Soil Moisture

1
Soil Moisture Extremes in European Climate
Heather S. Ashton12, Chris M. Taylor2, Pier-Luigi
Vidale1 and Julia Slingo1 1Department of
Meteorology, University of Reading, UK 2CEH
Wallingford, UK h.s.ashton_at_reading.ac.uk
1. Motivations
4. Land Surface Modelling
Data from the Netherlands Loobos FLUXNET
observation site has been used to drive a
second-generation land surface scheme derived
from MOSES 2.1 of the UK Met Office Unified Model
The Joint UK Land Surface Simulator
(JULES). Land surface models are one of the
major causes of uncertainty in climate change
predictions (Gedney et al 2000). Therefore it was
important to compare output from JULES output
against observations from the site it represents,
in order to verify that it can simulate water
stressed conditions. JULES was run with a
standard set up, driven by meteorological data
for Loobos 1997 - 2003. This output was compared
with ground observations taken from Loobos during
the same years. A sensitivity analysis has been
carried to investigate the effect of changing the
soil hydraulic parameters (critical soil
moisture,?c and soil moisture availability
factor,?), the surface conductance (gs) and the
vegetation type (Leaf Area Index, LAI and Plant
Functional type fractions). The satellite
derived land surface temperatures from
MODIS/TERRA have been compared against modelled
land surface temperature over a range of soil
moistures. This will be used to improve the
modelled land surface temperature data at the
point scale.
The summer of 2003 was an extreme climatic
anomaly. In a large area of the European
continent, the mean summer temperatures (JJA)
exceeded the 1961-90 mean by 3C (figure 1).
Schar et al (2004) showed from historical
observations, that this event had a low
probability of returning.
Figure 1 Characteristics of the summer 2003
heatwave - JJA temperature anomaly with respect
to the 1961-1990 mean. Colour shading shows
temperature anomaly (ºC). Bold contours display
anomalies normalised by the 30-yr standard
deviation (Schar et al 2004)
Climate simulations have suggested that the
European summer climate might experience a
pronounced increase in year-to-year variability
in response to greenhouse-gas forcing
5. Results Model-Observations Comparison
Present Day
Future decades
The soil hydraulic parameters are used to define
how soil moisture affects evapotranspiration.
They are defined according to the soil texture (
sand, silt clay) at the location/grid box. Soil
is spatially very heterogeneous and hence soil
textures are not very well known and difficult to
estimate. JULES uses 3 texture classes (fine,
medium coarse) which is far fewer than found in
reality. The results shown in figure 6
illustrate how sensitive the fluxes are to
changes in the soil parameters. Under the default
parameter setup JULES is unable to simulate the
partition between latent heat (LE) and sensible
heat fluxes (H). The evaporative fraction
(LE/(LEH)) is overestimated for most of the
year. The consequences of this is an
underestimation of the surface temperatures. The
soil hydraulic parameters were reassessed, and a
more realistic representation of vegetation
fractions was implemented. This resulted in a
better match between modelled and observed
evaporative fraction.
Figure 2 Schematic diagrams showing the
distribution of average summer (JJA)
temperatures. 2003 stands out as the highest
temperature to date and sits out in the tail of
the frequency distribution. This event has a low
probability of return. Heatwave events such as
these may become more frequent in future decades.
Extreme temperatures are expected to increase
with an increase in the range of temperatures
hence the tail in the distribution will become
longer and the peak will lower. It is possible
that temperature extremes like those in the
summer of 2003 will become closer to average
summer temperatures. One possible cause of this
is the interactions between soil moisture and
atmospheric physics
Figure 6 Modelled and observed evaporative
fraction for (i) old parameter setup (green) (ii)
New parameter setup (blue), (iii) Flux tower
observations (red) and total soil moisture
availability factor (beta) in a 3m column (black)
2. Background
Soils have been shown to impact weather systems
via changes in surface heating. In 2003 the
precipitation deficit and prolonged hot
conditions at the surface led to the drying of
the soil. A reduction in soil moisture leads to
strongly reduced latent heat flux. This increases
the amount of sensible heat emitted from the
surface and increases surface temperatures. This
can then set up a positive soil moisture feedback
(Fischer et al 2006)
6. Results Model-Observations Comparison 2
The model observation comparison was continued
for more years worth of data to see if this
behaviour was repeated. Modelled and observed JJA
evaporative fractions from 7 years of data were
plotted against total modelled soil moisture
(figures 7 8). At low soil moistures (lt900mm)
modelled evaporative fraction is overestimated
compared to observations. In contrast at higher
soil moistures (gt900mm) modelled evaporative
fraction is underestimated compared to
observations. We are currently exploring whether
this behaviour can be observed in the modelled
and satellite surface temperatures. This will
give an idea of whether satellite data can be
used to improve the specification of soil
parameters and hence regional scale simulations.
Figure 3 The soil moisture feedback process
This soil moisture feedback is believed to have
contributed to the anomously high temperatures
experienced in the summer of 2003. Soil moisture
and surface fluxes are very heterogeneous over
the land surface. Observations of these are
sparse and are only representative on a local
scale. In order to understand how soil moisture
influences summer climate, one needs to correctly
simulate the soil moisture and surface fluxes
during summer time conditions using a land
surface model at a regional scale. How well can
land surface models can simulate these feedback
processes?
Figure 7 Plot showing the relationship between
observed/modelled evaporative fractions and
modelled soil moisture JJA 1997 - 2003.
3. Use of Earth Observation
Figure 8 Modelled and observed evaporative
fractions averaged over soil moisture intervals
Figure 5 Differences (2007 minus 2003) in
MODIS/TERRA 8day averaged FPAR over the period
28th July - 4th August 2003/07
Figure 4 Differences (2007 minus 2003) in
MODIS/TERRA land surface temperatures daily
averaged over 31st July - 4th August 2003/07
7. Conclusions Further Work
The modelled partitioning between latent and
sensible heat fluxes matches observations very
well. It has been shown that fluxes are very
sensitive to the specification of soil and
vegetation parameters. However it can be seen
that during dry spells modelled evaporative
fraction is still slightly overestimated compared
to the observations. Similarly, during wet spells
the modelled evaporative fraction is slightly
underestimated compared to observations. This has
been observed over multiple years. It has been
shown that MODIS/TERRA land surface temperatures
and FPAR are sensitive to changes in soil
moisture. Hence there are plans to look at
modelled and satellite derived land surface
temperatures to see if the same behaviour can be
observed. It can then be decided as to whether
this data can be used on a regional scale to
improve soil hydraulic parameter maps. Now need
to answer the question of whether satellite data
is useful to constrain the model at a regional
scale. Next step Need to look at another
site where the drying of soil is more pronounced.
Information about the surface on the large scale
can be detected from satellites. Land surface
temperatures and FPAR (Fraction of
Photosynthetically Active Radiation) retrieved
from MODIS/TERRA can provide information on the
moisture status at the surface. Dry soils have
higher surface temperatures and inhibit
vegetation growth, leading to a lowering of FPAR.
Data from 2003 and 2007 show how a variation in
surface moisture status can be detected by earth
observation. High pressure ridges and clear sky
conditions were frequent during the period from
28th July to 4th August 2003 and 2007. However
soil moisture conditions were contrasting. The
summer of 2003 was an extremely dry year in terms
of soil moisture (Fischer et al 2006) whereas in
contrast 2007 was very wet over the UK and France
with significant flooding occurring in the week
before this period. 2007 has also been noted for
its severe heatwaves over the Mediterranean. The
differences in soil moisture between 2003 and
2007 have lead to a contrast in surface
temperatures within the UK, France and Germany
showing a difference of up to 10C. A weaker
signal is seen in the FPAR with small increases
of up to 5 during 2007, resulting from the
increase in available soil moisture.
8. References
E. Fischer et al (2006). Soil Moisture-atmosphere
interactions during the 2003 European heatwave.
Journal of Climate (in press) N. Gedney et al
(2000). Characterising GCM land surface schemes
to understand their responses to climate change.
Journal of Climate. C. Schar et al (2004) The
role of increasing temperature variability in
European summer heatwaves. Nature 427, 332 - 336.
Acknowledgment Heather Ashton is funded by
CLASSIC