TraseH2O - Objective 1 - Consumption indicator

Introduction: Soybean consumption

This document looks at the combined indicators of consumption related to soy exports from Brazil. Information on production is linked to the supply chains of Trase (Brazil soy and beef) to look at actor performance (China, EU and Brazil at this time).

We look at 2015-2017 trade.

Importing country average

We first look at the benchmark of Brazil, China and the EU linking specific Brazilian municipality of export with the average soy WF and the composition of green and blue WF (in m³ per tonne). We basically calculate a Commodity Supply Mix.

Recall that the trase data that is imported here has the cattle deforestation 5y total exposure already embedded. Given that it is a 5y sum, you cannot sum across the years of interest and can only show individual years.

First, we focus on China and EU imports, as well as Brazilian domestic consumption (which might not be very relevant).

We need to keep in mind that there are unknown flows (which need to be reported as %total) for which we can:

not present a benchmark (should be a small percentage)
provide a country-wide benchmark (probably not worth it)

The volume of unknown flows are shown below:

## # A tibble: 3 × 4
##   year       vol    vol_tot pct_vol
##   <chr>    <dbl>      <dbl>   <dbl>
## 1 2015  6334513.  97464936.    0.06
## 2 2016  7090174.  96394820.    0.07
## 3 2017  8350842. 114732101.    0.07

So the total amount of unknowns represent 6-7% of total volume of soy in 2015-2017. We then look at the soybean WF benchmarks and derive an estimate of production that is susceptible to be irrigated. In more detail:

We then look at the total water, deforestation, GhG emissions associated with consumption (imports into China and the EU and Brazil’s domestic consumption) and calculate average footprints (Commodity supply mix).

Some notes regarding reporting irrigation values:

For any municipality that has irrigation, the soybean that is exported from this municipality is susceptible to be irrigated at the level of blue water calculated in the WF (m³ t^-1)
This does not mean that the entire production from that municipality is irrigated; similar to deforestation we are only talking about association to irrigation in the supply chain.

The above represents the total amount of soybean that is potential irrigated based on the municipalities for which irrigation was detected.

Additional thoughts

Taking the blue WF for that municipality and multiplying it by the volume exported, we get the potential blue water consumed for imports in Gm³. But, what does 0.36 Gm³ of blue water allocated to Chinese imports of soy actually mean?

If a municipality has 3 m³ t^-1 as a blue WF for soy, this means that for that municipality, the distributed blue WF for the given amount of production, considering the area irrigated, led to 3 m³ per tonne of soy produced in that municipality. So for each tonne of soy produced, there is a share of irrigation responsibility. This share is then allocated to the volume of soy exported to a given country, leading to 0.36 Gm³ of potentially used irrigation for soy allocated to Chinese imports. However, this still amounts to a given volume, and not an impact (similar to deforestation might represent).

Big numbers assigned to importing countries

Let’s dig into the WF assigned to consumption centers.

The above shows differences in soybean WF as they are associated with the export markets (of all soybean, rainfed and irrigated combined). We can provide another estimate that is focused only on blue water as as sort of risk to water scarcity.

Linking trade to production benchmarks

We revisit the production benchmarks and then check which municipalities are supplying each consumer country, and the associated metrics.

For now we focus exclusively on the top deforestation and top water scarcity footprint (and overlap)

Then, we need to relate the to the EU and China focused benchmarks which we calculate here.

We then connect to the “flags” that we assigned to soy production which included whether soy:

Was linked to a municipality above or below the 80th percentile cutoff (a municipality with no soy deforestation is included in the “< cutoff”)
the level of total water footprint (80%tile or not)
whether soy was irrigated or not

We place China and EU imports within these categories and also determine the volume of soy that would fall within each of the categories.

Important This approach links the benchmarks of production to the supply chains, which is not the same as deriving the benchmarks of consumption.

Each source of soy can be linked to a different strategy whether the impact is concentrated in deforestation of water scarcity. There are ways of addressing multiple impacts.

Again, the above is about looking at all of production, rather than all the soy imported. The actions are different and focus either on production, or on the supply chain.

Total amount of soy that is “exposed” to irrigation, i.e. that the soy supply is being sourced by municipalities that apply irrigation:

China: 12.3 Mt (33%) (2015), 13.1 Mt (34%) (2016), 12.8 Mt (27%) (2017)
EU: 4.1 Mt (30%) (2015), 5.1 Mt (40%) (2016), 3.3 Mt (26%) (2017)

Largest volumes of potentially irrigated soy going to China and the EU were in the bottom 80% of deforestation and WF and represented quite a large potentially affected volume:

China: 6.6 Mt (18%) (2015), 9.0 Mt (26%) (2016) and 8.7 Mt (18%) (2017) of imports
EU: 1.1 Mt (8%) (2015), 2.5 Mt (19%) (2016) and 1.2 Mt (10%) (2017) of imports

What could be considered the worst case scenario, high deforestation, high WF and irrigation:

China: 0.4 Mt (1%) (2015), 1.4 Mt (4%) (2016), and 0.4 Mt (0.9%) (2017) of imports
EU: 0.2 Mt (1.5%) (2015), 0.8 Mt (6%) (2016), and 0.3 Mt (2%) (2017) of imports

We could also show this in a bar graph, perhaps easiest to draw out (to be explored).

We then plot the water scarcity footprint benchmark of imports into China and the EU and the maps showing the largest water scarcity footprints, which is different from the tables above that show the overlap of deforestation, WF and irrigation.

The above curve can be used to highlight specific municipalities and volumes where soy is being sourced. The idea is to superimpose the volumes above benchmark with the production benchmark and then show the following:

Municipalities where the soy with the highest WSF is being sourced from by China and the EU within the context of all the WSF of soy production
provide information on deforestation that is also associated with this production and highlight the river basins where impact is highest

Below is an example of the sourcing for China and the EU in 2017.

The above graphs gives the positioning and size of imports from the municipalities with the largest impacts to water scarcity. We can now show this spatially.

We can also calculate the volume of soy that falls within the largest water scarcity footprint, regardless of deforestation.

## # A tibble: 6 × 4
##   year  economic_bloc  vol_sum def_exp_ha
##   <chr> <chr>            <dbl>      <dbl>
## 1 2015  CHINA         2930674.     42018.
## 2 2015  EU             819906.     18318.
## 3 2016  CHINA         2696901.     19197.
## 4 2016  EU            1057828.     15220.
## 5 2017  CHINA         2610040.     33904.
## 6 2017  EU             760548.     16878.

The following plots show the 80th percentile WSF of all of soybean production that is irrigated, but then highlights the specific municipalities that are exporting to either China or the EU (highlighted in blue outline).

Plot WSF for EU in 2015:

Plot WSF for EU in 2017:

Calculating water savings

Then we can calculate the potential water savings based on application of a median beef WF in cases where the WF is within the cutoff (now 80th percentile). So we carry out the following:

Identify which municipalities sourcing soy to CHINA and the EU are in the “> cutoff WF” category
Replace the WF by the median
Calculate the total volume of water associated with trade and compare to actual in those municipalities
Compare the water savings to the total virtual water (keeping in mind that only a portion is “saved”)

## # A tibble: 6 × 6
##   economic_bloc tot_km3_from_median_wf tot_km3_ref median_wf km3_savings year 
##   <chr>                          <dbl>       <dbl>     <dbl>       <dbl> <chr>
## 1 CHINA                          11.3        13.1       1928       1.77  2015 
## 2 EU                              4.52        5.29      1928       0.770 2015 
## 3 CHINA                          11.3        15         1918       3.66  2016 
## 4 EU                              6.04        7.92      1918       1.88  2016 
## 5 CHINA                          14.5        16.6       1803       2.18  2017 
## 6 EU                              5.78        6.64      1803       0.860 2017

Sankey representation (figure 3)

We then create a Sankey to summarise all of resource use (i.e. total water use, total deforestation, etc.).

Statistical analyses

We can now run some statistics on the numbers, with potentially two tests to carry out:

Compare soy imports into China and the EU with and without deforestation, regardless of where there is irrigation: this test will include values of 0 in both areas, so comparing the overall soy production
Compare soy imports into China and the EU with and without deforestation, considering only the irrigated areas: this will remove the municipalities with 0 irrigation and will exclusively compare irrigated areas.

Note that the code should be reviewed to confirm the results. At this stage we will not include this analysis in the paper.

Considering all soy imported, regardless of irrigation**

Here we check the hypothesis:

The soy imports into China and the EU have the same WF and WSF whether or not the soy is linked to deforestation

Summary of results:

CHINA: 2017 - all signif. differences, except WF results at ha_limit == 0 2016 - all signif. differences in WF, but not in WSF 2015 - all signif. differences in WF, but not WSF
EU: 2017 - all signif. differences, except WF results at ha_limit == 0 2016 - WF show signif. differences only 2015 - WF show signif. differences and WSF at ha_limit == 0 only

So 2017 seems to be an odd year, perhaps due to El Niño?

Considering soy production only in municipalities that have irrigation