GMDGeoscientific Model DevelopmentGMDGeosci. Model Dev.1991-9603Copernicus PublicationsGöttingen, Germany10.5194/gmd-9-2499-2016High-resolution land surface fluxes from satellite and reanalysis data (HOLAPS v1.0): evaluation and uncertainty assessmentLoewAlexanderalexander.loew@lmu.dePengJianBorscheMichaelLudwig-Maximilians Universität München, Luisenstr. 37, 80333
Munich, GermanyMax Planck Institute for Meteorology, Bundesstr. 53, 20146 Hamburg,
GermanyDeutscher Wetterdienst, National Climate Monitoring, Frankfurter Str.
135, 63067 Offenbach, GermanyAlexander Loew (alexander.loew@lmu.de)27July201697249925328December201521December20157July20168July2016This work is licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by/3.0/This article is available from https://gmd.copernicus.org/articles/9/2499/2016/gmd-9-2499-2016.htmlThe full text article is available as a PDF file from https://gmd.copernicus.org/articles/9/2499/2016/gmd-9-2499-2016.pdf
Surface water and energy fluxes are essential components of the Earth system.
Surface latent heat fluxes provide major energy input to the atmosphere.
Despite the importance of these fluxes, state-of-the-art data sets of surface
energy and water fluxes largely differ. The present paper introduces a new
framework for the estimation of surface energy and water fluxes at the land
surface, which allows for temporally and spatially high-resolved flux
estimates at the quasi-global scale (50∘ S, 50∘ N)
(High resOlution Land Atmosphere Parameters from Space – HOLAPS v1.0). The framework makes use of existing long-term satellite and
reanalysis data records and ensures internally consistent estimates of the
surface radiation and water fluxes. The manuscript introduces the technical
details of the developed framework and provides results of a comprehensive
sensitivity and evaluation study. Overall the root mean square difference
(RMSD) was found to be 51.2 (30.7) W m-2 for hourly (daily)
latent heat flux, and 84 (38) W m-2 for sensible heat flux when
compared against 48 FLUXNET stations worldwide. The largest uncertainties of
latent heat flux and net radiation were found to result from uncertainties in
the solar radiation flux obtained from satellite data products.
Introduction
Water and energy fluxes between the land surface and atmosphere are essential
components of the Earth system. At the ecosystem scale, the land–atmosphere
fluxes have been mainly measured by a network of flux-tower sites within the
frame of FLUXNET (Baldocchi et al., 2001; Baldocchi, 2008). However, to
generate global data sets of water and energy fluxes, the use of satellite
data as well as models has become indispensable.
Different approaches exist to infer land turbulent surface fluxes by either
one of the following methods (Kalma et al., 2008; Wang and Dickinson, 2012):
(1) simulations by an offline land surface model (Roads and Betts, 2000);
(2) empirical statistical models, e.g., obtained by machine learning
techniques or neural networks (Jung et al., 2011); (3) surface energy balance
models forced either by satellite remote sensing or reanalysis data
(Bastiaanssen et al., 1998; Su, 2002); (4) methods based on Penman–Monteith
or Priestley–Taylor equations (Fisher et al., 2008; Miralles et al.,
2011; Mu et al., 2007; Zhang et al., 2015); and (5) spatial variability
methods (Roerink et al., 2000; Peng et al., 2013b; Peng and Loew, 2014).
Novel long-term satellite data records as well as increasing computing
capacities allow one to generate spatially (< 10 km) and temporally
(< 3 h) high-resolved estimates of surface fluxes at the global scale.
The currently existing global data sets have spatial resolutions between 0.01
and 2.5∘ and are focused on hourly to monthly timescales (Fisher et
al., 2008; Miralles et al., 2011, 2016; Mu et al., 2007; Vinukollu et al., 2011;
Zhang et al., 2010). The multidecadal trends in global
terrestrial latent heat flux have also been investigated and analyzed based
on these newly generated products (Jung et al., 2011; Mao et al., 2015;
Miralles et al., 2014; Zhang et al., 2015; Zhang et al., 2016). For field and
continental scale agricultural applications, ALEXI/DisALEXI (Anderson et al.,
2007; Norman et al., 2003) already have the ability to provide very high
spatial resolution surface fluxes (up to 10 m resolution) with the use of
thermal observations from a combination of polar and geostationary orbiting
satellites (Anderson et al., 2011).
The Global Energy and Water Cycle Experiment (GEWEX) LandFlux initiative aims
for the analysis of existing global land surface flux products and the
generation of new data sets of land surface fluxes (McCabe et al., 2016). A
comparison of existing global latent heat flux data sets from either land
surface models, reanalysis or satellite estimates was conducted within the
GEWEX LandFlux-EVAL initiative (Mueller et al., 2011; Jiménez et al.,
2011) and a synergy data set has been compiled, which provides latent heat
fluxes at monthly timescale and a spatial resolution of 1∘ (Mueller
et al., 2013).
However, large discrepancies remain in the existing data products. The global
mean latent heat flux over land was diagnosed as 45 ± 5 W m-2
with a spread as large as 20 W m-2 and substantial regional and
seasonal differences (Jiménez et al., 2011).
These discrepancies might be either related to the different methods applied
to estimate the surface fluxes as well as due to different ancillary data sets
used (Ershadi et al., 2014; Vinukollu et al., 2011). Recently, McCabe et
al. (2016) examined the performance of four commonly used methods for the
estimation of surface evaporation with FLUXNET tower-based and
globally-gridded forcing data. They found that the root mean square difference (RMSD) ranges from 61
to 101 W m-2 for 3-hourly data from 45 FLUXNET towers.
As a parallel and complementary effort to the GEWEX LandFlux initiative, the
ESA WACMOS-ET project aimed to identify the appropriate methods for the
estimation of latent heat flux and maximizing the use of European Earth
Observation data sets. The accuracy of the WACMOS-ET results have been
validated against a set of FLUXNET sites. Compared to McCabe et al. (2016), a
set of different FLUXNET sites and forcing data sets are investigated by Michel
et al. (2016). They found accuracies between 40.8 and 88.5 W m-2 for
3-hourly values comparing against data from 24 eddy-covariance towers
(Miralles et al., 2016; Michel et al., 2016). Another important finding from
both recent projects is that no single algorithm always outperforms any other
method. In addition, the existing models do not capture well the early-morning and late-afternoon transitions in the atmospheric boundary layer
(Ershadi et al., 2014). In order to develop a more accurate global latent heat flux product, improvement of the parameterization and sensitivity analysis of
the model to forcing data set are still needed (McCabe et al., 2016; Michel et
al., 2016).
Only a limited numbers of studies provide evaluation of latent as well as
sensible heat fluxes. Previous studies estimated sensible heat flux at
the regional scale and validated against limited in situ measurements with
accuracies ranging from ∼ 10 to ∼ 100 W m-2 (Jia et al.,
2003; Marx et al., 2008; Tang et al., 2011b; Wang et al., 2013; Zhuang et
al., 2016).
The present paper introduces a novel framework for the generation of global
high-resolution land surface fluxes from satellite and reanalysis data sets.
The High resOlution Land Atmosphere surface Parameters from Space (HOLAPS)
framework makes use of meteorological drivers coming from globally available
satellite and reanalysis data sets and integrates many of the different
components developed in previous studies within a single framework. A
state-of-the-art land surface scheme is used for the estimation of the
surface energy and water fluxes. HOLAPS allows for internally consistent
estimates of the surface radiation and water fluxes at high temporal
(< 1 h) and spatial (< 5 km) resolutions. In particular, the
shortwave and longwave surface radiation fluxes are consistently estimated,
which is often not the case when satellite-based forcing data from different
sources are used, as these can differ, e.g., in their cloud coverage or
characterization of the atmospheric humidity profile. The different
components of the HOLAPS framework are easily exchangeable as they are
coupled through well-defined interfaces. This allows, for instance, for the
integration of different approaches for the estimation of surface turbulent
fluxes while building on the general HOLAPS infrastructure for providing all
required forcing data. The required drivers for HOLAPS comprise satellite
data at different processing levels and reanalysis data for a limited
number of variables. The modular framework allows for integrating different land
surface schemes.
HOLAPS drivers, estimated surface fluxes and surface fluxes and modules.
The objectives of the present study are mainly 2-fold. First, we introduce
and validate the surface fluxes from the novel HOLAPS framework at
quasi-global scales (50∘ S, 50∘ N). Second, we
perform a thorough sensitivity analysis of the impact of different forcing
data sets on the accuracy of surface heat flux estimates. The latter is
motivated by the question: how much uncertainty is introduced when using
globally available satellite and reanalysis data as a driver for land surface
models compared to local measurements. The HOLAPS results are validated using
tower-based eddy-covariance measurements for a wide range of ecosystems and
climates.
We first briefly introduce the overall HOLAPS concept and framework developed
in Sect. 2. The data sets and methods are introduced in Sects. 3 and 4,
respectively, followed by the summary and conclusions.
Model
The HOLAPS v1.0
framework is used for the estimation of quasi-global surface water and
energy fluxes. It is based on a state of the art land surface model and was
in particular designed to make use of satellite and reanalysis data as
drivers as well as to maximize internal consistency of the different energy
and water fluxes. HOLAPS is used for the estimation of quasi-global surface
fluxes at high spatial and temporal resolutions. It is based on a radiation
module, a planetary boundary layer model, a soil module and a general module
for the exchange of energy and moisture at the surface layer. All framework
components are modular and are easily exchangeable.
Forcing data and variable interdependencies in the HOLAPS model. Only
major output variables are illustrated. Details for model formulations can
be found in Appendix B.
Figure 1 shows the general surface state and fluxes simulated by HOLAPS and
Fig. 2 shows the general interdependency between the different variables as
described briefly in the following. A very detailed technical documentation
of the entire model formulation is provided in Appendix B.
The all sky surface solar irradiance Rg (W m-2) is either obtained
from remote sensing products or is directly calculated internally by the
HOLAPS radiation module using the MAGIC radiative transfer model (Mueller et
al., 2009). The algorithm requires information on aerosol properties and surface
albedo (α) as well as total column water vapor content (TCW)
(kg m-2). Aerosol properties are taken from an aerosol climatology
(Kinne et al., 2013). Total column water vapor content can be either used
from climatologies or reanalysis data. Details on the accuracy of the MAGIC
radiative transfer model is provided by Posselt et al. (2012). When radiation
data are used as input, the radiation module calculates in addition the cloud
coverage, which is further required for the calculation of consistent longwave
radiation fluxes.
The land surface scheme is explicitly coupled to a one-dimensional (1-D) mixed layer model for
the planetary boundary layer (PBL), which is used to calculate the surface
downwelling radiation consistently with the surface heat fluxes. As the PBL
temperature and height are directly linked to the surface turbulent fluxes, a
combination of the surface heat fluxes with a PBL model helps to better
constrain the surface heat flux estimates as has been shown in previous
studies, e.g., the ALEXI model (Margulis and Entekhabi, 2001; Anderson et
al., 2007). However, while it has been shown that such an approach helps to
better constrain the surface heat fluxes, it is rarely used in common methods
for the estimation of surface heat fluxes. A mixed boundary layer model is
used within HOLAPS (Kim and Entekhabi, 1998; Margulis and Entekhabi, 2001;
Smeda, 1979), which calculates the boundary layer height and temperature using prognostic equations (see Sect. B2.6 in Appendix B), whereas
the boundary layer temperature can be nudged towards available air
temperature observations. The soil temperature is calculated using a
force-restore approach (Ren and Xue, 2004), which gives the surface
temperature (TS) that is required for the calculation of the
longwave surface net radiation budget.
The surface water fluxes comprise vegetation interception and soil moisture
dynamics as well as evaporation and transpiration processes. The currently
implemented land surface scheme calculates the latent heat flux following
the Priestley–Taylor formulation. The surface aerodynamic and canopy
resistances are estimated as a function of wind speed, air temperature, soil
moisture and surface solar radiation flux. Calculated sensible heat flux
feeds directly back into the PBL model, which constrains the diurnal
evolution of the surface fluxes as discussed earlier.
The present paper will focus exclusively on the validation of HOLAPS v1.0
results using in situ flux-tower measurements as well as the assessment of
the sensitivity of HOLAPS to forcing perturbations, namely, different forcing
data sets. An assessment of spatiotemporal dynamics estimated from HOLAPS and
cross-comparison against other existing global data sets, e.g., the
LandFlux-EVAL data set (Mueller et al., 2013), will be performed in a separate
study. All symbols used throughout the manuscript are summarized in
Appendix A.
Data
The HOLAPS framework was in particular designed to (a) make use of globally
available satellite and reanalysis data and (b) ensure internally consistent
flux estimates. The drivers required to force HOLAPS are summarized in
Table 1. These consist of satellite remote sensing and reanalysis data sets,
which have been thoroughly validated and which are briefly introduced in the
following. The data sets have in common: they (a) provide long-term
observations of the required driver variables and (b) provide this
information at comparably high temporal and spatial resolutions, which is a
major prerequisite. Data sets that are based on geostationary satellite
measurements are therefore given preference. Static information on land cover
and soil properties is required as well. All data need to be regridded to
the computational grid, and temporal interpolation to the HOLAPS timescale is
required. Details about the employed interpolation techniques are provided in
Appendix B8.
FLUXNET data
Measurements of surface turbulent fluxes are obtained from eddy-covariance
towers of the FLUXNET network. These measure the exchange of carbon dioxide,
water vapor and energy between terrestrial ecosystems and the atmosphere
(Baldocchi, 2003). Standard meteorological measurements are collected as at
most stations. The most comprehensive compilation of these flux-tower
measurements is available from the “La Thuile 2007” database (Papale et
al., 2012).
A subset of FLUXNET stations was used for the analysis in the present study.
Stations were selected where (a) all variables required to run the HOLAPS
model (Table 1) were available, (b) the station provided data with limited
data gaps (> 80 % coverage). All data are carefully quality
checked and available quality flags are applied to ensure the highest quality of
the reference data.
Overview of data sets used as drivers for HOLAPS.
VariableData setSpatialresolutionTemporal resolutionSpatialcoverageTemporal coverageReferencePrecipitationTMPA v70.25∘3 h±50∘Jan 1998–presentHuffman et al. (2007)Surface solar radiation fluxMETEOSAT SARAH SIS2.5 kmhourlyMeteosatJan 1983–Dec 2013Müller et al. (2015)TOA reflectanceGRIDSAT8 km3 hGlobalJan 1980–presentKnapp et al. (2011)TemperatureERA-InterimT255 (∼ 80 km)6 hGlobalJan 1979–presentDee et al. (2011)Wind speedERA-InterimT255 (∼ 80 km)6 hGlobalJan 1979–presentDee et al. (2011)Total column water vaporERA-InterimT255 (∼ 80 km)6 hGlobalJan 1979–presentDee et al. (2011)PressureERA-InterimT255 (∼ 80 km)6 hGlobalJan 1979–presentDee et al. (2011)Soil textureHWSDn/aStaticGlobal–FAO (2012)Surface albedoGlobalbedo1 km8 daysGlobalJan 1998–Dec 2011Muller et al. (2012)Leaf area indexMODIS Beijing Normal University1 km8 daysGlobalJan 2000–Dec 2015Yuan et al. (2011)
The stations used in the present study are depicted in Fig. 3. A major number
of stations are located in Europe and North America, and only a few stations
are located in other regions. Table C1 lists all stations (N=48) that
fulfilled the above-described criteria and provides detailed information
about data availability and relevant references for each station. The total
number of measurement years, which is used for the present analysis, is M=101
years. FLUXNET data are currently distributed under different data policies.
For the present study we only use data from stations that provide their
data under a “free fair use” license (http://www.fluxdata.org).
Distribution of FLUXNET stations used in this study. Light green
corresponds to latitudes between 50∘ N and 50∘ S, which corresponds to the
coverage of the TMPA precipitation data (see text). Stations in red cannot
be used when forced with TMPA data. Light orange indicates approximate
coverage of Meteosat data.
Eddy-covariance measurements are subject to uncertainties from various
sources. A common problem is that the eddy-covariance measurements typically
do not allow one to close the surface energy balance (RN-G-H-LE=0) (see
Table A1 for definition of acronyms throughout the paper). The energy
imbalance for eddy-covariance measurements can be as high as 20 to 30 %
on average (e.g., Wilson et al., 2002). The reason for this energy balance
closure problem is still not fully understood and subject of ongoing research
(e.g., Ingwersen et al., 2015). Several approaches have been developed to
empirically correct for the energy closure (Foken et al., 2011; Twine et al.,
2000; Wilson et al., 2002; Ingwersen et al., 2015). A simple energy balance
correction (Bowen ratio method) is applied in this study following the
approach as described in Twine et al. (2000). Further uncertainties in the
FLUXNET data occur under stable conditions, as the eddy-covariance method
requires turbulent conditions (Berbigier et al., 2001). It should be noted
that the eddy-covariance measurements are less accurate under rainfall
conditions. Previous studies have therefore removed measurements during rain
events (Ershadi et al., 2014; Michel et al., 2016). As we applied the quality
flags available from the FLUXNET data, many rainfall events were masked
already. A sensitivity study was performed to evaluate if additional masking
of rainfall events affects the results of the present study, but no
deterioration of the HOLAPS performance during rainfall events could be
identified. Therefore, we did not explicitly exclude any rainfall data from
the analysis.
Large-scale forcing data
In the following we will briefly summarize the different forcing data sets
used within the HOLAPS framework.
Radiation data
The surface solar radiation flux (Rg) is either prescribed from existing
satellite data products or can be calculated internally within the HOLAPS
framework (cf. Appendix B2.2). In both cases a maximum consistency between
the shortwave and longwave radiation fluxes is ensured as the same ancillary
data (TCW, cloud fractional coverage) are used. This explicit internal
consistency of the radiation flux estimates is unique to the HOLAPS
framework.
As the surface solar radiation is a major input to the surface energy
balance, it is expected that uncertainties in radiation data will also
affect the accuracy of the derived water and energy fluxes. Different
approaches to estimate Rg are therefore analyzed in the present study.
The following radiation data sets are used:
FLUXNET: The radiation data measured at each FLUXNET station are
used as a reference as these local measurements are expected to provide the
most accurate surface solar radiation estimates for the FLUXNET locations.
They also capture local changes in Rg at high temporal frequencies
(e.g. cloud shadowing) and might also be affected by local effects like
topographic conditions.
CM SAF-SIS: The EUMETSAT Climate Monitoring Satellite Application
Facility (CM-SAF) has specialized in the generation of long-term climate data
records from satellites. As part of their suite of radiation data products
(www.cmsaf.eu), the CM SAF provides solar incoming surface (SIS) radiation
data at hourly timescales and with a spatial resolution of
0.03∘ (Posselt et al., 2012; Müller et al., 2015) for all sky conditions. The CM SAF-SIS
is based on data from the series of METEOSAT satellites. It therefore
provides only a limited area coverage (see Fig. 3).
GRIDSAT: The Gridded Satellite data set (GRIDSAT) (Knapp et al., 2011) provides a long-term (January 1980 to
present) record of top-of-atmosphere (TOA) radiances in the visible and
thermal spectral domains. It is based on the International Satellite Cloud
Climate Project (ISCCP) (Rossow and Schiffer, 1991; Knapp, 2008) and provides
data every 3 h on an equal angular grid with a resolution
of 0.07∘.
The TOA radiances in the visible channels are used to estimate a cloud
effective albedo (CAL) (Posselt et al., 2012), which is then used subsequently
for the calculation of Rg and cloud cover fraction (cf. Sect. B2.2).
Precipitation data
Satellite precipitation data sets are produced from satellite only or combined
satellite and ground-based measurements at a variety of spatial (0.25 to
2∘) and temporal (3-hourly to monthly) resolutions at the global
scale. Ground-based precipitation estimates, e.g., from ground-based rain
radars provide even higher temporal and spatial resolution, but are available
only for limited areas. A comprehensive review and inter-comparison of
existing satellite-based precipitation products and their application is
provided by Kidd et al. (2012) and Kucera et al. (2013)
The TRMM Multisatellite Precipitation Analysis (TMPA) product (3B42 v7) is
used for the present study (Huffman et al., 2007). It combines microwave
sounding and infrared observations and compensates product biases using rain
gauge information on monthly timescales. TMPA provides 3-hourly precipitation
information at a spatial resolution of 0.25∘. It has been available since
1998 and covers the geographical extent of 50∘ N,
50∘ S.
The high temporal frequency of the measurements is a major advantage for flux
estimates and the main reason why TMPA is currently used within HOLAPS. The
spatial extent of TMPA, however, currently limits the application of HOLAPS to
that same extent (±50∘ latitude).
Vegetation data
Leaf area index (LAI) data products from the Moderate Resolution
Spectroradiometer (MODIS) instruments (Justice et al., 2002) are used in the
present study. We use an enhanced product from Beijing Normal
University
http://globalchange.bnu.edu.cn/research/lai
(Yuan
et al., 2011), which provides enhanced temporal and spatial consistency of the
MODIS LAI fields by post-processing the original MOD15A2 products (Myneni et
al., 2002). This results in much more consistent LAI fields than in the
original product, which contains abrupt changes in the time series. Surface
albedo information is obtained from the ESA GlobAlbedo project (Muller et
al., 2012; Potts et al., 2013). Both, LAI data and surface albedo are
available every 8 days. As both variables are varying slowly in time, they
are linearly interpolated to the model time step. However, it needs to be
emphasized that the used LAI and albedo products are not necessarily
consistent between each other, as they are derived from different instruments
and using different inversion techniques. Such a consistency of land surface
parameters could only be achieved through joint surface parameter retrieval
approaches, e.g., that provided by Pinty et al. (2011), and is part of ongoing
research activities, e.g., within the QA4ECV project
(http://www.qa4ecv.eu/).
Reanalysis data
A number of additional fields (temperature, wind speed, total column water
vapor path, pressure) are required from global reanalysis as these variables
are not available from remote sensing data at the required temporal and
spatial scales. Therefore, ERA-Interim reanalysis (Dee et al., 2011) fields are
used for that purpose, which provide 6-hourly data on a regular global grid
with 512×256 grid points, which corresponds to a spatial sampling of
∼ 0.7∘. The reanalysis fields are remapped to the flux-tower
locations using bilinear interpolation. The scale mismatch between the used
reanalysis field data and the local scale HOLAPS simulations might result in
additional uncertainty in the simulations and is investigated in the present
study.
List of performed model experiments. Includes the number of
stations and station years as well as the data source: F is FLUXNET data; S is satellite data for precipitation and radiation; additional data from
satellites for albedo and LAI, and from ECMWF reanalyses for temperature,
total column water vapor, and wind speed.
CoverageExperimentNumber of Precipitation Radiation Temperature Wind speed stationsyearsFSFSFSFSGlobalCTRL_G48101xxxxGRIDSAT_G48101xxxxMetosat diskCTRL_M1937xxxxMETEOSAT_M1937xxxxGRIDSAT_M1937xxxx±50∘CTRL_503061xxxxGRIDSAT_503061xxxxTmpa_503061xxxxTa_503061xxxxWind_503061xxxxMetosat disk & ±50∘CTRL_M_501017xxxxMETEOSAT_M_501017xxxxGRIDSAT_M_501017xxxxLand cover data
Global land cover information is available with a spatial resolution of 300 m
from the ESA Climate Change Initiative land cover project (Bontemps et al.,
2012; Defourny et al., 2014). The land cover information is used for the
spatial discretization of land-cover-dependent parameters in HOLAPS, e.g.,
roughness length or surface resistance parameters. These are summarized in
Table B1.
However, for the present study, no global land cover data set is used as the
experiments conducted are only performed on the point scale. The land cover
type is known for each FLUXNET station and is therefore used in the present
study.
Soil data
Information on soil properties is obtained from the Harmonized World Soil
Database (HWSD) (FAO, 2012). The HWSD is based on soil mapping units with
varying sizes. Thus, no fixed resolution can be given, but the map is gridded
with a spatial spacing of 30 arcsec. The information on soil texture (sand,
clay content) is used to derive soil hydrological properties using
pedo-transfer functions (Cosby et al., 1984; Rawls and Brakensiek, 1985; Lee,
2005).
As the HWSD is a global data set, the local soil properties might differ from
those of the used mapping units. Further uncertainties are introduced by
the applied pedo-transfer functions to derived soil hydraulic parameters from
soil texture information (e.g., Wösten et al., 2001).
MethodsExperimental setup
To quantify the accuracy of HOLAPS and the uncertainties related to the usage
of different satellite and reanalysis data sets as drivers, we conduct a series
of sensitivity experiments. Using the different data sets introduced in
Sect. 3.2, we aim to investigate the uncertainty introduced by replacing a
locally measured forcing with satellite-based drivers. First a control
simulation (CTRL) is conducted, which is based exclusively on local
measurements from FLUXNET only. This allows one to quantify HOLAPS accuracies
without additional uncertainties from the satellite and reanalysis data sets.
Thus, the CTRL simulation is considered as the baseline accuracy of the
current HOLAPS land surface scheme. For each site multiple years are used for
the simulations (see Table 2). Results are then compared against reference
measurements from FLUXNET and the accuracy of the simulations is quantified
using various skill scores (cf. Sect. 4.2).
Further experiments are conducted by replacing individual drivers (e.g.,
radiation, precipitation) with data from either satellite observations or
reanalysis. This allows one to quantify the additional uncertainty introduced by
the usage of these particular data products. The different experiment names
allow one to identify the variable that was replaced by satellite/reanalysis
data (e.g., experiment Ta= air temperature was replaced).
However, as the different data sets cover different spatial domains
(cf. Fig. 3) we generated subsets of stations representing the following
different spatial domains:
Global (G): global coverage uses the maximum number of FLUXNET stations
available.
±50∘ (50): as the precipitation data currently used
are available only between 50∘ S and 50∘ N, we use this
spatial domain to analyze the sensitivity to changes in the precipitation
forcing.
Meteosat disc (M): the analysis of the impact of satellite surface
radiation data sets on HOLAPS results is investigated for the Meteosat spatial
domain, as long-term radiation data sets are only available from the CM SAF
for Meteosat so far.
A few FLUXNET stations are located within the Meteosat disc, but within
latitudes of 50∘ S, 50∘ N. For these stations we
conducted additional simulations (M_50).
Control simulations are conducted for all of these different spatial domains.
As a consequence a total of four different control simulations with a different
number of stations are conducted. All the other experiments were also
performed for these different spatial subsets where applicable. The
differences between the same experiment type, at different spatial domains
provides additional information on the variability of the error metrics as a
function of the number of FLUXNET stations used. Table 2 summarizes all
experiments conducted and the number of stations and simulation years.
While this experimental setup allows one to quantify the impact of different
drivers on the HOLAPS results, it does not allow one to explicitly disentangle
different components of the overall mismatch between reference data and model
results, which are affected by, e.g., model parameterization uncertainties,
uncertainties in ancillary data (e.g., soil information) and spatial
representativeness of the used reference and forcing data as well as
uncertainties in the reference data itself. This could be achieved, e.g., by
perturbing the model input parameters and usage of different ancillary
data sets. Nevertheless, for the present study we keep the HOLAPS model setup
fixed as described in Appendix B.
Analysis
We compare the net radiation and HOLAPS turbulent heat fluxes with the
corresponding reference data from FLUXNET at hourly, daily and monthly
timescales using standard statistical skill scores. The variance of the
difference between the model simulations and FLUXNET data is a function of
(a) the uncertainties of the HOLAPS model itself, (b) the sensitivity of the
HOLAPS model to uncertainties in the forcing data (including
representativeness error) and (c) uncertainties in the FLUXNET
reference data. Uncertainties in the FLUXNET measurements might also result
from varying temporal and spatial footprints of the flux-tower measurements
(Chen et al., 2011).
Statistical metrics
The mean squared difference between in situ observations (x) and model
results (y) is given as
MSD=RMSD2=1N∑iNxi-yi2.
The RMSD is defined as the square root of
Eq. (1). For the calculation of the centered root mean square difference
(cRMSD), the bias is removed in advance. It is then defined as
cRMSD=1N∑i=1Nxi-x¯-yi-y¯2,
whereas the overbar indicates temporal averaging. This is also related to the
Pearson correlation coefficient (r) (Taylor, 2001).
Comparison of surface net radiation flux (RN) between FLUXNET
measurements and HOLAPS estimates for the CTRL experiment: (a) hourly and
(b) daily timescales. Colors indicate the frequency of occurrence of values
(data density). Units in W m-2.
The above-defined metrics (r, cRMSD, RMSD) are calculated for each FLUXNET
station over the entire analysis period. We then normalize each metric by the
corresponding metric obtained from the control experiment to obtain relative
deviations of the error skill scores of an experiment and the same score from
the CTRL simulation for the same station.
Temporal aggregation and data gaps
The comparison between FLUXNET and HOLAPS is performed on hourly, daily and
monthly timescales and the above metrics are calculated for these different
aggregation periods.
As the FLUXNET measurements also contain data gaps these might introduce
sampling biases. A traceable approach is therefore required to derive the
temporally aggregated reference. A daily mean is therefore only calculated if
at least 16 h (i.e., two-thirds) of valid data were available from the FLUXNET
measurements on that particular day. Given half-hourly data, this requires
that at least 32 valid data samples are available from the eddy-covariance
data set. Once daily mean fluxes have been calculated these are used to
estimate monthly mean statistics. A monthly mean is calculated if at least
two-thirds of the days of a month contained valid values. This approach was chosen
as the data gaps might introduce biases for daily and monthly values and it
was found that the calculated error statistics could be largely influenced by
a few dates with insufficient reference data. Therefore, the chosen approach
provides a traceable procedure to provide reference data for different
temporal resolutions.
Results
The HOLAPS validation results are summarized in the following. We hereby
focus on the accuracy of the surface energy and water fluxes estimated by
HOLAPS and evaluate the surface net radiation (RN), solar radiation
(Rg) and the surface latent (LE) and sensible heat (H) fluxes
for all experiments.
Evaluation of surface net radiation (RN)
The estimated surface net radiation from all 48 stations is compared against
the corresponding measurements from FLUXNET in Fig. 4 for the CTRL experiment
and all FLUXNET stations. Overall, HOLAPS provides very accurate estimates of
RN at hourly and daily timescales. The correlation between
reference data and HOLAPS is r=0.96 (0.91) for hourly (daily) data. All
correlations are significant (p<0.05). The corresponding RMSD is 54.5
(27.2) W m-2 for hourly (daily) data with almost no bias.
Box plots of validation statistics that are calculated at each
station for surface net radiation (RN) for hourly data and all
experiments investigated: (a) RMSD, (b) cRMSD, (c) correlation coefficient.
The box corresponds to the inner-quartile range of the data and the red line
indicates the median value. Numbers indicate number of model years for each
experiment.
However, as these statistics are based on the entire data record from all
FLUXNET stations, the accuracy of HOLAPS net radiation is also validated for
each of the stations individually. Statistics for the RMSD, cRMSD and
correlation that are calculated at each station are summarized in Fig. 5 for
all experiments introduced in Sect. 4.1 for hourly timescales. The
corresponding error statistics for daily and monthly fluxes are provided in
the Appendix D.
Box plots of (a) RMSD and (b) cRMSD for hourly surface solar
radiation flux (Rg).
Comparison of HOLAPS latent heat flux for (a) hourly and (b) daily
timescale for the CTRL experiment using results from all stations and years.
Units in W m-2.
Comparable accuracies are obtained for all CTRL simulations, which are based
on a different number of stations (varying spatial coverage). Using satellite
and reanalysis data as drivers for temperature, precipitation or wind speed,
the net radiation accuracies show only minor changes. Larger sensitivity of
HOLAPS is observed when replacing the local surface solar radiation with
satellite-based surface radiation data (METEOSAT, GRIDSAT experiments). The
RMSD for surface net radiation ranges between 62 and 103 W m-2 for
the majority of the stations compared to 30 to 59 W m-2 for the
other experiments, which corresponds to a significant increase in
uncertainty.
While the correlation coefficients for the different CTRL simulations are
very high (r>0.95), the correlation coefficients for the experiments
using METEOSAT or GRIDSAT radiation are lower, still amounting to r>0.8 for most cases. Only minor differences can be observed
between the RMSD and cRMSD, which indicates that the hourly estimates of
RN have only a small bias.
Box plots of (a) RMSD, (b) cRMSD and (c) correlation coefficient
for HOLAPS hourly latent heat flux.
The accuracy of the daily and monthly net surface radiation shows a picture similar
to the hourly values (see Figs. D1 and D2). The RMSD for the daily
fluxes ranges between 18 and 52 W m-2 for the majority of the
results and correlations are typically larger than r=0.95. In the cases
where satellite data are used as a radiation driver, the RMSD also increases and
the correlation coefficient reduces. However, for monthly mean fluxes
(Fig. D2) the discrepancy between CTRL simulations and the METEOSAT and
GRIDSAT experiments reduces.
Evaluation of surface solar radiation flux
(Rg)
As shown before, major uncertainties in the surface net radiation flux are
introduced by using satellite radiation products within HOLAPS. The accuracy
of the radiation data itself are therefore investigated at
the FLUXNET stations in the following. Figure 6 shows the RMSD and cRMSD for hourly surface
global radiation fluxes. For the CTRL simulations, the deviations are close
to zero as these experiments are based on the same radiation data as that
used as reference. Minor deviations still occur in these cases as the FLUXNET
measurements are not available at exactly the same time steps as HOLAPS
simulations. As HOLAPS interpolates the driver data to equal time steps,
small interpolation differences might occur, which result in non-zero RMSD
values.
The RMSD of the satellite radiation data (METEOSAT, GRIDSAT) ranges between
75 and 143 W m-2 at hourly timescales. This is partly related to a
negative bias between the FLUXNET radiation data and the satellite radiation
data. Thus, the deviations in the radiation data have by far the strongest
effect on the surface net radiation flux and are also likely to affect the
surface turbulent heat flux estimates, which will be analyzed subsequently.
Evaluation of latent (LE) and sensible (H) heat fluxes
The overall relationship between HOLAPS latent heat flux estimates and
FLUXNET measurements is illustrated in Fig. 7. The RMSD is 51.2, 30.7 and
26.3 W m-2 for the hourly, daily and monthly flux estimates, respectively, for the
CTRL_G simulations. The correlation coefficient is 0.87 for hourly data,
0.79 for daily and 0.81 for monthly data.
Comparison of HOLAPS sensible heat flux for (a) hourly and
(b) daily timescale for the CTRL experiment using results from all stations and
years. Units in W m-2.
Error statistics for all experiments are provided in Fig. 8. The increased
uncertainty in the surface solar radiation and thus RN has a direct
effect on the accuracy of the latent heat flux estimates. Correlation
coefficients are the smallest for the experiments that use satellite surface
solar radiation data. However, the correlations are still high with r>0.74 for most of the stations and experiments. The RMSD for the
CTRL simulations ranges between 35 and 52 W m-2 for the majority of
the cases. The largest RMSD is observed for the METEOSAT and GRIDSAT experiments.
However, results from the experiments when replacing the air temperature and
wind speed with reanalysis data show that this also introduces uncertainties
in the latent heat flux estimates. The RMSD ranges between 40 and
62 W m-2 for these experiments. Corresponding results for daily and
monthly timescales are provided in Figs. D3 and D4.
The overall error statistics for the sensible heat flux in the CTRL_G
simulations are shown in Fig. 9. The RMSD ranges from 79.1 W m-2
(hourly) to 36.0 W m-2 (daily). The error statistics for all
experiments are shown in Fig. 10 and show a result similar to the latent
heat flux error statistics with worse statistics for the experiments with
satellite radiation data as a forcing. The daily and monthly comparison
results are shown in Figs. D5 and D6.
Box plots of (a) RMSD, (b) cRMSD and (c) correlation coefficient
for HOLAPS sensible heat flux.
In principle, the accuracy of the results obtained might depend on
additional factors, e.g., the land cover type, the cloudiness of the sky
or the local time. Additional analysis of the HOLAPS results were therefore
performed to analyze in more detail the impact of these additional factors.
Overall HOLAPS accuracies for RN, LE and Rg, at
hourly (h), daily (d) and monthly (m) timescales for the CTRL, GRIDSAT and
METEOSAT experiments.
VariableExperimentRMSD W m-2cRMSD W m-2RhdmhdmhdmRNCTRL_G54.527.222.754.527.122.70.960.910.91GRIDSAT_G98.140.927.397.238.623.50.890.790.90LECTRL_G51.230.726.349.126.921.80.870.790.81GRIDSAT_G61.833.125.560.831.022.80.790.710.80RgMETEOSAT_M83.924.715.383.623.513.20.940.970.99GRIDSAT_M109.652.931.8106.546.117.00.910.870.98
In order to explore if the model performance is influenced by the biome
types, the overall HOLAPS error statistics across biomes are shown in Figs. E1
and E2. It can be seen that the performance of HOLAPS is generally stable
across biomes. Relatively high RMSD (∼ 60 W m-2) was found
over croplands, deciduous broadleaf forests and savannas.
Michel et al. (2016) investigated the accuracy of surface latent heat flux at
specific times of a day. We therefore also investigated if the HOLAPS error
statistics vary between daytime and nighttime compared to the entire day.
The day and night separation was based on a global radiation threshold of
20 W m-2 as suggested by Reichstein et al. (2005). Figures E3 and E4
show the HOLAPS latent heat flux error statistics over daytime and nighttime.
Compared to full-day statistics (r=0.87, RMSD = 51.2 W m-2),
the daytime has a slightly worse performance (r=0.81,
RMSD = 67.9 W m-2), while nighttime has the worst performance (r=0.35, RMSD = 21.1 W m-2). The small RMSD of nighttime is due to
the overall small fluxes during nighttime and the low correlation values
might be caused by both errors from model and measurements.
The influence of clouds on the performance of HOLAPS has also been explored
in the present study. According to Peng et al. (2013a), the clearness index
KT (the ratio of the global solar radiation measured at the surface to the
total solar radiation at the top of the atmosphere) was used to separate
clear-sky conditions (0.65<KT≤1) from partly cloudy skies
(0.15 < KT ≤ 0.65) and cloudy conditions (0 ≤ KT ≤ 0.15). The error statistics of hourly latent heat flux for different cloud
coverage are shown in Figs. E5–E7. It can be seen that the best model
performance occurs under clear-sky condition, and the model performance
decreases with the increase of cloudiness.
Summary of HOLAPS accuracies
So far we have summarized the overall accuracies of HOLAPS for the different
experiments. As the HOLAPS framework is designed to be used at the global
scale with a maximum of satellite and reanalysis data as drivers, we
summarize in the following the accuracy of the HOLAPS results for the
GRIDSAT_G experiment, which corresponds to the case where only satellite
and reanalysis drivers are used for HOLAPS flux estimates. Results are
compared against the accuracy of the CTRL_G experiment that exclusively uses
FLUXNET station data and the same stations. The overall
accuracies at hourly, daily and monthly timescales for these two experiments
are summarized in Table 3.
On monthly timescales, the results for the latent heat flux of the CTRL
simulations and GRIDSAT-based estimates are rather comparable. The
correlation is r=0.80 and r=0.81 and RMSDs are 25.5 and
26.3 W m-2 for the GRIDSAT_G and CTRL_G experiments,
respectively. However, at the hourly and daily timescales the RMSD can be
10–20 % larger for the GRIDSAT_G experiment than for the CTRL_G
experiment, which is likely to be a result of the uncertainties of the
surface shortwave radiation fluxes.
The accuracy of the two surface solar radiation data sets was estimated for the
stations that were located within the Meteosat footprint. The RMSD and
correlations for Rg are summarized in Table 3 as well. For the METEOSAT
experiment, the hourly (daily, monthly) RMSD for the surface solar radiation
flux is 83.9 (24.7, 15.3) W m-2, whereas it is 109.6 (52.9,
31.8) W m-2 for GRIDSAT.
Discussion
The HOLAPS framework provides estimates of surface net radiation and latent
heat flux at accuracies that are comparable to those obtained in other
studies (Ershadi et al., 2014; McCabe et al., 2016; Miralles et al., 2016).
It was found that the major source of uncertainty is the surface solar
radiation data used as a forcing. When using tower only measurements (CTRL),
the RMSD of HOLAPS latent heat flux is 51.2 (30.7) W m-2 for hourly
(daily) fluxes. Michel et al. (2016) and Miralles et al. (2016) evaluated the
performance of four different algorithms to estimate the surface latent heat
flux, within the WACMOS-ET project, using either tower-based forcings or
satellite data. As this is probably one of the most comprehensive studies
existing, we compare our results against results from that study. The RMSD
for the algorithms investigated in the study of Michel et al. (2016) ranges
between 40.8 and 88.5 W m-2 when comparing their results at 3-hourly
time step and using tower data as a driver. At daily timescales, the RMSD
obtained for the same four algorithms ranged between 22.7 W m-2 and
52.2 W m-2. Correlations were found to range between 0.76 and 0.88
(0.66 and 0.78) for 3-hourly (daily) values. Under the support of the GEWEX
LandFlux project, McCabe et al. (2016) evaluated the same methods but with a
different number of tower stations. They found that the correlations range
from 0.71 to 0.85, and RMSD range from 61 to 101 W m-2 for
tower-based 3-hourly data. Similar statistic scores (RMSD between 64 and
105 W m-2) have also been reported by Ershadi et al. (2014), who
also evaluated similar methods (SEBS, PT-JPL, PM, advection-aridity) with
tower-based half-hourly or hourly data. For HOLAPS we have provided the
accuracy measures when using all data samples (all stations + all years)
at once. These were provided in Table 3. The HOLAPS hourly (daily) RMSD is
51.2 (30.7) W m-2 with correlations of r=0.87 (r=0.79).
However, these values are not exactly comparable with the study of Miralles et
al. (2016) as (a) the HOLAPS statistic is based on hourly values instead of
3-hourly values for the WACMOS-ET project. Further, the information provided
by Michel et al. (2016) is given as the mean value from results of all
investigated stations. Thus, instead of calculating the RMSD for all data
samples, these authors calculated first the error statistics and then
provided the mean skill score. When following an approach similar to the
48 stations investigated in the present study, the mean RMSD of HOLAPS
corresponds to 46.6 (26.5) W m-2 with mean correlations of r=0.89 (0.85) for hourly (daily) timescales. Thus, following an approach
similar to the one by Michel et al. (2016) the results of the present study are very
similar to those of WACMOS-ET.
Similar differences are also obtained when using satellite data as a driver for
the latent heat flux estimates. The RMSD obtained for 3-hourly (daily)
estimates by Michel et al. (2016) ranges from 47.6 to 88.5 (24.5 to
59.0) W m-2 while HOLAPS hourly (daily) RMSD is
62.3 (29.1) W m-2 with correlations of r=0.79 (r=0.72), whereas
Michel et al. (2016) found correlations of 0.69<r<0.82 (0.59<r<0.79) for 3-hourly (daily) comparisons. Overall, HOLAPS seems
to provide improved correlations, which might be due to the enhanced temporal
resolution of HOLAPS. It needs to be emphasized, however, that results of the
present study are not fully comparable with Michel et al. (2016), due to the
different temporal sampling, and the different number of stations
investigated (N=48 in this study instead of N=24).
Overall, a small bias was observed, for both the simulations with flux-tower
and satellite forcings (see Table 3). While the CTRL and GRIDSAT experiments
differ on hourly and daily timescales, the RMSD for the monthly results is
very similar. This indicates that the uncertainties due to the large-scale
forcing are minimized at longer timescales.
Replacing station precipitation data with the TMPA large-scale satellite
forcing as well as using ERA-Interim for temperature and wind speed has a minor
effect on the accuracy of the results obtained. By far the largest
uncertainties are introduced when using satellite-based surface solar
radiation data, whereas similar accuracies are obtained using either the
METEOSAT or GRIDSAT data. The accuracy for the surface solar radiation flux
from METEOSAT was found to have an RMSD of 83.9 (24.7) W m-2 for
hourly (daily) timescales using the FLUXNET stations located within the
Meteosat footprint (N=19), which is slightly larger than the daily RMSD of
17.9 W m-2 reported by Müller et al. (2015) based on BSRN
observations. As a further improvement of the surface solar radiation flux is
expected to improve the latent heat flux estimates, a thorough investigation
of the impact of different surface solar radiation data set will be performed
in a future study. This could then also include the analysis of reanalysis-based radiation data, which was excluded from the present study, as
Posselt et al. (2012) had already shown that the METEOSAT radiation data used
in the present study has an overall better agreement with ground measurements
than the ERA-Interim reanalysis radiation data. Overall, best results were
obtained for clear-sky conditions. Decreasing performance of HOLAPS estimates
was observed for increased cloudiness, which is likely to be caused by the
increased uncertainties in the satellite-based radiation data under cloudy-sky conditions. No systematic differences between different biome types could
be identified in this study. A more comprehensive sensitivity analysis of
HOLAPS to different biome-specific model parameters might be subject of a
further study, where the vegetation parameter of each biome will be perturbed
and the relevant HOLAPS performance will be assessed.
Conclusions
This study has introduced a new framework for the estimation of high-resolution land surface water and energy fluxes, HOLAPS v1.0. The framework
was developed to make use of existing satellite data records and to allow
for the generation of temporal and spatial high-resolved and consistent
quasi-global water and energy fluxes. Key features of the HOLAPS framework
comprise
internally consistent estimation of shortwave and longwave radiation
fluxes;
capability to directly use top-of-atmosphere radiances for surface solar
flux estimations;
constrained surface fluxes using a mixed boundary layer model in
combination with the surface flux estimates;
flexible framework for the generation of high-resolution land surface
energy and water fluxes that allows one to use a multitude of different land
surface schemes within the same framework.
This study analyzed the accuracy of HOLAPS v1.0 using data from 48 eddy-covariance towers. A sensitivity analysis was performed to investigate the
tradeoff in using satellite data as drivers instead of locally measured tower-based data. The results of this study can be summarized as follows:
The accuracy of the HOLAPS surface fluxes was found to be comparable or
even better than results obtained in other studies for the surface net
radiation as well as turbulent fluxes.
The hourly (daily) RMSD for the surface net radiation flux was 54.5
(27.2) W m-2 with correlations of r=0.96 (r=0.91) when using
tower data as drivers for HOLAPS.
For the latent heat flux, the obtained RMSD was 51.2 (30.7) W m-2
with r=0.87 (r=0.79) and 79.1 (36.0) W m-2 for the sensible
heat flux at hourly (daily) timescales.
Using satellite and reanalysis data as only drivers, the RMSD and
correlations were found to be 61.8 W m-2 and r=0.79 (33.1, r=0.71) for the latent heat flux
Accuracy of turbulent flux estimates decreases with increasing cloudiness
due to higher uncertainties in the surface solar radiation flux, which is
consistent with previous studies.
The largest uncertainties resulted from the uncertainties of the surface
solar radiation flux. However, on monthly timescales, these uncertainties
were minimized, which indicates that comparable accuracies can be obtained
when using satellite-based drivers instead of local in situ data.
A first quasi-global data set generated using HOLAPS v1.0 is planned to be
released to the scientific community after a thorough validation and cross-comparison
against other data sets, e.g., the LandFlux-Eval (Mueller et
al., 2013) data. Further improvements of the HOLAPS framework will comprise
the capability to assimilate land surface temperature data from geostationary
satellite observations to better constrain the surface latent heat flux
estimates as well as the usage of new satellite observations, e.g.,
provided by the new SENTINEL series of satellites. Recent advances in
available computational resources allow for the first time to exploit these
high spatial resolution sensors at a global scale and might lead to
operational services provided, e.g., in the frame of Copernicus services.
A major constraint is nevertheless the lack of consistent and harmonized
geostationary satellite data records. The mosaic of geostationary
satellites, known as GEORING, is currently operated by individual space
agencies and so far no long-term climate or operational data set of harmonized
and well-intercalibrated geostationary radiance and brightness temperature
data is available at the original sensor resolution. The GRIDSAT data set,
used in the present study is currently the only long-term GEORING data set
available, but is limited in its spatial resolution. Further developments
towards Fundamental Climate Data records from geostationary satellite data
are therefore required.
Further studies using HOLAPS will investigate the potential to use
the novel SENTINEL data streams and to further reduce the dependency on
reanalysis data by using, e.g., the total column water vapor information from
satellite data and exploit the potential of internally consistent land
surface parameters currently developed, e.g., by different European
projects (QA4ECV, MULTIPLY).
While the present study provides a sensitivity analysis of using the HOLAPS
framework with different forcing data, it would be important to conduct
further in-depth studies to disentangle the different components of the
overall error budget (model uncertainties, forcing uncertainties, scale
mismatches, reference data uncertainty), which still remains a major
challenge to be addressed by the research community.
Code availability
The HOLAPS code used for this manuscript can be accessed via
https://github.com/pygeo/holaps.
Acronyms
Acronyms used throughout the text are summarized in the following table.
Acronyms used throughout the manuscript.
SymbolVariableUnitGeneral variables cpHeat capacity of dry airJ kg-1 K-1ρDensity of dry airkg m-3ΔSlope of water vapor saturation curvePa K-1γPsychrometer constantPa K-1αpt=1.26Priestley–Taylor parameter–ΛLeaf area indexm2 m-2εsurface emissivity–σ=5.670373×10-8Stefan–Boltzmann constantW m-2 K-1tTimesg=9.80665Gravity accelerationm s-2TaAir temperature (2 m)KPPrecipitation ratem s-1QRunoff (fast, slow, percolation)m s-1ETEvapotranspiration fluxm s-1λLatent heat vaporizationJ kg-1Radiation module CALEffective cloud albedo [0, …, 1]–aSurface albedo–cCloud cover fraction [0, …, 1]–RN, RN,SRN,CSurface net radiation, soil/canopy net radiationW m-2RgRgclearShortwave downwelling flux, clear-sky downwelling fluxW m-2L↓Lslab↓Longwave downwelling flux, clear-sky longwave downwelling fluxW m-2kClear-sky index [0…1]–TCWTotal column water vapor contentkg m-2PBL module HvVirtual heat fluxW m-2HtopEntrainment fluxW m-2δθmMixed layer inversion strengthKθmBoundary layer potential temperatureKkvon Karman constant (≈0.41)–ζ=0.01Dissipation parameter–Turbulent flux module uWind speedm s-1LE LEI LES LECLatent heat flux, subscripts indicate: interception, soil, canopyW m-2ETEvapotranspirationm s-1ETIEvapotranspiration from canopy interception storagem s-1hVegetation heightmHSensible heat fluxW m-2GSoil heat fluxW m-2u∗Friction velocitym s-1fcVegetation cover fraction–raAerodynamic surface resistances m-1Ψm,hStability correction functionsRiRichardson number–z0,mz0,hRoughness lengths for momentum and heatmϕVegetation inhibition function–RrAerodynamic resistances m-1RCCanopy resistances m-1rradRadiation stress factorW m-2rminrmaxMinimum and maximum canopy resistances m-1γθmPotential temperature lapse rateK m-1zvegVegetation heightm
Continued.
symbolvariableunitWater flux and soil module I, ImaxCanopy interception storage, maximum interception storagemCGThermal inertial coefficientK m2 J-1ΓThermal inertiaJ m-2 K-1 s-0.5d=1.5 mSoil temperature damping scale depthmγSSoil temperature lapse rateK m-1DThroughfall and drainage of water from the canopy layer to the soilm s-1TSSurface temperatureKTdDeep soil temperatureKzVertical coordinate (e.g., boundary layer height, soil depth)mmvVolumetric soil moisturem3 m-3ΘRelative degree of saturation for soil moisture–KUnsaturated soil conductivitym s-1ΨSoil suction pressure headmWWater storage in soilmDetailed HOLAPS model description
The different components of the HOLAPS framework and its land surface model
are described in detail in the following sections. The variable definitions
used and their units are summarized in Table A1.
HOLAPS runtime environment
The general workflow of the HOLAPS runtime environment is illustrated in
Fig. B1. After specifying the model setup by the user, the HOLAPS main
controller checks the availability of all required data and then launches
subprocesses to run the model. Required forcing data are read for each time
step and interpolated in space and time if required. Surface water and
energy fluxes are calculated for each time step. Results are then written to
netCDF files and additional statistics are calculated if required.
HOLAPS sub-modules
The different sub-modules used within HOLAPS are described in the following.
Surface energy balance
The surface energy balance is given as
RN-LE-H-G=0.RN is estimated from the shortwave and longwave radiation fluxes
as
RN=1-αRg+εL↓-εσTS4.
The ground heat flux G is obtained through the coupling of the surface
energy balance model to a soil model that simulates the surface temperature
temporal evolution (see Sect. B2.3).
Radiation moduleShortwave solar surface radiation fluxes
The shortwave clear-sky solar radiation flux (Rgclear) is estimated
using the MAGIC radiative transfer model (Mueller et
al., 2009). The shortwave surface downwelling solar flux (Rg) for all
sky conditions is then obtained from the clear-sky downwelling solar flux
and the clear-sky index k as (Posselt et al., 2012)
HOLAPS runtime environment.
Rg=k(CAL)Rgclear.
The clear-sky index is related to CAL through the following relationship
(Hammer et al., 2003)
k=1.2CAL≤-0.21-CAL-0.2<CAL≤0.8a+b⋅CAL+c⋅CAL20.8<CAL≤1.10.05CAL>1.1,
where a=2.0667, b=-3.667, c=1.6667.
Longwave surface radiation fluxes
The longwave surface downwelling radiation flux (L↓) depends
on the near-surface moisture and temperature profile as well as the cloud
coverage. The clear-sky longwave downwelling radiation flux
Lslab↓ is calculated using the
PBL model (Margulis and Entekhabi, 2001).
Lslab↓ is then corrected for
cloud coverage as (Brubaker and Entekhabi, 1995)
L↓=Lslab↓(1+0.17c2).
Soil module
The surface temperature TS [K] is obtained by a revised force restore
approach (Ren and Xue, 2004) as
∂TS∂t=CGRN-LE-H-ωTS-Td-πdγS-AB′′sin[ωt+a′′],
where A (K) is the diurnal temperature amplitude of TS, CG=2Γ86 400π-1 (K m2 J-1) is
the thermal inertia coefficient and Γ is the thermal inertia, which
is estimated as function of soil moisture conditions (Murray
and Verhoef, 2007) and ω=2π86 400 is the diurnal
angular frequency. The parameters B′′ and a′′ in
Eq. (B6)
are set to a′′=0.45π and B′′=0.158 (Ren and Xue, 2004). The
prognostic equation for the deep soil layer temperature Td is
∂Td∂t=-1τTd-TS+γSπd,
where d is the soil temperature damping scale depth with typical values on
the order of d=0.15 (m). The lapse rate between the mean surface and
deep-layer temperature γS (K m-1) is estimated from the
differences between TS and Td and τ=86 400 (s) is the time
period, 1 day in our case.
Water balance module
The surface water balance is defined as
P-∂I∂t-Q-ET-∂W∂t=0.
The soil moisture dynamics is calculated using a multilayer soil scheme,
discretized into five layers. The soil layers have a thickness of
dz= [0.05, 0.1, 0.25, 0.6, 1.0] (m). Soil moisture fluxes
between the different soil layers are simulated by solving numerically the
Richards equation (Richards, 1931) whereas only vertical moisture fluxes are
considered:
∂mv∂t=∂∂zK(mv)∂ψ∂z+1.
The water fluxes between the different soil layers is solved using a
numerical approach. The net soil water flux in a soil layers is hereby
determined by the fluxes into and from the layers above any below, whereas
the model allows for both downward (percolation) and upward (capillary
rise) fluxes. Surface runoff Q is obtained as the excess of water that can
not infiltrate the soil when maximum infiltration capacity is reached. The
relationship between volumetric soil moisture content and soil suction head
ψ is calculated using the model of van Genuchten (1980).
The water interception by the canopy is estimated by (Valente et al., 1997)
∂I∂t=P-ETI-D,
where ETI=λ-1LEI is the transpiration from the
canopy interception storage and D is the through fall and drainage of water
from the canopy layer to the soil.
Turbulent flux module
For a vegetated patch with fractional vegetation coverage fc the
surface latent heat flux is calculated as the weighted sum of the evaporation
from soil (LES) and the transpiration from the canopy
(LEC) as well as evaporation from water intercepted by the canopy
layer (LEI) as
LE=1-fcLES+fc1-wILEC+wILEI,
where wI=(I/Imax)b is a weighting factor dependent on the current
canopy interception storage I, the potential maximum interception storage
Imax(Λ) (von Hoyningen-Huene, 1981) and an empirical parameter
b=0.5 (Chen and Dudhia, 2001). The vegetation cover fraction fc
is obtained from leaf area index (Λ) as (Norman et al., 1995)
fc=1-e-0.5Λ,
which assumes a random leaf distribution with spherical leaf angle
distribution. The different latent heat flux components in Eq. (B11) are then
estimated using the Priestley–Taylor approach as
LES=ϕαptRN,SΔΔ+γLEC=ϕαptRN,CΔΔ+γLEI=αptRN,CΔΔ+γ
where
αpt=1.26 is the Priestley–Taylor parameter for equilibrium
evapotranspiration and Δ, γ are the slope of the water vapor
saturation curve and psychrometer constant (Pa K-1), respectively. The
inhibition function 0≤ϕ≤1 describes the reduction
of LE due to limiting factors like radiation, temperature and soil moisture.
The soil net radiation is estimated as (Norman et al., 1995)
RN,S=RNe0.9ln(1-fc)
and the canopy net radiation is then calculated as
RN,C=RN-RN,S.
The sensible heat flux is estimated as
H=ρcp(TS-Ta)/ra,
where the aerodynamic surface resistance ra (s m-1) is
calculated as
ra=logz-dz0,m-Ψmlogz-dz0,h-Ψhk2uz-1,
where k≈0.41 is the von Karman constant and uz corresponds to
the win speed at canopy height and is obtained from wind speed data assuming
a logarithmic wind profile and a displacement height d corresponding to
two-thirds
of the vegetation height (Maidment, 1993). The stability correction functions
Ψm,h are calculated after (Paulson, 1970) using the Richardson
number Ri as an indicator for atmospheric stability. The roughness
lengths for momentum and heat (z0,m, z0,h) are
parameterized for each land cover type (Table B1).
Surface inhibition functions
The canopy inhibition function 0≤φc≤1 is defined as (Chen
and Dudhia, 2001)
ϕ=1+ΔRr-11+ChRC+ΔRr-1,
where Rr is a function of surface air temperature and pressure,
Ch is the surface exchange coefficient for heat and moisture and
RC is the canopy resistance, given as
RC=rminΛfS↓fTafmv
with
fS↓=rminrmax-1+ff1+fffTa=1-0.0016298-Ta-273.152fmv=lnw0wfw0+(wf-w0)exp(-μΘ)lnwf
with ff=1.1S↓Λrrad, where
rrad is a radiation-specific parameter W m-2 and
rmin and rmax are the minimum and maximum canopy
resistance (s m-1), which are all land cover specific parameters
(Table B1). The relative degree of soil saturation is given by Θ and
w0=1, wf=800 and μ=12 are empirical parameters (Anderson et
al., 2007). fTa and fS↓ are based on Chen and
Dudhia (2001).
Planetary boundary layer module
The prognostic equations of the PBL model are given by (Kim and Entekhabi,
1998; Smeda, 1979)
∂z∂t=2G∗-D1-δD2θmgzδθm+Hvρcpδθm,ρcpzdθdt=H-Htop-R
with
R=a(θm-θs)
with the proportionality constant a=10-5 (s-1) (Smeda, 1979).
Alternative approaches to simulate the radiative cooling have been proposed
(Kim and Entekhabi, 1998; Margulis and Entekhabi, 2001). The relationship
between PBL air temperature (T) and θ is given by
θ=TP0P-1R/cp
with R/cp≈0.286 for air. The details of the model formulations
are based on Smeda (1979) and are given as follows:
G∗=u∗2,D1=u∗2u(1-e-ζz),D2=0.4gzθmHvρcp,Hv=H+0.61θmcpET≈H+0.07LE
with δ=0 in stable conditions and δ=1 in unstable
conditions. We set ζ=0.01 to ensure a realistic collapse of the PBL
(Kim and Entekhabi, 1998).
Land cover specific parameters.
Land coverαptrminrmaxrradz0,mz0,hzvegBare soil1.264005000–0.0010.001–Cropland1.26405000300.010.0010.2Deciduous broadleaf forest0.911005000301.00.115Coniferous forest0.911505000301.40.1415Coniferous forest or deciduous0.911505000301.20.1415Deciduous broadleaf forest and broad leaf/mixed forest0.911005000301.00.115Grassland1.264050001000.010.0010.2Savanna1.2630050001000.010.0010.4Deciduous broadleaf forest and broad leaf/mixed forest0.911005000301.00.115
Summary of spatial and temporal interpolation methods used within
the HOLAPS framework for different driver variables.
VariableSpatial interpolationmethodTemporal interpolation methodPrecipitationBilinearLast_validSurface solar radiation fluxBilinearLast_valid_same_time (interpolation of clear-sky index)TemperatureBilinearLast_validWind speedBilinearLast_validTotal column water vaporBilinearLast_validPressureBilinearLast_validSoil textureNearest neighborn/aLand covern/an/aSurface albedoBilinearLast_validLeaf area indexBilinearLast_valid
During daytime, the growth of the PBL is determined by the right side in
Eq. (B30). During the transition between unstable and stable conditions, the
PBL collapses because of turbulence dissipation. The PBL height during this
transition phase is given as (Smeda, 1979)
z=-2G∗-D1ρcpθmHvg
when assuming that Htop=0. Equation (B29) is applied in this
transition phase until
dzdt-Hvρcpδθm≤0.05Hvρcpδθm.
The mixed layer is capped by an inversion with inversion strength δθm (K), which determines the entrainment of overlying dry air from
the free atmosphere as (McNaughton and Spriggs, 1986)
Htop=-ρcpδθmdzdt.
Dry air entrainment causes the inversion strength itself to change according
to
dδθmdt=γθmdzdt-dθdt,
where γθm K m-1 is the potential temperature lapse
rate above the PBL and is assumed to be constant.
Model parameterization
The land cover specific model parameters are summarized in Table B1. They are based on the publications of Chen and Dudhia (2001) and
Hagemann (2002).
Interpolation methods
Different interpolation approaches are used to interpolate the input data
onto the HOLAPS computational grid and time step. The used techniques are
summarized in Table B2 for each of the HOLAPS drivers. The nearest neighbor
remapping as well as bilinear interpolation are currently used for spatial
remapping. The temporal interpolation is based on a linear interpolation of
measurements (y1, y2) between two observation times (t1t2)y=wy2+1-wy1,
whereas the weight w depends on the sampling times and the actual model
time step.
To handle data gaps, the HOLAPS framework currently provides the following
options:
ignore: the data gap is ignored and filled by interpolation.
last_valid: last valid value of the variable is used and the
data gap is filled with this value.
last_valid_same_time: use the last valid data at the same
time of the day. This option is in particular useful for data that show a
strong diurnal dynamics (e.g., radiation). In that case, using the last valid
value would lead to erroneous diurnal forcing data when data gaps of a few
hours occur, which can be quite often the case when using, e.g., FLUXNET data.
climatology: a climatological mean annual cycle i used for the
calculations.
Interpolation methods can be easily changed by the user in configuration
files.
The special case of radiation data
No direct interpolation is performed for the radiation data, as a linear
approximation might not be sufficient to capture the diurnal cycle of the
surface solar radiation flux. Instead, the clear-sky index (k) is
interpolated in time and then used to calculate Rg using Eq. (B3).
FLUXNET stations
List of FLUXNET stations investigated. The coverage term specifies
the location of each FLUXNET station. ±50∘ refers to
the station being within the latitudes 50∘ N, 50∘ S, while Meteosat indicates the station is within the coverage of Meteosat.
For details on the spatial coverage see Fig. 3.
No.Station IDLatLongYears Coverage Reference200320042005Global±50∘Meteosat1ATNeu47.1211.32XXXXXWohlfahrt et al. (2008)2AUHow-12.49131.15XXXXXHutley et al. (2000)3AUTum-35.66148.15XXXXXLeuning et al. (2005)4BEBra51.314.52XXXXGond et al. (1999)5BEVie50.316.00XXXAubinet et al. (2001)6CAMan55.88-98.48XXDunn et al. (2007)7CAMer45.41-75.52XXXXLafleur (2003)8CANS155.88-98.48XXXGouldon et al. (2006)9CANS255.91-98.52XXXXGouldon et al. (2006)10CANS355.91-98.38XXXGouldon et al. (2006)11CANS455.91-98.38XXGouldon et al. (2006)12CANS555.86-98.49XXXXGouldon et al. (2006)13CANS655.92-98.96XXXXGouldon et al. (2006)14CANS756.64-99.95XXXGouldon et al. (2006)15CAQcu49.27-74.04XXXXX16CASF354.09-106.01XXXXMkhabela et al. (2009)17CHOe147.297.73XXXXAmmann et al. (2007)18CZBK149.5018.54XXXX19DEGri50.9513.51XXXGilmanov et al. (2007)20DEHai51.0810.45XXXXXKnohl et al. (2003)21DEMeh51.2810.66XXXXScherer-Lorenzen etal. (2007)22DETha50.9613.57XXXXX23DEWet50.4511.46XXXXXRebmann et al. (2010)24FRHes48.677.06XXXXXXGranier et al. (2000)25FRLBr44.72-0.77XXXXBerbigier et al. (2001)26FRPue43.743.60XXXXXAllard et al. (2008)27HUBug46.6919.60XXXXXXNagy et al. (2007)28ITCpz41.7112.38XXXXXGarbulsky et al. (2008)29ITRo242.3911.92XXXXTedeschi et al. (2006)30ITSRo43.7310.28XXXXChiesi et al. (2005)31NLCa151.974.93XXXXXBeljaars andBosveld (1997)32NLLoo52.175.74XXXXDolman et al. (2002)33USARM36.61-97.49XXXXXFischer et al. (2007)34USAud31.59-110.51XXXXTang et al. (2011a) and Yang et al. (2008)35USBkg44.35-96.84XXXXZhang et al. (2008)36USBo140.01-88.29XXXXXMeyers (2004)37USFPe48.31-105.10XXXXGilmanov et al. (2005) and Zhang et al. (2008)38USGoo34.25-89.87XXX39USHo145.20-68.74XXXXHollinger et al. (2004)40USHo245.21-68.75XXXXHollinger et al. (2004)41USLos46.08-89.98XXXXX42USMOz38.74-92.20XXXGu et al. (2006, 2007)43USNe141.17-96.48XXXXVerma et al. (2005)44USNe241.16-96.47XXXXVerma et al. (2005)45USNe341.18-96.44XXXXVerma et al. (2005)46USOho41.55-83.84XXXX47USTon38.43-120.97XXXXBaldocchi et al. (2004)48USWCr45.81-90.08XXXXCook et al. (2004)Ancillary HOLAPS evaluation results
Similar error statistic for RN like Fig. 5 but for daily
timescales: (a) RMSD, (b) cRMSD, (c) correlation
coefficient.
Similar error statistic for RN like Fig. 5 but for monthly
timescales: (a) RMSD, (b) cRMSD, (c) correlation
coefficient.
Similar error statistic for LE like in Fig. 8 but for daily values:
(a) RMSD, (b) cRMSD, (c) correlation coefficient.
Similar error statistic for LE like in Fig. 8 but for monthly
values: (a) RMSD, (b) cRMSD, (c) correlation
coefficient.
Similar error statistics for sensible heat flux like in Fig. 10 but for daily values: (a) RMSD, (b) cRMSD, (c) correlation coefficient.
Similar error statistic for sensible heat flux like in Fig. 10 but for monthly values: (a) RMSD, (b) cRMSD, (c) correlation coefficient.
Performance of HOLAPS over different biomes, specific times and
cloudiness conditions
Comparison of HOLAPS latent heat flux for different biomes using results from all stations and years: Dbf is deciduous broadleaf forest, Ebf is evergreen broadleaf forest,
Enf is evergreen needleleaf forest. Units in W m-2.
Error statistic for hourly latent heat flux over different biomes: (a) RMSD, (b) cRMSD, (c) correlation coefficient. Dbf is deciduous broadleaf forest,
Ebf is evergreen broadleaf forest, Enf is evergreen needleleaf forest.
Error statistics for HOLAPS latent heat flux over daytime:
(a) comparison using results from all stations and years,
(b–d) box plots of validation statistics that are
calculated at each station.
Error statistics for HOLAPS latent heat flux over nighttime:
(a) comparison using results from all stations and years,
(b–d) box plots of validation statistics that are
calculated at each station.
Error statistics for HOLAPS latent heat flux over clear-sky
condition: (a) comparison using results from all stations and years,
(b–d) box plots of validation statistics that are
calculated at each station.
Error statistics for HOLAPS latent heat flux over partly cloudy-sky
condition: (a) comparison using results from all stations and years,
(b–d) box plots of validation statistics that are
calculated at each station.
Error statistics for HOLAPS latent heat flux over cloudy-sky
condition: (a) comparison using results from all stations and years,
(b–d) box plots of validation statistics that are
calculated at each station.
Acknowledgements
This study was supported through the Cluster of Excellence CliSAP (EXC177),
University of Hamburg, funded through the German Science Foundation (DFG),
which is gratefully acknowledged. Dissemination of the FLUXNET data through
http://www.fluxdata.org/ and the work of the individual PIs for FLUXNET
stations are very much appreciated. EUMETSAT Satellite Application Facility
on Climate Monitoring (CM SAF) climate data products were used by permission
of Deutscher Wetterdienst. We thank ECMWF for the use of their ERA-Interim
reanalysis, NASA for the dissemination and use of the TMPA satellite
product, NCDC at NOAA for the dissemination and use of the GridSat radiation
satellite product, Beijing Normal University for the provision and use of
their enhanced MODIS LAI data and the MODIS land team as well as the ESA
Globalbedo project for the provision of land surface
parameters. The article processing charges for
this open-access publication were covered by the Max Planck
Society. Edited by: J. Kala
Reviewed by: D. G. Miralles, M. McCabe, and one anonymous referee
References
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,
2008.
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, 2007.Anderson, M. C., Norman, J. M., Mecikalski, J. R., Otkin, J. A., and Kustas,
W. P.: A climatological study of evapotranspiration and moisture stress
across the continental United States based on thermal remote sensing: 1.
Model formulation, J. Geophys. Res.-Atmos., 112, D10117, 10.1029/2006JD007506, 2007.Anderson, M. C., Kustas, W. P., Norman, J. M., Hain, C. R., Mecikalski, J.
R., Schultz, L., González-Dugo, M. P., Cammalleri, C., d'Urso, G.,
Pimstein, A., and Gao, F.: Mapping daily evapotranspiration at field to
continental scales using geostationary and polar orbiting satellite imagery,
Hydrol. Earth Syst. Sci., 15, 223–239, 10.5194/hess-15-223-2011, 2011.
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, 2001.Baldocchi, D.: “Breathing” of the terrestrial biosphere: lessons learned
from a global network of carbon dioxide flux measurement systems, Aust. J.
Bot., 56, 1–26, 10.1071/BT07151, 2008.
Baldocchi, D., Falge, E., Gu, L., Olson, R., Hollinger, D., Running, S.,
Anthoni, P., Bernhofer, C., Davis, K., Evans, R., Fuentes, J., Goldstein, A.,
Katul, G., Law, B., Lee, X., Malhi, Y., Meyers, T., Munger, W., Oechel, W.,
Paw, K. T., Pilegaard, K., Schmid, H. P., Valentini, R., Verma, S., Vesala,
T., Wilson, K., and Wofsy, S.: FLUXNET: A New Tool to Study the Temporal and
Spatial Variability of Ecosystem–Scale Carbon Dioxide, Water Vapor, and
Energy Flux Densities, B. Am. Meteorol. Soc., 82, 2415–2434, 2001.
Baldocchi, D. D.: Assessing the eddy covariance technique for evaluating
carbon dioxide exchange rates of ecosystems: past, present and future, Glob.
Change Biol., 9, 479–492, 2003.
Baldocchi, D. D., Xu, L., and Kiang, N.: How plant functional-type, weather,
seasonal drought, and soil physical properties alter water and energy fluxes
of an oak–grass savanna and an annual grassland, Agr. Forest Meteorol., 123,
13–39, 2004.
Bastiaanssen, W., Menenti, M., Feddes, R., and Holtslag, A.: A remote sensing
surface energy balance algorithm for land (SEBAL). 1. Formulation, J.
Hydrol., 212, 198–212, 1998.
Beljaars, A. and Bosveld, F.: Cabauw data for the validation of land surface
parameterization schemes, J. Climate, 10, 1172–1193, 1997.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, 2001.
Bontemps, S., Defourny, P., Brockmann, C., Herold, M., Kalogirou, V., and
Arino, O.: New global land cover mapping exercise in the framework of the ESA
Climate Change Initiative, Geoscience and Remote Sensing Symposium (IGARSS),
2012 IEEE International, 2012, 44–47, 2012.
Brubaker, K. L. and Entekhabi, D.: An analytic approach to modeling
land–atmosphere interaction: 1. Construct and equilibrium behavior, Water
Resour. Res., 31, 619–632, 1995.
Chen, B., Coops, N. C., Fu, D., Margolis, H. A., Amiro, B. D., Barr, A. G.,
Black, T. A., Arain, M. A., Bourque, C. P.-A., Flanagan, L. B., Lafleur, P.
M., McCaughey, J. H., and Wofsy, S. C.: Assessing eddy-covariance flux tower
location bias across the Fluxnet-Canada Research Network based on remote
sensing and footprint modelling, Agr. Forest Meteorol., 151, 87–100, 2011.Chen, F. and Dudhia, J.: Coupling an Advanced Land Surface–Hydrology Model
with the Penn State–NCAR MM5 Modeling System. Part I: Model Implementation
and Sensitivity, Mon. Weather Rev., 129, 569–585,
10.1175/1520-0493(2001)129<0569:CAALSH>2.0.CO;2,
2001.
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, 2005.
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, 2004.
Cosby, B., Hornberger, G., Clapp, R., and Ginn, T.: A statistical exploration
of the relationships of soil moisture characteristics to the physical
properties of soils, Water Resour. Res., 20, 682–690, 1984.Dee, D. P., Uppala, S. M., Simmons, A. J., Berrisford, P., Poli, P.,
Kobayashi, S., Andrae, U., Balmaseda, M. A., Balsamo, G., Bauer, P.,
Bechtold, P., Beljaars, A. C. M., van de Berg, L., Bidlot, J., Bormann, N.,
Delsol, C., Dragani, R., Fuentes, M., Geer, A. J., Haimberger, L., Healy, S.
B., Hersbach, H., Hólm, E. V., Isaksen, L., Kållberg, P., Köhler,
M., Matricardi, M., McNally, A. P., Monge-Sanz, B. M., Morcrette, J. J.,
Park, B. K., Peubey, C., de Rosnay, P., Tavolato, C., Thépaut, J. N., and
Vitart, F.: The ERA-Interim reanalysis: configuration and performance of the
data assimilation system, Q. J. Roy. Meteor. Soc., 137, 553–597,
10.1002/qj.828, 2011.
Defourny, P., Kirches, G., Brockmann, C., Boettcher, M., Peters, M.,
Bontemps, S., Lamarche, C., Schlerf, M., and Santoro, M.: Land Cover CCI:
Product User Guide Version 2, 2014.
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, 2002.
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, 2007.Ershadi, A., McCabe, M. F., Evans, J. P., Chaney, N. W., and Wood, E. F.:
Multi-site evaluation of terrestrial evaporation models using FLUXNET data,
Agr. Forest Meteorol., 187, 46–61, 10.1016/j.agrformet.2013.11.008,
2014.
FAO: Harmonized World Soil Database (version 1.2),
FAO/IIASA/ISRIC/ISS-CAS/JRC, FAO, Rome, Italy and IIASA, Laxenburg, Austria,
1–43, 2012.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, 2007.Fisher, J. B., Tu, K. P., and Baldocchi, D. D.: Global estimates of the
land–atmosphere water flux based on monthly AVHRR and ISLSCP-II data,
validated at 16 FLUXNET sites, Remote Sens. Environ., 112, 901–919,
10.1016/j.rse.2007.06.025, 2008.
Foken, T., Aubinet, M., Finnigan, J. J., Leclerc, M. Y., Mauder, M., and Paw
U, K. T.: Results Of A Panel Discussion About The Energy Balance Closure
Correction For Trace Gases, B. Am. Meteorol. Soc., 92, ES13–ES18, 2011.Garbulsky, M. F., 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.Gilmanov, T. G., Tieszen, L. L., Wylie, B. K., Flanagan, L. B., Frank, A. B.,
Haferkamp, M. R., Meyers, T. P., and Morgan, J. A.: Integration of CO2
flux and remotely-sensed data for primary production and ecosystem
respiration analyses in the Northern Great Plains: potential for quantitative
spatial extrapolation, Glob. Ecol. Biogeogr., 14, 271–292, 2005.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.,
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, 2007.
Gond, V., De Pury, D. G. G., Veroustraete, F., and Ceulemans, R.: Seasonal
variations in leaf area index, leaf chlorophyll, and water content;
scaling-up to estimate fAPAR and carbon balance in a multilayer, multispecies
temperate forest, Tree Physiol., 19, 673–679, 1999.
Gouldon, M. L., Winston, G. C., McMillan, A. M. S., Litvak, M. E., Read, E.
L., Rocha, A. V., and Rob Elliot, J.: An eddy covariance mesonet to measure
the effect of forest age on land – atmosphere exchange, Glob. Change Biol.,
12, 2146–2162, 2006.
Granier, a., Ceschia, E., Damesin, C., Dufrene, E., Epron, D., Gross, P.,
Lebaube, S., Le Dantec, V., Le Goff, N., 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, 2000.
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., 111, 1–13, 2006.
Gu, L., Meyers, T., Pallardy, S. G., Hanson, P. J., Yang, B., Heuer, M.,
Hosman, K. P., Liu, Q., Riggs, J. S., Sluss, D., and Wullschleger, S. D.:
Influences of biomass heat and biochemical energy storages on the land
surface fluxes and radiative temperature, J. Geophys. Res., 112, 1–11, 2007.
Hagemann, S.: An improved land surface parameter dataset for global and
regional climate models, Max-Planck-Institute for Meteorology, Hamburg,
Germany,
2002.Hammer, A., Heinemann, D., Hoyer, C., Kuhlemann, R., Lorenz, E., Müller,
R., and Beyer, H. G.: Solar energy assessment using remote sensing
technologies, Remote Sens. Environ., 86, 423–432,
10.1016/S0034-4257(03)00083-X, 2003.Hollinger, D. Y., Aber, J., Dail, B., Davidson, E. a., Goltz, S. M., Hughes,
H., Leclerc, M. Y., Lee, J. T., Richardson, S. 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, 2004.
Huffman, G. J., Bolvin, D. T., Nelkin, E. J., Wolff, D. B., Adler, R. F., Gu,
G., Hong, Y., Bowman, K. P., and Stocker, E. F.: The TRMM Multisatellite
Precipitation Analysis (TMPA): Quasi-Global, Multiyear, Combined-Sensor
Precipitation Estimates at Fine Scales, J. Hydrometeorol., 8, 38–55, 2007.
Hutley, L. B., O'Grady, A. P., and Eamus, D.: Evapotranspiration from
Eucalypt open-forest savanna of Northern Australia, Funct. Ecol., 14,
183–194, 2000.Ingwersen, J., Imukova, K., Högy, P., and Streck, T.: On the use of the
post-closure methods uncertainty band to evaluate the performance of land
surface models against eddy covariance flux data, Biogeosciences, 12,
2311–2326, 10.5194/bg-12-2311-2015, 2015.Jia, L., Su, Z., van den Hurk, B., Menenti, M., Moene, A., De Bruin, H. A.
R., Yrisarry, J. J. B., Ibanez, M., and Cuesta, A.: Estimation of sensible
heat flux using the Surface Energy Balance System (SEBS) and ATSR
measurements, Phys. Chem. Earth, 28, 75–88,
10.1016/S1474-7065(03)00009-3, 2003.
Jiménez, C., Prigent, C., Mueller, B., Seneviratne, S. I., McCabe, M. F.,
Wood, E. F., Rossow, W. B., Balsamo, G., Betts, a. K., Dirmeyer, P. a.,
Fisher, J. B., Jung, M., Kanamitsu, M., Reichle, R. H., Reichstein, M.,
Rodell, M., Sheffield, J., Tu, K., and Wang, K.: Global intercomparison of 12
land surface heat flux estimates, J. Geophys. Res., 116, 1–27, 2011.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., 116, G00J07,
10.1029/2010JG001566,
2011.
Justice, C., Townshend, J., Vermote, E., Masuoka, E., Wolfe, R. E., Saleous,
N., Roy, D. P., and Morisette, J. T.: An overview of MODIS Land data
processing and product status, Remote Sens. Environ., 83, 3–15, 2002.Kalma, J., McVicar, T., and McCabe, M.: Estimating Land Surface Evaporation:
A Review of Methods Using Remotely Sensed Surface Temperature Data, Surv.
Geophys., 29, 421–469, 10.1007/s10712-008-9037-z, 2008.
Kidd, C., Bauer, P., Turk, J., Huffman, G. J., Joyce, R., Hsu, K.-L., and
Braithwaite, D.: Intercomparison of High-Resolution Precipitation Products
over Northwest Europe, J. Hydrometeorol., 13, 67–83, 2012.Kim, C. P. and Entekhabi, D.: Feedbacks in the Land-Surface and Mixed-Layer
Energy Budgets, Bound.-Lay. Meteorol., 88, 1–21,
10.1023/A:1001094008513, 1998.
Kinne, S., O'Donnel, D., Stier, P., Kloster, S., Zhang, K., Schmidt, H.,
Rast, S., Giorgetta, M., Eck, T. F., and Stevens, B.: MAC-v1: A new global
aerosol climatology for climate studies, Journal of Advances in Modeling
Earth Systems, 5, 704–740, 2013.Knapp, K. R.: Scientific data stewardship of international satellite cloud
climatology project B1 global geostationary observations, J. Appl. Remote
Sens., 2, 023548, 10.1117/1.3043461, 2008.
Knapp, K. R., Ansari, S., Bain, C. L., Bourassa, M. A., Dickinson, M. J.,
Funk, C., Helms, C. N., Hennon, C. C., Holmes, C. D., Huffman, G. J., Kossin,
J. P., Lee, H.-T., Loew, A., and Magnusdottir, G.: Globally Gridded Satellite
Observations for Climate Studies, B. Am. Meteorol. Soc., 92, 893–907, 2011.
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, 2003.Kucera, P. A., Ebert, E. E., Turk, F. J., Levizzani, V., Kirschbaum, D.,
Tapiador, F. J., Loew, A., and Borsche, M.: Precipitation from Space:
Advancing Earth System Science, B. Am. Meteorol. Soc., 94, 365–375,
10.1175/BAMS-D-11-00171.1, 2013.
Lafleur, P. M.: Interannual variability in the peatland-atmosphere carbon
dioxide exchange at an ombrotrophic bog, Global Biogeochem. Cy., 17, 1–14,
2003.Lee, D.-H.: Comparing the inverse parameter estimation approach with
pedo-transfer function method for estimating soil hydraulic conductivity,
Geosci. J., 9, 269–276, 10.1007/BF02910587, 2005.
Leuning, R., Cleugh, H. A., Zegelin, S. J., and Hughes, D.: Carbon and water
fluxes over a temperate Eucalyptus forest and a tropical wet/dry savanna in
Australia: measurements and comparison with MODIS remote sensing estimates,
Agr. Forest Meteorol., 129, 151–173, 2005.
Maidment, D.: Handbook of Hydrology, McGraw-Hill Education,
1993.Mao, J., Fu, W., Shi, X., Ricciuto, D. M., Fisher, J. B., Dickinson, R. E.,
Wei, Y., Shem, W., Piao, S., and Wang, K.: Disentangling climatic and
anthropogenic controls on global terrestrial evapotranspiration trends,
Environ. Res. Lett., 10, 094008, 10.1088/1748-9326/10/9/094008, 2015.Margulis, S. A. and Entekhabi, D.: A Coupled Land Surface–Boundary Layer
Model and Its Adjoint, J. Hydrometeorol., 2, 274–296,
10.1175/1525-7541(2001)002<0274:ACLSBL>2.0.CO;2,
2001.Marx, A., Kunstmann, H., Schüttemeyer, D., and Moene, A. F.: Uncertainty
analysis for satellite derived sensible heat fluxes and scintillometer
measurements over Savannah environment and comparison to mesoscale
meteorological simulation results, Agr. Forest Meteorol., 148, 656–667,
10.1016/j.agrformet.2007.11.009, 2008.McCabe, M. F., Ershadi, A., Jimenez, C., Miralles, D. G., Michel, D., and
Wood, E. F.: The GEWEX LandFlux project: evaluation of model evaporation
using tower-based and globally gridded forcing data, Geosci. Model Dev., 9,
283–305, 10.5194/gmd-9-283-2016, 2016.McNaughton, K. G. and Spriggs, T. W.: A mixed-layer model for regional
evaporation, Bound.-Lay. Meteorol., 34, 243–262, 10.1007/BF00122381,
1986.
Meyers, T.: An assessment of storage terms in the surface energy balance of
maize and soybean, Agr. Forest Meteorol., 125, 105–115, 2004.Michel, D., Jiménez, C., Miralles, D. G., Jung, M., Hirschi, M., Ershadi,
A., Martens, B., McCabe, M. F., Fisher, J. B., Mu, Q., Seneviratne, S. I.,
Wood, E. F., and Fernández-Prieto, D.: The WACMOS-ET project – Part 1:
Tower-scale evaluation of four remote-sensing-based evapotranspiration
algorithms, Hydrol. Earth Syst. Sci., 20, 803–822,
10.5194/hess-20-803-2016, 2016.Miralles, D. G., Holmes, T. R. H., De Jeu, R. A. M., Gash, J. H., Meesters,
A. G. C. A., and Dolman, A. J.: Global land-surface evaporation estimated
from satellite-based observations, Hydrol. Earth Syst. Sci., 15, 453–469,
10.5194/hess-15-453-2011, 2011.Miralles, D. G., van den Berg, M. J., Gash, J. H., Parinussa, R. M., de Jeu,
R. A. M., Beck, H. E., Holmes, T. R. H., Jiménez, C., Verhoest, N. E. C.,
Dorigo, W. A., Teuling, A. J., and Johannes Dolman, A.: El Niño–La
Niña cycle and recent trends in continental evaporation, Nature Clim.
Change, 4, 122–126, 10.1038/nclimate2068, 2014.Miralles, D. G., Jiménez, C., Jung, M., Michel, D., Ershadi, A., McCabe,
M. F., Hirschi, M., Martens, B., Dolman, A. J., Fisher, J. B., Mu, Q.,
Seneviratne, S. I., Wood, E. F., and Fernández-Prieto, D.: The WACMOS-ET
project – Part 2: Evaluation of global terrestrial evaporation data sets,
Hydrol. Earth Syst. Sci., 20, 823–842, 10.5194/hess-20-823-2016, 2016.
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, 2009.Mu, Q., Heinsch, F. A., Zhao, M., and Running, S. W.: Development of a global
evapotranspiration algorithm based on MODIS and global meteorology data,
Remote Sens. Environ., 111, 519–536, 10.1016/j.rse.2007.04.015, 2007.
Mueller, B., Seneviratne, S. I., Jimenez, C., Corti, T., Hirschi, M.,
Balsamo, G., Ciais, P., Dirmeyer, P., Fisher, J. B., Guo, Z., Jung, M.,
Maignan, F., McCabe, M. F., Reichle, R., Reichstein, M., Rodell, M.,
Sheffield, J., Teuling, a. J., Wang, K., Wood, E. F., and Zhang, Y.:
Evaluation of global observations-based evapotranspiration datasets and IPCC
AR4 simulations, Geophys. Res. Lett., 38, 1–7, 2011.Mueller, B., Hirschi, M., Jimenez, C., Ciais, P., Dirmeyer, P. A., Dolman, A.
J., Fisher, J. B., Jung, M., Ludwig, F., Maignan, F., Miralles, D. G.,
McCabe, M. F., Reichstein, M., Sheffield, J., Wang, K., Wood, E. F., Zhang,
Y., and Seneviratne, S. I.: Benchmark products for land evapotranspiration:
LandFlux-EVAL multi-data set synthesis, Hydrol. Earth Syst. Sci., 17,
3707–3720, 10.5194/hess-17-3707-2013, 2013.Mueller, R. W., Matsoukas, C., Gratzki, A., Behr, H. D., and Hollmann, R.:
The CM-SAF operational scheme for the satellite based retrieval of solar
surface irradiance — A LUT based eigenvector hybrid approach, Remote Sens.
Environ., 113, 1012–1024, 10.1016/j.rse.2009.01.012, 2009.
Muller, J.-P., López, G., Watson, G., Shane, N., Kennedy, T., Yuen, P.,
Lewis, P., Fischer, J., Guanter, L., and Domench, C.: The ESA GlobAlbedo
Project for mapping the Earth's land surface albedo for 15 Years from
European Sensors, Geophys. Res. Abstr.,
EGU2012-10969, EGU General Assembly 2012, Vienna, Austria, 2012.Müller, R., Pfeifroth, U., Träger-Chatterjee, C., Trentmann, J., and Cremer,
R.: Digging the METEOSAT Treasure-3 Decades of Solar Surface Radiation, Remote Sens.,
7, 8067, 10.3390/rs70608067, 2015.Murray, T. and Verhoef, A.: Moving towards a more mechanistic approach in the
determination of soil heat flux from remote measurements: I. A universal
approach to calculate thermal inertia, Agr. Forest Meteorol., 147, 80–87,
10.1016/j.agrformet.2007.07.004, 2007.Myneni, R. B., Hoffman, S., Knyazikhin, Y., Privette, J. L., Glassy, J.,
Tian, Y., Wang, Y., Song, X., Zhang, Y., Smith, G. R., Lotsch, A., Friedl,
M., Morisette, J. T., Votava, P., Nemani, R. R., and Running, S. W.: Global
products of vegetation leaf area and fraction absorbed PAR from year one of
MODIS data, Remote Sens. Environ., 83, 214–231,
10.1016/S0034-4257(02)00074-3, 2002.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.Norman, J., Anderson, M., Kustas, W., French, A., Mecikalski, J., Torn, R.,
Diak, G., Schmugge, T., and Tanner, B.: Remote sensing of surface energy
fluxes at 101-m pixel resolutions, Water Resour. Res., 39, 1221, 10.1029/2002WR001775,
2003.Norman, J. M., Kustas, W. P., and Humes, K. S.: Source approach for
estimating soil and vegetation energy fluxes in observations of directional
radiometric surface temperature, Agr. Forest Meteorol., 77, 263–293,
10.1016/0168-1923(95)02265-Y, 1995.
Papale, D., Agarwal, D., Baldocchi, D., Cook, R., Fisher, J., and van Ingen,
C.: Database Maintenance, Data Sharing Policy, Collaboration, in: Eddy
Covariance, edited by: Aubinet, M., Vesala, T., and Papale, D., Springer
Atmospheric Sciences, Springer Netherlands, 399-424, 2012.Paulson, C. A.: The Mathematical Representation of Wind Speed and Temperature
Profiles in the Unstable Atmospheric Surface Layer, J. Appl. Meteorol., 9,
857–861,
10.1175/1520-0450(1970)009<0857:TMROWS>2.0.CO;2,
1970.Peng, J. and Loew, A.: Evaluation of Daytime Evaporative Fraction from MODIS
TOA Radiances Using FLUXNET Observations, Remote Sens., 6, 5959, 10.3390/rs6075959,
2014.Peng, J., Borsche, M., Liu, Y., and Loew, A.: How representative are
instantaneous evaporative fraction measurements of daytime fluxes?, Hydrol.
Earth Syst. Sci., 17, 3913–3919, 10.5194/hess-17-3913-2013, 2013a.Peng, J., Liu, Y., Zhao, X., and Loew, A.: Estimation of evapotranspiration
from MODIS TOA radiances in the Poyang Lake basin, China, Hydrol. Earth Syst.
Sci., 17, 1431–1444, 10.5194/hess-17-1431-2013, 2013b.Pinty, B., Andredakis, I., Clerici, M., Kaminski, T., Taberner, M.,
Verstraete, M., Gobron, N., Plummer, S., and Widlowski, J. L.: Exploiting the
MODIS albedos with the Two-stream Inversion Package (JRC-TIP): 1. Effective
leaf area index, vegetation, and soil properties, J. Geophys. Res.-Atmos.,
116, D09105, 10.1029/2010JD015372, 2011.
Posselt, R., Mueller, R. W., Stöckli, R., and Trentmann, J.: Remote
sensing of solar surface radiation for climate monitoring – the CM-SAF
retrieval in international comparison, Remote Sens. Environ., 118, 186–198,
2012.Potts, D. R., Mackin, S., Muller, J. P., and Fox, N.: Sensor Intercalibration
Over Dome C for the ESA GlobAlbedo Project, IEEE T. Geosci. Remote Sens.,
51, 1139–1146, 10.1109/TGRS.2012.2217749, 2013.
Rawls, W. J. and Brakensiek, D.: Prediction of soil water properties for
hydrologic modeling, in: Watershed Management in the Eighties, edited by: Jones, E. B.,
293–299, American Society of Civil Engineers, New York, USA,
1985.Rebmann, C., Zeri, M., Lasslop, G., Mund, M., Kolle, O., Schulze, E.-D., and
Feigenwinter, C.: Treatment and assessment of the CO2-exchange at a
complex forest site in Thuringia, Germany, Agr. Forest Meteorol., 150,
684–691, 2010.
Reichstein, M., Falge, E., Baldocchi, D., Papale, D., Aubinet, M., Berbigier,
P., Bernhofer, C., Buchmann, N., Gilmanov, T., and Granier, A.: On the
separation of net ecosystem exchange into assimilation and ecosystem
respiration: review and improved algorithm, Glob. Change Biol., 11,
1424–1439, 2005.Ren, D. and Xue, M.: A Revised Force–Restore Model for Land Surface
Modeling, J. Appl. Meteorol., 43, 1768–1782, 10.1175/JAM2161.1, 2004.Richards, L. A.: Capillary conduction of liquids through porous mediums, J.
Appl. Phys., 1, 318–333, 10.1063/1.1745010, 1931.Roads, J. and Betts, A.: NCEP–NCAR and ECMWF Reanalysis Surface Water and
Energy Budgets for the Mississippi River Basin, J. Hydrometeorol., 1, 88–94,
10.1175/1525-7541(2000)001<0088:NNAERS>2.0.CO;2,
2000.Roerink, G. J., Su, Z., and Menenti, M.: S-SEBI: A simple remote sensing
algorithm to estimate the surface energy balance, Phys. Chem. Earth Pt. B,
25, 147–157, 10.1016/S1464-1909(99)00128-8, 2000.
Rossow, W. W. B. and Schiffer, R. A. R.: ISCCP Cloud Data Products, B. Am.
Meteorol. Soc., 72, 2–20, 1991.
Scherer-Lorenzen, M., Schulze, E., Don, A., Schumacher, J., and Weller, E.:
Exploring the functional significance of forest diversity: A new long-term
experiment with temperate tree species (BIOTREE), Perspect. Plant Ecol., 9,
53–70, 2007.Smeda, M.: A bulk model for the atmospheric planetary boundary layer,
Bound.-Lay. Meteorol., 17, 411–427, 10.1007/BF00118608, 1979.Su, Z.: The Surface Energy Balance System (SEBS) for estimation of turbulent
heat fluxes, Hydrol. Earth Syst. Sci., 6, 85–100,
10.5194/hess-6-85-2002, 2002.Tang, R., Li, Z.-L., and Chen, K.-S.: Validating MODIS-derived land surface
evapotranspiration with in situ measurements at two AmeriFlux sites in a
semiarid region, J. Geophys. Res., 116, D04106, 10.1029/2010JD014543, 2011a.Tang, R., Li, Z.-L., Jia, Y., Li, C., Sun, X., Kustas, W. P., and Anderson,
M. C.: An intercomparison of three remote sensing-based energy balance models
using Large Aperture Scintillometer measurements over a wheat–corn
production region, Remote Sens. Environ., 115, 3187-3202,
10.1016/j.rse.2011.07.004, 2011b.
Taylor, K. E.: Summarizing multiple aspects of model performance in a single
diagram, J. Geophys. Res., 106, 7183–7192, 2001.
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, 2006.
Twine, T. E., Kustas, W. P., Norman, J. M., Cook, D. R., Houser, P. R.,
Meyers, T. P., Prueger, J. H., Starks, P. J., and Wesely, M. L.: Correcting
eddy-covariance flux underestimates over a grassland, Agr. Forest Meteorol.,
103, 279–300, 2000.Valente, F., David, J. S., and Gash, J. H. C.: Modelling interception loss
for two sparse eucalypt and pine forests in central Portugal using
reformulated Rutter and Gash analytical models, J. Hydrol., 190, 141–162,
10.1016/S0022-1694(96)03066-1, 1997.
van Genuchten, M. T.: A closed-form equation for predicting the hydraulic
conductivity of unsaturated soils, Soil Sci. Soc. Am. J., 44, 892–898, 1980.
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, 2005.Vinukollu, R. K., Wood, E. F., Ferguson, C. R., and Fisher, J. B.: Global
estimates of evapotranspiration for climate studies using multi-sensor remote
sensing data: Evaluation of three process-based approaches, Remote Sens.
Environ., 115, 801–823, 10.1016/j.rse.2010.11.006, 2011.
von Hoyningen-Huene, J.: Die Interzeption des Niederschlags in
landwirtschaftlichen Pflanzenbeständen, Arbeitsbericht Deutscher Verband
für Wasserwirtschaft und Kulturbau, DVWK, 1981.Wang, K. and Dickinson, R. E.: A review of global terrestrial
evapotranspiration: Observation, modeling, climatology, and climatic
variability, Rev. Geophys., 50, RG2005, 10.1029/2011RG000373, 2012.
Wang, Y., Li, X., and Tang, S.: Validation of the SEBS-derived sensible heat
for FY3A/VIRR and TERRA/MODIS over an alpine grass region using LAS
measurements, Int. J. Appl. Earth Obs., 23, 226–233,
10.1016/j.jag.2012.09.005, 2013.
Wilson, K., Goldstein, A., Falge, E., Aubinet, M., Baldocchi, D., Berbigier,
P., Bernhofer, C., Ceulemans, R., Dolman, H., Field, C., Grelle, A., Ibrom,
A., Law, B. E., Kowalski, A., Meyers, T., Moncrieff, J., Monson, R., Oechel,
W., Tenhunen, J., Valentini, R., and Verma, S.: Energy balance closure at
FLUXNET sites, Agr. Forest Meteorol., 113, 223–243, 2002.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., 113, D08110, 10.1029/2007JD009286, 2008.Wösten, J. H. M., Pachepsky, Y. A., and Rawls, W. J.: Pedotransfer
functions: bridging the gap between available basic soil data and missing
soil hydraulic characteristics, J. Hydrol., 251, 123–150,
10.1016/S0022-1694(01)00464-4, 2001.Yang, F., Zhu, A.-X., Ichii, K., White, M. A., Hashimoto, H., and Nemani, R.
R.: Assessing the representativeness of the AmeriFlux network using MODIS and
GOES data, J. Geophys. Res., 113, G04036, 10.1029/2007JG000627, 2008.
Yuan, H., Dai, Y., Xiao, Z., Ji, D., and Shangguan, W.: Reprocessing the
MODIS Leaf Area Index products for land surface and climate modelling, Remote
Sens. Environ., 115, 1171–1187, 2011.Zhang, K., Kimball, J. S., Nemani, R. R., and Running, S. W.: A continuous
satellite-derived global record of land surface evapotranspiration from 1983
to 2006, Water Resour. Res., 46, W09522, 10.1029/2009WR008800, 2010.Zhang, K., Kimball, J. S., Nemani, R. R., Running, S. W., Hong, Y., Gourley,
J. J., and Yu, Z.: Vegetation Greening and Climate Change Promote
Multidecadal Rises of Global Land Evapotranspiration, Scientific Reports, 5,
15956, 10.1038/srep15956, 2015.
Zhang, X., Berhane, T., and Seielstad, G.: Comparision of Landsat and MODIS
Estimates of Heat Fluxes: Effect of Surface Heterogeniety, IGARSS 2008–2008
IEEE International Geoscience and Remote Sensing Symposium, 759–762, 2008.Zhang, Y., Peña-Arancibia, J. L., McVicar, T. R., Chiew, F. H. S., Vaze,
J., Liu, C., Lu, X., Zheng, H., Wang, Y., Liu, Y. Y., Miralles, D. G., and
Pan, M.: Multi-decadal trends in global terrestrial evapotranspiration and
its components, Scientific Reports, 6, 19124, 10.1038/srep19124, 2016.Zhuang, Q., Wu, B., Yan, N., Zhu, W., and Xing, Q.: A method for sensible
heat flux model parameterization based on radiometric surface temperature and
environmental factors without involving the parameter KB-1, International
Journal of Applied Earth Observation and Geoinformation, 47, 50–59,
10.1016/j.jag.2015.11.015, 2016.