The Global Land Evaporation Amsterdam Model (GLEAM) is a set of
algorithms dedicated to the estimation of terrestrial evaporation and
root-zone soil moisture from satellite data. Ever since its development in
2011, the model has been regularly revised, aiming at the optimal
incorporation of new satellite-observed geophysical variables, and improving
the representation of physical processes. In this study, the next version of
this model (v3) is presented. Key changes relative to the previous version
include (1) a revised formulation of the evaporative stress, (2) an optimized
drainage algorithm, and (3) a new soil moisture data assimilation system.
GLEAM v3 is used to produce three new data sets of terrestrial evaporation
and root-zone soil moisture, including a 36-year data set spanning
1980–2015, referred to as v3a (based on satellite-observed soil moisture,
vegetation optical depth and snow-water equivalent, reanalysis air
temperature and radiation, and a multi-source precipitation product), and two
satellite-based data sets. The latter share most of their forcing, except for
the vegetation optical depth and soil moisture, which are based on
observations from different passive and active C- and L-band microwave
sensors (European Space Agency Climate Change Initiative, ESA CCI) for the
v3b data set (spanning 2003–2015) and observations from the Soil Moisture
and Ocean Salinity (SMOS) satellite in the v3c data set (spanning
2011–2015). Here, these three data sets are described in detail, compared
against analogous data sets generated using the previous version of GLEAM
(v2), and validated against measurements from 91 eddy-covariance towers and
2325 soil moisture sensors across a broad range of ecosystems. Results
indicate that the quality of the v3 soil moisture is consistently better than
the one from v2: average correlations against in situ surface soil moisture
measurements increase from 0.61 to 0.64 in the case of the v3a data set and
the representation of soil moisture in the second layer improves as well,
with correlations increasing from 0.47 to 0.53. Similar improvements are
observed for the v3b and c data sets. Despite regional differences, the
quality of the evaporation fluxes remains overall similar to the one obtained
using the previous version of GLEAM, with average correlations against
eddy-covariance measurements ranging between 0.78 and 0.81 for the different
data sets. These global data sets of terrestrial evaporation and root-zone
soil moisture are now openly available at
Climate change alters the complex interplay between land and atmosphere,
significantly impacting different processes in the global hydrological cycle
However, the climate community is becoming increasingly aware of the crucial
role that terrestrial evaporation plays in the Earth system, acting as a link
in hydrological and biogeochemical cycles, and being a driver of air
humidity, cloud formation, temperature, or precipitation
Schematic of the four modules of GLEAM.
Existing algorithms share the overarching objective of producing consistent,
long-term, global data sets of terrestrial evaporation, but the methods and
input data sets used in these models differ markedly
The Global Land Evaporation Amsterdam Model – GLEAM
The main goal of this study is to present the new version of GLEAM and the
resulting evaporation and root-zone soil moisture data sets, including a
global validation using a large database of soil moisture measurements from
2325 in situ sensors, and evaporation measurements from 91 eddy-covariance
towers. In addition, the quality of these data sets is compared against
analogous data sets generated using the former version of GLEAM, allowing us
to evaluate the added value of the new formulations. The paper is organized
as follows: Sect.
GLEAM separately derives the different components of terrestrial evaporation,
i.e. transpiration, bare soil evaporation, open-water evaporation,
interception loss, and sublimation (
Estimates of
Schematic of the water-balance module implemented in GLEAM v3 for
the fraction of tall vegetation.
Figure
At every daily time step
In previous model versions, the entire volume of net precipitation (i.e.
precipitation minus interception loss, plus snowmelt) was first stored in the
top layer, which subsequently drained to field capacity into the next soil
layer
The rationale behind this simple drainage algorithm is that the downward flux
of water is expected to increase if (1) the relative soil moisture content is
higher (physically resulting in increased hydraulic conductivities), and
(2) the difference in soil moisture between source and sink is larger
(resulting in higher differences in soil-water potential). This empirical
drainage algorithm is preferred over well-known alternatives such as the
Richards equation
The original Kalman filter approach to assimilate microwave soil moisture
observations – typically sensitive to the first few
centimetres of the soil – into GLEAM
was replaced in favour of a simple Newtonian nudging algorithm in v2
As most assimilation algorithms require bias-free observations in reference
to the modelled states, a bias removal algorithm prior to (or during) the
assimilation step has to be applied. However, no standard procedure exists to
correct these constant or seasonally varying biases
Illustration of the stress function implemented in GLEAM v3 for the fractions of short and tall vegetation (colours relate to the VOD). Left-hand side: pixel with high range in VOD (
Finally, in contrast to the assimilation of soil moisture observations in all
model layers in GLEAM v2
Water availability, heat stress, or phenological constraints acting on
evaporation are generally combined in a single empirical stress factor
accounting for the decrease in potential evaporation
Figure
Finally, for the bare soil fraction,
List of the selected forcing data sets together with their references, the original spatial resolution, and period of availability. The first column indicates the use of these data in GLEAM.
Table
Radiation inputs are based on measurements from the Clouds and Earth's
Radiant Energy System (CERES) onboard Terra and Aqua
As discussed in Sect.
Using various combinations of the forcing data, three different data sets of
terrestrial evaporation and root-zone soil moisture are produced using
GLEAM v3 (see also Table
Average validation statistics for the different soil moisture data
sets (v3a, v3b, and v3c) and for the first two model layers (
For validation purposes, in situ soil moisture and evaporation measurements
from different global networks are processed. Soil moisture measurements are
sourced from the database of the International Soil Moisture Network (ISMN,
Difference in quality between the v3 and v2 data sets of surface
soil moisture
(
Average anomaly correlations for different soil moisture data sets
(v3a and v3b) and for the first two model layers (
Table
As indicated by the statistics in Table
Impact of the data assimilation system in GLEAM v3 on the surface
soil moisture for the CONUS. Left-hand side figures show the difference in
correlations against in situ measurements for the GLEAM v3 surface soil
moisture data sets with and without (open loop) the assimilation of
satellite-derived soil moisture
(
For comparison, Table
Finally, to better evaluate the skill of GLEAM v3 in capturing the effect of
specific precipitation events on the estimated soil moisture – without the
influence of the seasonal cycle – correlations between the anomaly time
series of GLEAM soil moisture and the anomaly time series of in situ soil
moisture are also calculated (
The left-hand side panel in Fig.
The negative impact of assimilating satellite observations of surface soil
moisture in the v3a data set is partly explained by the high quality of the
GLEAM open-loop soil moisture compared to the quality of the satellite-based
soil moisture data set (ESA CCI SM v2.3): correlations are significantly
better for the open loop at 73 % of the individual sites. The high quality
of the model open loop in these regions is largely due to the accuracy of the
MSWEP precipitation forcing in the CONUS domain; this is illustrated in the
central panel in Fig.
Impact of the data assimilation system in GLEAM v3 on the surface
soil moisture for the CONUS. Left-hand side figures show the difference in
anomaly correlations against in situ measurements for the GLEAM v3 surface
soil moisture data sets with and without (open loop) the assimilation of
satellite-derived soil moisture
(
Finally, it may be argued that differences in quality between the
satellite-derived and modelled soil moisture should be reflected in the
TCA-based quality factor (
Average validation statistics for the different evaporation data
sets (v3a, v3b, and v3c) against in situ measurements:
Table
Average anomaly correlations for different evaporation data sets
(v3a and v3b) against in situ measurements:
Difference in quality between the v3 and v2 data sets of terrestrial evaporation (
Time series of GLEAM and in situ measured evaporation for two in situ validation sites: US-Ne3 (left) and AU-ASM (right).
As an example, Fig.
Global maps of terrestrial evaporation (top row) and the partitioning into its different components, i.e. forest interception loss (second row), transpiration (third row), and bare soil evaporation (bottom row) for the v3a data set. On top, the multi-annual total flux of evaporation for the v3a data set (left) and the difference with the v2a data set (right) are shown. The other maps show the percentage of the total flux in the v3a data set coming from the different components (left) and the difference with the same maps for the v2a data set (right).
The top row in Fig.
The remaining maps in Fig.
The available range of satellite-observable geophysical variables that relate
to the process of evaporation – such as soil moisture, air temperature, and
net radiation – is continuously growing and the quality of these data sets
is constantly improving. As a result, models aiming at the accurate
estimation of terrestrial evaporation from satellite observations need to be
updated to optimally incorporate these new data. Concurrently, as our
knowledge of the relevant physical processes advances based on new
experimental evidence, these simple retrieval models should aim to increase
their realism. With the overarching goal of improving our understanding of
continental evaporation, a next version of the Global Land Evaporation
Amsterdam Model (GLEAM v3) – a set of algorithms dedicated to the estimation
of global terrestrial evaporation from satellite data – is presented in this
paper. Three major modifications are included: (1) a revised representation
of the evaporative stress, (2) an optimized water-balance module, and (3) a
new soil moisture data assimilation strategy. Using GLEAM v3, three novel
data sets of root-zone soil moisture and terrestrial evaporation are
presented. The first data set (v3a) spans the 36-year period 1980–2015, has
a global coverage, and is produced using satellite-observed soil moisture,
vegetation optical depth and snow-water equivalent, reanalysis air
temperature and radiation, and a multi-source precipitation product. The
remaining two data sets (v3b and v3c) are produced using satellite-based
forcing only, with their difference being the use of SMOS-based VOD and soil
moisture (v3c), as opposed to the corresponding CCI forcing (v3b). Both data
sets are quasi-global (50
Results based on the validation of these three data sets against an extensive
set of in situ measured evaporation and soil moisture point to a slightly
higher quality of the v3a soil moisture data set as compared to the other two
data sets, while the quality of the modelled evaporation is rather similar
across all three. The higher accuracy of the v3a soil moisture is explained
by the high quality of the MSWEP precipitation forcing over the regions where
soil moisture probes are located, compared to the satellite-based forcing in
the v3b and v3c data sets. Results, however, might be biased given that the
vast majority (i.e. more than 75 %) of the in situ soil moisture sites are
located in the CONUS, where gauge-based precipitation products are known to
outperform satellite products
Based on the results in this study, it can be concluded that the
modifications in GLEAM have led to a more realistic representation of
physical processes and an overall increased quality of the data sets,
particularly in the case of the root-zone soil moisture. Following the
advances in satellite technology and the increased availability of new data,
GLEAM will be further optimized in coming years. Current activities
concentrate on the incorporation of new constraints on evaporation, the
application of GLEAM to higher resolutions and near-real time, and the
improved partitioning of evaporation into its different components.
Meanwhile, the data sets of terrestrial evaporation and root-zone soil
moisture presented in this study have been made available for studies of
hydrological cycle dynamics and climate model benchmarking using
The model code of GLEAM v3 is available upon request
from the corresponding author. Data sets described in this paper can be
freely accessed from
List of the FLUXNET sites used in this study together with their FLUXNET code (ID), IGBP land cover (LC), and official reference (or principal investigator (PI)).
All authors have been involved in interpreting the results, discussing the findings, and editing the manuscript. B. Martens, H. Lievens, D. G. Miralles, and N. E. C. Verhoest designed the modifications in GLEAM v3. B. Martens, R. v. d. Schalie, H. E. Beck, and W. A. Dorigo processed the forcing and validation data. B. Martens did the analyses. B. Martens and D. G. Miralles designed the layout of the paper and wrote the draft of the manuscript.
The authors declare that they have no conflict of interest.
This work has been funded by the ESA's Support To Science Element through the SMOS+ET II project (IPL-POE-2015-723-LG-cb-LE). D. G. Miralles acknowledges support from the European Research Council (ERC) under grant agreement no. 715254 (DRY-2-DRY). H. Lievens is a postdoctoral research fellow of the Research Foundation Flanders (FWO). W. A. Dorigo is supported by personal grant “TU Wien Wissenschaftspreis 2015” and the ESA's Climate Change Initiative for Soil Moisture (contract no. 4000112226/14/I-NB). The authors would like to thank the principal investigators of the International Soil Moisture Network (ISMN). This work used eddy-covariance data acquired and shared by the FLUXNET community, including these networks: AmeriFlux, AfriFlux, AsiaFlux, CarboAfrica, CarboEuropeIP, CarboItaly, CarboMont, ChinaFlux, Fluxnet-Canada, GreenGrass, ICOS, KoFlux, LBA, NECC, OzFlux-TERN, TCOS-Siberia, and USCCC. The FLUXNET eddy-covariance data processing and harmonization were carried out by the ICOS Ecosystem Thematic Center, the AmeriFlux Management Project, and the Fluxdata project of FLUXNET, with the support of CDIAC, and the OzFlux, ChinaFlux, and AsiaFlux offices. The authors also acknowledge the Instituto Nacional Technologica Agropecuaria for making the eddy-covariance data of AR-Vir publicly available.Edited by: H. McMillan Reviewed by: two anonymous referees