Research Topics
Research
by the Crops and Climate Group falls within three broad topics:
1.
Development of a combined crop and climate forecasting system
Food
production systems are inherently sensitive to variability in weather
and climate with critical thresholds above which crop productivity can
be seriously affected.
Recently, we have developed a new combined weather and crop
forecasting system, based on the General Large Area Model for annual
crops (GLAM). GLAM can take its input data directly from numerical
weather prediction and climate models, thus providing a new approach to
investigating the impact of climate variability and change on crops.
2.
Coupling crop models to numerical climate models
We
have coupled the GLAM crop model to the land surface scheme of the UK
Meteorological Office Unified Model (MOSES-2). Thus, we can begin to
represent crop/atmosphere interactions in a realistic manner within
General Circulation Model runs, and to explore the ways in which changes
in land use feedback on the local climate.
3.
Assessments of the impacts of climate variability and change on crops
Assessments
of the impacts of climate change on food supply in the future will
therefore require crop models that capture the effects of climate change
and variability. The GLAM crop model and its coupling with a GCM land
surface scheme open
up a broad range of opportunities for improved assessments of the
impacts of climate variability and change on food production. We are
currently undertaking a studies in China, India, and across tropical
regions.
Further
Details
1.
Development of a combined crop and climate forecasting system
Much
research to date has provided estimates of the impacts of climate
variability and change on food systems. This has demonstrated the
importance of changes in seasonal mean weather, and of short-term
extremes such as drought and high temperatures, on the productivity of
crops. A common approach has been to simulate the impacts of
human-induced climate change on crop productivity using crop simulation
models driven by weather data downscaled from General Circulation Models
(GCMs).
An important consequence of this approach is that differences in
the spatial and temporal scales of crop and climate models may introduce
uncertainties into assessments of the impacts of climate change. Most
crop models are designed to run at small spatial scales. They can
provide good simulations of crop productivity at the scale of fields,
but not necessarily for regions. However, policy decisions on the
stabilisation of greenhouse gases require regional assessments of
impacts on food systems. Thus, to provide this information, crop model
outputs have to be aggregated to a regional scale, resulting in a loss
of information. An alternative approach is to design a crop model to
operate on spatial and temporal scales close to the scale of the GCM
output (Challinor et al. 2003). By using a large area process-based crop
model, a more integrated modelling approach may reduce uncertainties in
predictions of crop productivity over regions, and so perhaps provide
better estimates of the vulnerability of crops to climate change.
The
scientific basis for a large-area crop model has been established by
looking at the relationship between crop yield and weather data on a
number of spatial scales (Challinor et al., 2003). Such a large-area
model has the advantage that using a process-based model which operates
on the spatial scale of the GCM avoids the need for downscaling of
weather data, whilst maintaining a process-based modelling approach.
Intra-seasonal variability can also be represented and the impact of
temperature threshold exceedance is simulated. Further, full integration
of the crop and climate models allows the GCM to capture feedbacks
between the crop and the climate and also diurnal temperature
variability, which is important in determining the impact of temperature
threshold exceedance.
The
General Large-Area Model for Annual Crops (GLAM) is a
process-based crop model. It has a daily time-step, allowing it to
resolve the impacts of sub--seasonal variability in weather. It has a
soil water balance with 25 layers which simulates evaporation,
transpiration and drainage. Roots grow with a constant extraction--front
velocity and a profile linearly related to Leaf Area Index (LAI). LAI
evolves using a constant maximum rate of change of LAI modified by a
soil water stress factor. Separate simulation of biomass accumulation,
by use of transpiration efficiency allows Specific Leaf Area (SLA, the
mass of leaf per unit area of leaf) to be used as an internal
consistency check: leaf area and leaf mass can be derived independently
of each other and used to calculate values of SLA which can be compared
to typical observed values. Quantitative methods to simulate and predict
the impacts of high temperature episodes have been included in the
second version of GLAM, which is called GLAM-HTS (high temperature
stress).
The
geographical focus of work to date with GLAM is the tropics. Much of the
world’s food is grown in this region. Also, there is a well-documented
dependence on rainfed agriculture across much of the tropics. Farmers
rely on monsoon rains to bring sufficient water for crop cultivation.
Preliminary work focussed on simulations in the current climate as
predictive skill here is seen as a pre-requisite for predictive skill in
future climates. The figure below shows the ability of GLAM to capture
interannual variability in yields of groundnut across India when driven
with observed weather data.
Satellite
rainfall estimates is useful in crop model for operational crop yield
forecast in data sparse regions of the Sahel where gauge observations
are sparse or unreliable. Currently decadal (10-day) rainfall estimates
from Meteosat CCD had been routinely provided by TAMSAT which has been
useful in crop monitoring for several international and national
agencies. GLAM has the potential to be deployed in operation for crop
yield forecast at national and sub-national level owing to its design
for large area applicability. Therefore, investigations have been
conducted on the possibility of using Meteosat CCD on GLAM for crop
yield prediction in the Sahel. Specific areas of interest include:
Extending the TAMSAT method from 10 day to daily rainfall estimates from
CCD data for use in GLAM at a daily timestep to better account for
intermittences in daily rainfall; investigating the effects of
uncertainties of rainfall estimates on crop model yield under an
ensemble framework using Monte Carlo simulation in generating
realisations of daily rainfall condition with CCD data.
2.
Coupling crop models to numerical climate models
In
recent decades the role of the land surface in the determination of
local weather and climate has been increasingly recognised. As
a consequence, general circulation models now include sophisticated
parameterisations of the Earth’s land surface and its interaction with
the overlying atmosphere. Despite
occupying approximately 10% of the land surface, the representation of
croplands in GCMs is almost non-existent. Instead significant cultivated
areas are commonly represented by grasslands.
However, both the seasonality and magnitude of the growth of many
crop species is different to that of natural grasslands, due to
differences in plant physiology and morphology, and the managed nature
of agro-ecosystems. Therefore,
a representation of crop growth and development has been added to
HadAM3; crop growth and development routines of the General Large Area
Model for annual crops (GLAM) were incorporated into the Met Office
Surface Exchange Scheme (MOSES). In
the new version of the model, the crop grows in accordance with the
simulated weather of HadAM3, thus altering the surface characteristics
such as leaf area and canopy height, which may in turn influence the
state of the overlying atmosphere.
The
figure above shows how the coupled crop model represents the variability
in groundnut yield within one GCM grid box. The red line is the observed
yield, the black line is the simulated yield with the grey shading
showing the spatial standard deviation.
Initially,
the groundnut version of GLAM has been incorporated as an example of the
behaviour of an annual crop in a GCM.
To satisfactorily include croplands in GCMs, the representation
of crop management needs consideration.
However, this study focuses on potential production as determined
by the climate. A potential
groundnut growing region was determined based on temperature and soil
moisture requirements. With
crops occupying this region, an AMIP integration of the coupled
crop-climate model was completed. The
simulated crop growing seasons exhibited large interannual variations in
both growing season frequency and crop size, implying that an
interactive rather than prescribed representation of crops is required
for the complete examination of crop-climate interactions.
3.
Assessments of the impacts of climate variability and change on crops
Two
aspects of climate change are important for the assessment of impacts on
crop yield. Firstly, shifts in mean climate (including changes in CO2
levels) and in sub-seasonal weather variability will impact crop
yield. GLAM already contains parameterisations to account for these
effects. Secondly, climate threshold exceedance, such as
episodes of high temperature that coincide with critical phases of the
crop cycle, can have a huge impact on crop yield. GLAM now has
parameterisations for high temperature stress.
Reanalysis
data such as ERA40 can be used to simulate crop yields under current
climates. An example of the
use of reanalysis data is shown below as correlations
between observed and simulated yields (dots indicate 95% significance).
The simulated yields were formed from an ensemble mean GLAM simulation
of crop yield in Gujarat, India. Time series of yield were formed by
driving the crop model with each individual ensemble member.
By
using these methods, we can examine the impacts of changes in mean
climate, changes in the occurrence of extremes, and temperature
threshold exceedance.
|
