This document explores the details of the data that describe cattle population in Brazil with the purposes of calculating both water use and GhG emissions from the cattle herd in (at least) 2017. The datasets used below are the following:
We rely on work by Erasmus in the following script: https://github.com/sei-international/TRASE/blob/6227bf8de8135c8c31250e93eb944dccd8d0df75/trase/data/brazil/beef/production/cattle_production_2010_2019.Rmd which calculated cattle production.
Some important conclusions from his work:
We can use the cattle production data that he generated (tonnes of carcass per municipality and per year) for the purposes of linking the indicators work to the beef produced and exported.
Here we are really focused on generating the indicator at the municipality level for both water use and GhG emissions.
We explore three possible estimates of cattle population at municipality level following:
For the calculation of liveweights, Erasmus used some information from Schielein and Börner (2018) (see SEI-PCS documentation), but we can now update this information with the new regional liveweights available by MCTI.
The classifications of animal stages are different based on input datasets:
For GhG emissions, there are sub-category emissions factors that are specific to age groups and sex. Using different population breakdowns will help identify possible windows of emissions and water use.
According to Embrapa Link the categories for slaughter are as follows:
According to the Censo Agropecuario (2006), we have the following categories:
First we look at the estimate using the IBGE Census 2006 information applied to the animal population (PPM). The result is the estimated calf and cattle population in our years of interest at the municipal level in the different phases.
We follow the MCTI (2020) approach which does annual accounting of age groups:
There are municipalities that did not have any information on population in 2006, but do have a small population in the years of interest. In that case, we take the state-wide percentages available from the census to derive the cattle sub-categories.
This first table provides the state average population breakdown, but the values are available at the municipality level. When the values do not make sense, we will use these state averages instead.
We now apply the above method to derive the population in each municipality and state between 2010 and 2019:
There are some municipalities/years that seem to have a prevalence of dairy cows which messed up the totals, but these are relatively small (12 municipalities in various years). Some of those were removed at the totals did not make sense when applying the state-wide average percent population breakdown.
The code provides two tables:
Key assumptions with this approach:
In this approach, we are provided the state average of the age groups to which we add the average breakdown of females >= 2 years (non dairy cows) determined above. This will help provide an additional estimate of cattle population, but using more recent data (also more aggregated).
These results would correct the general population but there would still be a assumption on the breakdown of adults into females for slaughter.
The code provides 1 tables:
Key assumptions with this approach:
In this approach we use the slaughtered animal population to derive the population that is slaughtered in respective age groups.
We use the SIGSIF information to estimate the cattle development phases based on a cattle balance. We combine this information to PPM to split the cattle population by IBGE (from PPM) into three stages:
\[Hm,y = Cm,y + Mm,y + Fm,y\]
where Hm,y is the total herd in a given municipality m and year y, Cm,y is the calf population derived above (as % of cattle herd population in a given year), Mm,y is the cattle population in the mid-stage (out to pasture, 1-2 years of age), and Fm,y is the total cattle population in the finishing stage (male and female 2 years of age and older).
In a first step we obtain the slaughtered cattle population in year y and equate it to the population in the finishing stage in year y.
We use the corrected Anualpec slaughter data: (1) correction on animal slaughter in years prior to 2011 using a correction IBGE/Anualpec), (2) use SIGSIF to correct for origin of animal slaughter and provide a revised state slaughter rate.
First, we use Erasmus’ method to correct Anualpec data prior to 2011 (if needed for the study, otherwise use Anualpec as is):
Then, we correct for cattle movement across states and provide a revised slaughter rate (per state) to allow for a better estimate of the per state offtake rate so that can help derive the animal population in the “finishing stage”.
We can now estimate the number of animals in their stages using the SIGSIF.
Then we use the Anualpec (2017) relative population of calf to separate the calfs from the remaining population and assign the rest of the population to the mid-level (1-2 years of age) cattle group.
The code provides 1 tables:
Key assumptions with this approach:
Compare the results at the state level (then move down to include the third estimate)
Then generate tables for deeper dive at the state level.
MCTI (2020) method without dairy cows
MCTI (2020) method with dairy cows
Anualpec approach*
Lathuilliere et al (2019) approach
Let’s now compare the estimates per municipality in a scatter plot.
Now let’s get the total population according to the methods
## Joining with `by = join_by(YEAR)`
## Joining with `by = join_by(YEAR)`
## Joining with `by = join_by(YEAR)`
## # A tibble: 10 × 5
## YEAR TOTAL_BEEF_CENSUS TOTAL_BEEF_DAIRY_CENSUS TOTAL_BEEF_DAIRY_ANUALPEC
## <dbl> <dbl> <dbl> <dbl>
## 1 2010 186393858 209298172 209541097
## 2 2011 189362466 212570930 212815289
## 3 2012 188250011 211032947 211279126
## 4 2013 188580423 211514208 211764364
## 5 2014 189107064 212114865 212366113
## 6 2015 193868629 214960695 215220441
## 7 2016 198383961 217925318 218190739
## 8 2017 197902083 214738952 215003430
## 9 2018 197211437 213549279 213809407
## 10 2019 198101975 214391524 214659743
## # ℹ 1 more variable: TOTAL_BEEF_SIGSIF <dbl>
We then export the data with these important notes regarding column definitions:
cattle_population_beef_census.csv
cattle_population_beef_dairy_census
Column names have same meaning, except that the dairy cows are included (DAIRY added to the names)
cattle_population_beef_dairy_anualpec
Column names have same meaning, except that the dairy cows are included and the Anualpec method is used (ANUALPEC added to the names)
cattle_population_beef_sigsif These results are no longer considered as they deviate too much from other results and are not in line with “official” approach (MCTI)
## [1] "saved at C:/Users/MICHAE~1/AppData/Local/Temp/RtmpUrn7me/file5aa878ad73b1/cattle_population_beef_census.csv"
## [1] "saved at C:/Users/MICHAE~1/AppData/Local/Temp/RtmpUrn7me/file5aa878ad73b1/cattle_population_beef_dairy_census.csv"
## [1] "saved at C:/Users/MICHAE~1/AppData/Local/Temp/RtmpUrn7me/file5aa878ad73b1/cattle_population_beef_dairy_anualpec.csv"
Lathuillière et al (2019) Cattle production in Southern Amazonia: implications for land and water management Environ. Res. Lett. 14(11): 114025, doi: 10.1088/1748-9326/ab30a7.
MCTI (2020) Quarta Comunicação Nacional E Relatórios De Atualização Bienal Do Brasil À Convenção-Quadro Das Nações Unidas Sobre Mudança Do Clima. Quarto Inventário de Emissões de Remoções Antrópicas de Gases de Efeito Estufa Ministério da Ciência, Tecnologia e Inovações: Brasília, DF. Link https://www.gov.br/mcti/pt-br/acompanhe-o-mcti/sirene/publicacoes/relatorios-de-referencia-setorial