Intra- and inter-year variability of agricultural carbon footprints – A case study on field-grown tomatoes

Abstract

The performance of agricultural systems and their environmental impacts can vary considerably within a single crop supply chain, due to differences in farming practices, soil properties, and yearly climatic conditions. In this paper, we characterised the variability of carbon footprints of open-field tomato production by analysing a comprehensive farm dataset gathered over 4 years. We also assessed the importance of the different drivers of variability as compared to model uncertainties. The primary data used in this study were collected from 189 farms from the Extremadura region in Spain and Portugal over a period of four years, from 2012 to 2015. We modelled the carbon footprint of these farms using the Cool Farm Tool model developed by Hillier et al. (2011) and conducted statistical analysis on the results to understand the relative importance of inter-year and intra-year variability. We performed sensitivity analysis to understand how sensitive the results were to variability in the farmers' input parameters and to the uncertainty in model parameters. This was done by varying all factors one-at-a-time, and then by running a Monte Carlo simulation where all uncertainties were propagated simultaneously. Results clearly show significant inter-year and intra-year variability in carbon footprints of tomato production within the study region. We observed approximately 20% variation for each annual carbon footprint (intra-year variability), resulting in an overall 28% coefficient of variation in the aggregated footprint across the different years. The carbon footprint of the whole tomato supply, calculated with the 4-year dataset, showed a weighted geometric mean of 51 kg CO2-eq/t and a weighted GSD of 1.32, meaning a 95% confidence interval of 29–89 kg CO2-eq/t. Results also show that small farms were characterised by a larger variability than larger ones. This highlights the need to weight farm results by production volumes if the objective is to obtain a carbon footprint for the total production in a given region. The carbon footprint was found to be mainly sensitive to variability in farm practices, notably extent of pump irrigation and choice and amount of fertiliser used, with model uncertainties influencing the results to a relatively smaller extent. Further work is needed to extend these findings to other crops, regions and impact categories.