Despite the large quantity of research undertaken into the sustainability of
food production and transportation systems, there is currently little consensus on
the total contribution that greenhouse gas (GHG) emissions make to the overall
GHG budget of food production systems. To date, most research has focused on
the miles that food has travelled and the energy put into the production of
pesticides and fertilisers associated with crop production. Understanding whether
food imported from distant countries has a higher GHG footprint than locally
produced food remains a very topical issue. Our fundamental lack of knowledge
of this issue is limiting policy development in this area. Due to difficulties in
field measurements mathematical models such as DNDC (DeNitrification
DeComposititon) are being used to predict GHG emissions from different
ecosystems. In this thesis, a combination of field measurements and model
simulations were used to evaluate GHG emissions from different agricultural
production systems undertaken in different countries (UK, Spain and Kenya).
This thesis also considered the accuracy of the model by undertaking a sensitivity
analysis and evaluating the outputs from different model versions. In addition,
the accuracy of using a Q10 value approach to predict organic matter degradation was also evaluated.
Overall, the results suggested that different model versions gave varying
outputs, suggesting that predictions of GHG emissions obtained with models
such as DNDC should be treated with caution. However, the model did predict
similar results to those obtained in the field, although the model outputs tended to be higher. For comparison of GHG emissions from vegetable types grown in
different geographical regions, no specific region produced lower GHG results
when averaged across all crops. However, when individual crops were
considered, Spain had the highest GHG emissions. The models showed different
degrees of sensitivity to different inputs, with some not showing any variation at
all. In the Q10 evaluation experiments the Q10 values varied greatly, though all
gave results above the standard Q10 of 2. Further research is needed into the
accuracy of climate and farm management models, and whether or not it is
necessary to compare large data sets when considering different vegetable types
and areas.