17 Apr 2018

Behind the scenes

Improve agricultural data quality: multi cropping in Agri-footprint

We believe that transparency is essential in the development of sustainable agri-food chains, which is why we are more than happy to give you an impression of how we work ‘behind the scenes’ and share some of our insights and experiences with you.

Recently, Hans Blonk visited Brazil for a Global Feed LCA Institute (GFLI) stakeholder meeting. Improving the data on multi cropping will be one of the main topics of the upcoming GFLI data project in Brazil. Therefore, this time our colleague Paulina Gual would like to share some of our insights on the modelling of multi crop cultivation in Agri-footprint, the LCA Food Database.

Making life cycle modelling more accurately reflect agricultural reality is a major challenge in agricultural life cycle assessment (LCA). Yearly land occupation (ha/year) is particularly difficult to model, especially when multiple crop cultivation is involved. Agri-footprint process modelling gives a clear and easy to follow structure to cultivation processes using 1 ha as a base for calculating the total production (kg) for a specific crop/country combination using FAOstat yield data (kg/ha). Our methodology (Agri-footprint 4.0, the last public update) implicitly assumes a crop cycle is always exactly one year. This approach may be inadequate when considering multiple cropping cycles (same crop) or associated crop cycles (multiple crops) on the same parcel within a year.

Acknowledging the broadness of this modelling assumption and always seeking to improve the quality of our database, we set out to assess the limitations and possible implications of our method when representing land occupation. We assessed how well our model can reflect reality by estimating the possible scenarios for crop cultivation and identifying how well they were represented in Agri-footprint. It should be noted that these scenarios do not necessarily reflect actual cultivation practices in the countries covered by Agri-footprint; they just serve as means to assess the flexibility of our database to represent any possible situation. 

Scenario 1: One-year cycle or perennial cops

In one year, Crop A is grown in a specific plot during the whole year, or specific period, without recurrence.

  • AFP 3.0 modelling correctly adjusts to crops in this scenario.
  • Model does not include/consider fallow land or failed crops (FAOstat yield data account only for area harvested (FAO, 2017))

Scenario 2: Multiple crop cycles for the same crop in one year

Our second scenario deals with cases in which Crop A is grown and harvested in area L1 multiple times in one year. The yield definition in FAOstat clearly states that the land area is counted as many times as it is harvested (FAO, 2017).

In this scenario, FAOstat considers the total harvested area to be 3L1, but in reality only L1 is used in a year. 

  • AFP modelling overestimates yearly land occupation.
  • Model does not include/consider fallow land or failed crops (FAOstat yield data account only for area harvested (FAO, 2017)).

Scenario 3: Crop rotation, multiple crop cultivation on same land parcel within one year

In one year, Crop A and B are grown and harvested consecutively on L1. 

By using FAOstat yields, we consider L1 fully occupied by A and again by B. To emulate reality, L1 should be partially assigned to A and B. 

  • AFP modelling overestimates yearly land occupation.
  • Model does not include/consider fallow land or failed crops (FAOstat yield data account only for area harvested (FAO, 2017)).
  • L1 should be allocated to A and B.

Scenario 4: Crop rotation, multiple crop cultivation on the same land parcel during a period of more than one year

In one year, Crop A and B grown and harvested consecutively in L1. 

Again, L1 is considered 3 times, but the real land occupation during one year is L1. In this scenario, Crop B is grown between two harvest years. FAOstat (FAO, 2017) sets constraints on reporting yearly crop production and harvested area. Countries must report a specific crop in the year when the bulk of the harvest occurs. This simplifies considerations of partial year crops in AFP.

  • AFP modelling overestimates yearly land occupation.
  • Model does not include/consider fallow land or failed crops (FAOstat yield data account only for area harvested (FAO, 2017)).
  • L1 should be allocated to A and B.
  • AFP uses 5-year average crop yields, ensuring that crops cultivated between two harvest years are correctly considered if reported as per FAOstat guidelines.

What does this mean?

The Agri-footprint methodology adequately represents perennial crops, one-year cycle and mixed crops (as the harvest land for each is reported separately to FAO), with the disadvantage that using FAOstat yield data means only the harvested area is considered but not any failed crops or fallow land. Where crops are cultivated and harvested on the same plot of land, the Agri-footprint approach overestimates the area of land used because FAO counts the same area of land each time it is harvested.

To represent reality more accurately, we would need to acquire data on common crop cycles and associated crop combinations for each country covered by Agri-footprint. Not only this, a partition system would need to be defined to correctly allocate the land use to all the crops cultivated and harvested on the same plot of land during the course of a year. We would then either require information about the time the land was used (including resting periods) for each crop or use an allocation approach already in our database (economic, energy, mass).
The additional data required would be country specific and accessibility might vary from country to country in scope, methodology and quality, with some countries having a lot of information and others none at all. Differences in data availability and country specificity represent a challenge for inclusion in a life cycle inventory background database with a wide scope such as Agri-footprint.

The main challenge lies in the wide range of data sources that would be required to fully cover all crop/country combinations in Agri-footprint. Right now, to model land use we rely on one source, FAOstat, with the same scope, format and accuracy. Having to retrieve data for different countries, crops combination schemes and common crop cycles would require extensive data manipulation from sources with different levels of quality.

An initial way forward could be to focus on scenario 2 crops (multiple cycles for the same crop), which might be the easiest to model as no partition method would be required and it would be easier to investigate common crop cycle practices for a single crop. 

What we have learnt

  • For scenarios of multiple crop cultivation in a year, Agri-footprint modelling currently takes a conservative approach that ensures land use is accounted for every crop, with a clear and easy to follow process structure.
  • To be able to represent reality in multi-crop scenarios, we would have to know the crop cycles and commonly associated crops used in each country. Moreover, a partition approach would need to be devised to allocate land use between crops grown on the same parcel.
  • Agri-footprint aims to have a wide coverage, with a clear process structure, and needs country/crop specific data to represent multi-crop systems, making it difficult to expand into a full background database such as Agri-footprint.

Way forward

We are committed to continually improving the Agri-footprint database. Besides this evaluation we have also carried out many other self-assessment exercises and in the light of the results we have decided that the additional work needed to represent different land use scenarios for crop cultivation would involve a complete overhaul of our dataset, which is not feasible in the short term. For now, we will focus on the correct representation of scenario 1 and work out ways to include scenario 2 in future AFP updates.

We did not discover to what extent counting multiple harvest cycles would ‘inflate’ the harvest areas reported in FAOstat. We understand that our model overestimates reality, but need to further investigate exactly how many country/crops are impacted by this.

It is also important to highlight that the focus of this screening was to consider the land occupation implications of the current AFP modelling approach. However, the implications of double cropping affect different aspects of cultivation practices (fertiliser use, crop residue), which present a constant challenge to LCA methodologies that aim to accurately portray reality. This however, was not examined in this analysis. 

More information

Get in touch

Paulina Gual
Senior Consultant | Blonk Consultants

We are always looking for ways to improve our background data. Any suggestions or new information is welcome. Also, if you have any questions about double cropping in Agri-footprint, please get in touch with Paulina Gual