 2003.Syed, K. H., Flanagan, L. B., Carlson, P. J., Glenn, A. J., and
Gaalen, K. E. V.: Environmental control of net ecosystem CO2 exchange
in a treed, moderately rich fen in northern Alberta, Agr. Forest Meteorol.,
140, 97–114, 10.1016/j.agrformet.2006.03.022, 2006.Tedeschi, V., Rey, A., Manca, G., Valentini, R., Jarvis, P. G., and
Borghetti, M.: Soil respiration in a Mediterranean oak forest at
different developmental stages after coppicing, Glob. Change Biol., 12,
110–121, 10.1111/j.1365-2486.2005.01081.x, 2006.Thum, T., Aalto, T., Laurila, T., Aurela, M., Kolari, P., and
Hari, P.: Parametrization of two photosynthesis models at the canopy scale
in a northern boreal Scots pine forest, Tellus B, 59, 874–890,
10.1111/j.1600-0889.2007.00305.x, 2007.Tramontana, G., Jung, M., Schwalm, C. R., Ichii, K., Camps-Valls, G.,
Ráduly, B., Reichstein, M., Arain, M. A., Cescatti, A., Kiely, G.,
Merbold, L., Serrano-Ortiz, P., Sickert, S., Wolf, S., and Papale, D.:
Predicting carbon dioxide and energy fluxes across global FLUXNET sites with
regression algorithms, Biogeosciences, 13, 4291–4313,
10.5194/bg-13-4291-2016, 2016.Ueyama, M., Ichii, K., Iwata, H., Euskirchen, E. S., Zona, D.,
Rocha, A. V., Harazono, Y., Iwama, C., Nakai, T., and Oechel,
W. C.: Upscaling terrestrial carbon dioxide fluxes in Alaska with satellite
remote sensing and support vector regression, J. Geophys. Res.-Biogeo., 118,
1266–1281, 10.1002/jgrg.20095, 2013.Urbanski, S., Barford, C., Wofsy, S., Kucharik, C., Pyle, E.,
Budney, J., McKain, K., Fitzjarrald, D., Czikowsky, M., and Munger,
J. W.: Factors controlling CO2 exchange on timescales from hourly to
decadal at Harvard Forest, J. Geophys. Res.-Biogeo., 112, 1–25,
10.1029/2006JG000293, 2007.Vaccari, F. P., Lugato, E., Gioli, B., D'Acqui, L., Genesio, L.,
Toscano, P., Matese, A., and Miglietta, F.: Land use change and soil
organic carbon dynamics in Mediterranean agro-ecosystems: The case study
of Pianosa Island, Geoderma, 175–176, 29–36,
10.1016/j.geoderma.2012.01.021, 2012.Valentini, R., Angelis, P. d., Matteucci, G., Monaco, R., Dore, S.,
and Scarascia Mucnozza, G. E.: Seasonal net carbon dioxide exchange of a
beech forest with the atmosphere, Glob. Change Biol., 2, 199–207,
10.1111/j.1365-2486.1996.tb00072.x, 1996.van der Molen, M. K., van Huissteden, J., Parmentier, F. J. W., Petrescu, A.
M. R., Dolman, A. J., Maximov, T. C., Kononov, A. V., Karsanaev, S. V., and
Suzdalov, D. A.: The growing season greenhouse gas balance of a continental
tundra site in the Indigirka lowlands, NE Siberia, Biogeosciences, 4,
985–1003, 10.5194/bg-4-985-2007, 2007.Verma, S. B., Dobermann, A., Cassman, K. G., Walters, D. T., Knops,
J. M., Arkebauer, T. J., Suyker, A. E., Burba, G. G., Amos, B.,
Yang, H., Ginting, D., Hubbard, K. G., Gitelson, A. A., and
Walter-Shea, E. A.: Annual carbon dioxide exchange in irrigated and rainfed
maize-based agroecosystems, Agr. Forest Meteorol., 131, 77–96,
10.1016/j.agrformet.2005.05.003, 2005.Vickers, D., Thomas, C., and Law, B. E.: Random and systematic CO2
flux sampling errors for tower measurements over forests in the convective
boundary layer, Agr. Forest Meteorol., 149, 73–83,
10.1016/j.agrformet.2008.07.005, 2009.Walter, A., Christ, M. M., Barron-gafford, G. A., Grieve, K. A.,
Murthy, R., and Rascher, U.: The effect of elevated CO2 on diel leaf
growth cycle, leaf carbohydrate content and canopy growth performance of
Populus deltoides, Glob. Change Biol., 11, 1207–1219,
10.1111/j.1365-2486.2005.00990.x, 2005.Wang, W., Liang, S., and Meyers, T.: Validating MODIS land surface
temperature products using long-term nighttime ground measurements, Remote
Sens. Environ., 112, 623–635, 10.1016/j.rse.2007.05.024, 2008.Wilkinson, M., Eaton, E. L., Broadmeadow, M. S. J., and Morison, J. I. L.:
Inter-annual variation of carbon uptake by a plantation oak woodland in
south-eastern England, Biogeosciences, 9, 5373–5389,
10.5194/bg-9-5373-2012, 2012.Williams, C. A. and Albertson, J. D.: Soil moisture controls on
canopy-scale water and carbon fluxes in an African savanna, Water Resour.
Res., 40, 1–14, 10.1029/2004WR003208, 2004.Wilske, B., Lu, N., Wei, L., Chen, S., Zha, T., Liu, C., Xu,
W., Noormets, A., Huang, J., Wei, Y., Chen, J., Zhang, Z., Ni,
J., Sun, G., Guo, K., McNulty, S., John, R., Han, X., Lin, G.,
and Chen, J.: Poplar plantation has the potential to alter the water
balance in semiarid Inner Mongolia, J. Environ. Manage., 90, 2762–2770,
10.1016/j.jenvman.2009.03.004, 2009.Wilson, T. B. and Meyers, T. P.: Determining vegetation indices from
solar and photosynthetically active radiation fluxes, Agr. Forest Meteorol.,
144, 160–179, 10.1016/j.agrformet.2007.04.001, 2007.Wohlfahrt, G., Anderson-Dunn, M., Bahn, M., Balzarolo, M.,
Berninger, F., Campbell, C., Carrara, A., Cescatti, A.,
Christensen, T., Dore, S., Eugster, W., Friborg, T., Furger, M.,
Gianelle, D., Gimeno, C., Hargreaves, K., Hari, P., Haslwanter, A.,
Johansson, T., Marcolla, B., Milford, C., Nagy, Z., Nemitz, E.,
Rogiers, N., Sanz, M. J., Siegwolf, R. T., Susiluoto, S., Sutton,
M., Tuba, Z., Ugolini, F., Valentini, R., Zorer, R., and Cernusca,
A.: Biotic, Abiotic, and Management Controls on the Net Ecosystem CO2
Exchange of European Mountain Grassland Ecosystems, Ecosystems, 11,
1338–1351, 10.1007/s10021-008-9196-2, 2008a.Wohlfahrt, G., Hammerle, A., Haslwanter, A., Bahn, M., Tappeiner,
U., and Cernusca, A.: Seasonal and inter-annual variability of the net
ecosystem CO2 exchange of a temperate mountain grassland: Effects of
weather and management, J. Geophys. Res.-Atmos., 113, 1–14,
10.1029/2007JD009286, 2008b.
Xiao, J., Zhuang, Q., Baldocchi, D. D., Law, B. E., Richardson,
A. D., Chen, J., Oren, R., Starr, G., Noormets, A., Ma, S.,
Verma, S. B., Wharton, S., Wofsy, S. C., Bolstad, P. V., Burns,
S. P., Cook, D. R., Curtis, P. S., Drake, B. G., Falk, M., Fischer,
M. L., Foster, D. R., Gu, L., Hadley, J. L., Hollinger, D. Y.,
Katul, G. G., Litvak, M., Martin, T., Matamala, R., McNulty, S.,
Meyers, T. P., Monson, R. K., Munger, J. W., Oechel, W. C., Paw, U.
K. T., Schmid, H. P., Scott, R. L., Sun, G., Suyker, A. E., and
Torn, M. S.: Estimation of net ecosystem carbon exchange for the
conterminous United States by combining MODIS and AmeriFlux data,
Agr. Forest Meteorol., 148, 1827–1847, 2008.Xiao, J., Zhuang, Q., Law, B. E., Chen, J., Baldocchi, D. D.,
Cook, D. R., Oren, R., Richardson, A. D., Wharton, S., Ma, S.,
Martin, T. A., Verma, S. B., Suyker, A. E., Scott, R. L., Monson,
R. K., Litvak, M., Hollinger, D. Y., Sun, G., Davis, K. J.,
Bolstad, P. V., Burns, S. P., Curtis, P. S., Drake, B. G., Falk,
M., Fischer, M. L., Foster, D. R., Gu, L., Hadley, J. L., Katul,
G. G., Matamala, R., McNulty, S., Meyers, T. P., Munger, J. W.,
Noormets, A., Oechel, W. C., Paw U, K. T., Schmid, H. P., Starr,
G., Torn, M. S., and Wofsy, S. C.: A continuous measure of gross primary
production for the conterminous United States derived from MODIS and
AmeriFlux data, Remote Sens. Environ., 114, 576–591,
10.1016/j.rse.2009.10.013, 2010.
Xiao, J., Chen, J., Davis, K. J., and Reichstein, M.: Advances in upscaling
of eddy covariance measurements of carbon and water fluxes, J. Geophys.
Res.-Biogeo., 117, G00J01, 10.1029/2011JG001889, 2012.Yamamoto, S., Murayama, S., Saigusa, N., and Kondo, H.: Seasonal and
inter-annual variation of CO2 flux between a temperate forest and the
atmosphere in Japan, Tellus B, 51, 402–413,
10.3402/tellusb.v51i2.16314, 1999.Yan, Y., Zhao, B., Chen, J., Guo, H., Gu, Y., Wu, Q., and Li,
B.: Closing the carbon budget of estuarine wetlands with tower-based
measurements and MODIS time series, Glob. Change Biol., 14, 1690–1702,
10.1111/j.1365-2486.2008.01589.x, 2008.Yang, F., Ichii, K., White, M. A., Hashimoto, H., Michaelis, A. R.,
Votava, P., Zhu, A.-X., Huete, A., Running, S. W., and Nemani,
R. R.: Developing a continental-scale measure of gross primary production by
combining MODIS and AmeriFlux data through Support Vector Machine
approach, Remote Sens. Environ., 110, 109–122,
10.1016/j.rse.2007.02.016, 2007.Yang, Y., Anderson, M., Gao, F., Hain, C., Kustas, W., Meyers,
T., Crow, W., Finocchiaro, R., Otkin, J., Sun, L., and Yang, Y.:
Impact of Tile Drainage on Evapotranspiration in South Dakota, USA,
Based on High Spatiotemporal Resolution Evapotranspiration Time Series From a
Multisatellite Data Fusion System, IEEE J. Sel. Top. Appl., 10, 2550–2564,
10.1109/JSTARS.2017.2680411, 2017.Zhang, W. L., Chen, S. P., Chen, J., Wei, L., Han, X. G., and
Lin, G. H.: Biophysical regulations of carbon fluxes of a steppe and a
cultivated cropland in semiarid Inner Mongolia, Agr. Forest Meteorol.,
146, 216–229, 10.1016/j.agrformet.2007.06.002, 2007.Zscheischler, J., Mahecha, M. D., Avitabile, V., Calle, L., Carvalhais, N.,
Ciais, P., Gans, F., Gruber, N., Hartmann, J., Herold, M., Ichii, K., Jung,
M., Landschützer, P., Laruelle, G. G., Lauerwald, R., Papale, D., Peylin,
P., Poulter, B., Ray, D., Regnier, P., Rödenbeck, C., Roman-Cuesta, R.
M., Schwalm, C., Tramontana, G., Tyukavina, A., Valentini, R., van der Werf,
G., West, T. O., Wolf, J. E., and Reichstein, M.: Reviews and syntheses: An
empirical spatiotemporal description of the global surface–atmosphere carbon
fluxes: opportunities and data limitations, Biogeosciences, 14, 3685–3703,
10.5194/bg-14-3685-2017, 2017.