Creation of the WATCH Forcing Data and
Its Use to Assess Global and Regional
Reference Crop Evaporation over Land
during the Twentieth Century
Article
Published Version
Weedon, G. P., Gomes, S., Viterbo, P., Shuttleworth, W. J.,
Blyth, E., Österle, H., Adam, J. C., Bellouin, N., Boucher, O.
and Best, M. (2011) Creation of the WATCH Forcing Data and
Its Use to Assess Global and Regional Reference Crop
Evaporation over Land during the Twentieth Century. Journal
of Hydrometeorology, 12 (5). pp. 823-848. ISSN 1525-7541
doi: https://doi.org/10.1175/2011JHM1369.1 Available at
https://centaur.reading.ac.uk/34633/
It is advisable to refer to the publisher’s version if you intend to cite from the
work. See Guidance on citing
.
Published version at: http://dx.doi.org/10.1175/2011JHM1369.1
To link to this article DOI: http://dx.doi.org/10.1175/2011JHM1369.1
Publisher: American Meteorological Society
All outputs in CentAUR are protected by Intellectual Property Rights law,
including copyright law. Copyright and IPR is retained by the creators or other
copyright holders. Terms and conditions for use of this material are defined in
the End User Agreement
.
Creation of the WATCH Forcing Data and Its Use to Assess Global and Regional
Reference Crop Evaporation over Land during the Twentieth Century
G. P. WEEDON,* S. GOMES,
1
P. VITERBO,
1
W. J. SHUTTLEWORTH,
#
E. BLYTH,
@
H. O
¨
STERLE,
&
J. C. ADAM,** N. BELLOUIN,
11
O. BOUCHER,
11
AND M. BEST
11
* Met Office Hadley Centre, Joint Centre for Hydrometeorological Research, Wallingford, United Kingdom
1
Instituto Dom Luiz, University of Lisbon, Lisbon, Portugal
#
Department of Hydrology and Water Resources, The University of Arizona, Tucson, Arizona
@
Centre for Ecology and Hydrology, Joint Centre for Hydrometeorological Research, Wallingford, United Kingdom
&
PIK Potsdam, Potsdam, Germany
** Washington State University, Pullman, Washington
11
Met Office Hadley Centre, Exeter, United Kingdom
(Manuscript received 3 September 2010, in final form 24 February 2011)
ABSTRACT
The Water and Global Change (WATCH) project evaluation of the terrestrial water cycle involves using
land surface models and general hydrological models to assess hydrologically important variables including
evaporation, soil moisture, and runoff. Such models require meteorological forcing data, and this paper de-
scribes the creation of the WATCH Forcing Data for 1958–2001 based on the 40-yr ECMWF Re-Analysis
(ERA-40) and for 1901–57 based on reordered reanalysis data. It also discusses and analyses model-
independent estimates of reference crop evaporation. Global average annual cumulative reference crop
evaporation was selected as a widely adopted measure of potential evapotranspiration. It exhibits no
significant trend from 1979 to 2001 although there are significant long-term increases in global average
vapor pressure deficit and concurrent significant decreases in global average net radiation and wind speed.
The near-constant global average of annual reference crop evaporation in the late twentieth century masks
significant decreases in some regions (e.g., the Murray–Darling basin) with significant increases in others.
1. Introduction
As the earth’s whole climate system slowly changes
there are likely to be greater and faster regional changes.
Studies of the impacts of these changes on essential
services such as fresh water supply are being made by
many researchers (e.g., Harding et al. 2011) with the
change in evaporation being a key aspect. Observations
of large-scale evaporation over the last half century (the
most studied period) are, however, not available. Con-
sequently, models of evaporation are frequently used
as an alternative. In such models the key factors that
determine changes in evaporation are changes in mete-
orological factors such as radiation, wind speed, air
temperature, and humidity.
Studies have analyzed pan evaporation data (Roderick
and Farquhar 2002; Roderick et al. 2007) and reported
changes in the external drivers on evaporation when there
is no change in available water. In Australia these studies
have demonstrated that large-scale change in wind speed
(global stilling) is responsible for an observed drop in pan
evaporation, although decreases–increases in radiation
(global dimming–brightening) are perhaps responsible
for changes elsewhere. Shuttleworth et al. (2009) dem-
onstrated that it is not always possible to use pan evapo-
ration to diagnose large-scale change in external drivers
of actual evaporation. This is because some changes in the
drivers of pan evaporation are caused by feedbacks in the
atmospheric planetary boundary layer caused by altered
actual evaporation in the area surrounding the pan.
However, they also demonstrated that it is not possible to
assume that changes in pan evaporation are equal and
opposite to changes in surrounding actual evaporation, as
suggested by Bouchet (1963), since changes in the vari-
ables controlling evaporation are a mixture of regional
Corresponding author address: G. P. Weedon, Met Office Had-
ley Centre, Joint Centre for Hydrometerological Research, Ma-
clean Building, Wallingford OX10 8BB, United Kingdom.
E-mail: graham.weedon@metoffice.gov.uk
O
CTOBER 2011 W E E D O N E T A L . 823
DOI: 10.1175/2011JHM1369.1
Ó 2011 American Meteorological Society
atmospheric feedbacks superposed on modified large-
scale atmospheric circulation.
In their comprehensive review, Hobbins et al. (2008)
point out that researchers interested in global evaporation
need an accurate assessment of the external drivers of the
evaporation process. However, because of nonlinearity
in the relationships between the drivers of evaporation
(particularly temperature) it is not possible to make such
an assessment using daily average meteorological data.
Instead, accurate assessment requires data that resolve
the full diurnal cycle. This paper describes the creation of
the Water and Global Change (WATCH) Forcing Data
(WFD), a dataset that is available for the whole of the
twentieth century and that resolves the full diurnal cycle.
An analysis of changes in the external drivers of evapora-
tion that is relevant to both researchers and water-resource
engineers is also made.
The European Union WATCH project (www.eu-watch.
org) seeks to assess the terrestrial water cycle in the con-
text of global change in the twentieth and twenty-first
centuries. A major component of the study is use of land
surface models (LSMs) and general hydrological models
(GHMs) to calculate changes in hydrologically important
variables such as evaporation, soil moisture, and runoff
(Haddeland et al. 2011). For both types of model, mete-
orological ‘‘forcing’’ (or ‘‘driving’’) data (air temperature,
rainfall/snowfall, etc.) are required at subdaily time steps
for the LSMs and daily time steps for the GHMs. The
40-yr European Centre for Medium-Range Weather
Forecasts (ECMWF) Re-Analysis (ERA-40) product,
which provided the basis data used in the derivation of
the WFD, was derived from successive short-term in-
tegrations of a general circulation model (GCM) that
assimilated [via three-dimensional variational data as-
similation (3D-Var)] various satellite data along
with atmospheric soundings and land and sea surface
observations (Uppala et al. 2005). The reanalysis pro-
cedure used to create ERA-40 merged global subdaily
observations with a prior estimate based on short in-
tegrations of a comprehensive GCM, allowing for un-
certainties in each, using a GCM configuration that was
consistent, as opposed to the progressively refined and
improved GCMs that are used in routine weather fore-
casting. As explained below, the WFD were derived from
the surface variables of the ERA-40 reanalysis product for
the period 1958 to 2001, but from reordered ERA-40 data
for the period 1901 to 1957.
The several models involved in the WATCH project
calculate hydrological variables using the WFD in dif-
ferent ways, but a key aspect of the models is the way in
which evaporation is estimated (Haddeland et al. 2011).
LSMs typically estimate actual evaporation by evaluating
the energy balance at the subdaily time scale, whereas
GHMs typically require estimates of daily-average ‘‘po-
tential’’ evapotranspiration and then assess actual evapo-
ration by adjusting this estimate to allow for the water
availability. In this paper an assessment is made of changes
in global twentieth-century potential evaporation inde-
pendent of any specific LSM or GHM as estimated via the
WFD themselves. Consideration is also given to regional
variations in the selected large river basins shown in Fig. 1.
2. The WATCH Forcing Data
The WFD consist of subdaily, regularly (latitude–
longitude) gridded, half-degree resolution, meteoro-
logical forcing data. The variables included are (i) wind
speed at 10 m, (ii) air temperature at 2 m, (iii) surface
pressure, (iv) specific humidity at 2 m, (v) downward
longwave radiation flux, (vi) downward shortwave radia-
tion flux, (vii) rainfall rate, and (viii) snowfall rate. These
global data are stored at 67 420 points over land (excluding
FIG. 1. Location map for the FLUXNET sites used in Figs. 2–4 (indicated by plus signs) and for
the large river basins considered in Figs. 7–9 (indicated in black).
824 JOURNAL OF HYDROMETEOROLOGY VOLUME 12
the Antarctic), with the land–sea mask used being that
defined by the Climatic Research Unit (CRU; New et al.
1999, 2000) in netCDF format using the Assistance for
Land-Surface Modelling Activities (ALMA) conven-
tion (see http://www.lmd.jussieu.fr/;polcher/ALMA/).
Variables vi–viii are not readily interpolated and are
stored at three-hourly time steps as in the basic ERA-40
data, but to save space variables i–v are stored at
6-hourly time steps with code provided to give variable-
dependent interpolation to the three-hourly time step.
a. WATCH Forcing Data 1958–2001
1) I
NTRODUCTION
Generation of the WFD for the late twentieth century
described in detail by Weedon et al. (2010) adopted the
procedures described by Ngo-Duc et al. (2005) and
Sheffield et al. (2006), but with the changes summarized
in Table 1. Processing involved bilinear interpolation of
each variable from the 18 ERA-40 grid to the 0.58 CRU
land–sea mask. To maintain consistency, elevation cor-
rections were then made sequentially to the interpolated
temperature, surface pressure, specific humidity, and
downward longwave radiation (in that order, because
elevation correction of later variables requires use of
previously corrected variables).
In several respects the ERA-40 data product is su-
perior to the earlier National Center for Atmospheric
Research–National Centers for Environmental Pre-
diction (NCAR–NCEP) reanalysis used in deriving
other forcing datasets (e.g., Uppala et al. 2005), but the
2-m temperatures in ERA-40 are known to lack some
climatic trends and to exhibit an overall bias (Betts and
Beljaars 2003; Simmons et al. 2004; Hagemann et al.
2005) despite the assimilation of relevant surface ob-
servations. Comparison of diurnal extremes in near-
surface temperature in the NCAR–NCEP, ERA-40, and
the (more recent) Japanese Meteorological Agency
(JMA) 25-yr reanalysis (JRA-25) data reveals problems
in all 3 data products (Pitman and Perkins 2009), par-
ticularly with respect to minimum temperature. For this
reason the monthly average interpolated and elevation-
corrected temperatures from ERA-40 were also bias-
corrected (Weedon et al. 2010). Because the CRU3 data
(Brohan et al. 2006) were not available at 0.58 resolution
for all the required observations during creation of the
WFD, CRU TS2.1 gridded observations were used for
this bias correction (New et al. 1999, 2000; Mitchell and
Jones 2005).
The use of CRU observations for monthly bias cor-
rection inevitably incorporates inaccuracies related to
creation of the gridded products. Nevertheless, the
CRU interpolation methodology based on 1961–90
anomalies (New et al. 1999, 2000) includes allowance
for the ‘‘correlation length’’ of the variables involved,
and elevation corrections and inhomogeneities be-
tween stations have been adjusted while the variable
station coverage through time and spatially is docu-
mented by New et al. (1999, 2000) and Mitchell and
Jones (2005). Despite these limitations the CRU data-
set has been widely used for investigating global ter-
restrial changes through the twentieth century (e.g.,
De
´
ry and Wood 2005; Gedney et al. 2006; Dang et al.
2007; Piao et al. 2009).
The CRU temperature data used include some (albeit
rare) inhomogeneities. Specifically, there were steplike
TABLE 1. Creation of the meteorological variables in the WFD.
Meteorological
variable
Elevation correction after bilinear
interpolation Data used for monthly bias correction
10-m wind speed Nil Nil
2-m temperature Via environmental lapse rate CRU average temperature (corrected) and
average diurnal temperature range.
10-m surface pressure Via changes in 2-m temperature Nil
2-m specific humidity Via changes in 2-m temperature
and surface pressure
Nil
Downward longwave radiation Via fixed relative humidity, changes
in 2-m temperature, surface
pressure, and specific humidity
Nil
Downward shortwave radiation Nil CRU average cloud cover and effects of
changing atmospheric aerosol loading.
Rainfall rate Nil CRU number of ‘‘wet days’’, GPCCv4
precipitation totals, ERA-40 rainfall/total
proportion, and rainfall gauge catch corrections.
Snowfall rate Nil CRU number of ‘‘wet’’ days, GPCCv4
precipitation totals, ERA-40 snowfall/total
proportion, and snowfall gauge corrections.
O
CTOBER 2011 W E E D O N E T A L . 825
