Trase H2O First Pass

Summary

This document provides initial results for the TraseH2O project by following a “first pass” of expected results of Objective 1. Objective 1 of the project involves the following steps:

1. Obtain water use for crops for Brazilian municipalities
2. Obtain water use for cattle for Brazilian municipalities
3. Link water use for soybean and cattle per municipality with trase data
4. Perform a benchmark analysis considering water use, land use, deforestation and GhG emissions from deforestation

This “first pass” skips step 1 and 2 by using datasets from Flach et al (2020) and Lathuillière et al (2019) for Mato Grosso in order to get an early idea of the kind of analyses that could be done once the water use data is widely available for all of Brazil’s municipalities.

The document also includes important notes to consider for the project so that all the calculations can be reviewed.

Soybean water use and supply chain

In this first pass, we use the data from Flach et al (2020) to obtain ET from soy production in the municipalities of Mato Grosso. Flach et al (2020) use the EPIC model to determine the growing season ET (GSET, as actual ET) for soybean and maize in each Brazilian municipality (1990-2013). This actual ET already takes into account the water use from rainfall, but does not include any irrigation.

The ET results are effectively simulations of ET (un-validated) for each of the areas of soybean and maize in the Brazilian municipalities. The ET methods used is the Hargreaves method within EPIC. Simulations were run on 80,000 units in the country using 3 scenarios:

1. soybean in a single cropping system
2. maize in a single cropping system
3. soybean and maize in a double cropping system

Flach et al (2020) also did a sensitivity analysis on planting dates, and crop development cycle (harvest dates) to provide a range of ET for all possible situations in Brazil. Planting dates also observed the onset of the precipitation using anomalous accumulation of precipitation from Arvor et al (2014). The combination of these conditions provide a robust window of ET results for the crops of interest.

Meteorological data came from Xavier et al (2016) which is a gridded dataset derived from INMET data (1980-2013) and provide a limitation to the ET time series. Land use data was obtained from Dias et al (2016).

In the second pass, we will need to considering and research the following:

• revisit the availability of meteorological data to obtain simulations of crop ET beyond 2013
• explore the potential combination of ET obtained from remote sensing and land use maps
• check the possible use of land use maps from GLAD on soy extent in South America (unpublished) to derive ET of the growing season(s)
• obtain a full map of irrigation for soy in Brazil for the time series of interest
• expand on the separation of green and blue water for production per year (including irrigation potential)
• explore the uncertainty in the results and the allocation of the growing season with exports

Cattle water use and supply chain

In this first pass, we use the results from Lathuillière et al (2019) to estimate water consumption for beef in the municipalities of Mato Grosso. Lathuillière et al (2019) estimate water consumption by simulating the amount of water integrated into cattle or evaporated from the production system.

Water integrated into cattle include:

• moisture content in feed
• water drunk by the animal
• water generated by the animal’s digestive system

The results are provided per kg live weight (kg LW) and change according to the animal’s development cycle. Several scenarios were carried out to determine the animal’s water consumption: male vs. female development cycle, and pasture vs. feedlot finishing. For the purposes of this first pass, we can use the average water consumption of the animal as: 143 L per kg LW.

the pond area per municipality was estimated using remote sensing in order also estimate evaporation from ponds as a water consumptive use (estimated to be 126 L per kg LW in Mato Grosso as a whole). We expect this value to change per municipality based on the herd size, pond area and evaporation. The pond area estimate is available for the years 2000, 2005, 2010 and 2014 which does not match the time series available in trase at the moment (2015-2017). We therefore use 2014 as the most recent estimate for cattle water use and trade.

In the second pass, we will need to considering and research the following:

• Further development of the Brazilian beef trase data for the expanded time series (2010-2019). This is currently ongoing with work from Erasmus
• Choose the right time series for detecting cattle ponds in Brazil to match the ET0 data available from Xavier et al (2016) and the trase data available. This will likely be 2010-2013 data overall
• Think carefully about allocating water use for cattle to exports (similar to deforestation)
• Consider augmenting the CO2 emissions estimate on trase with CH4 emissions from cattle (to provide total CO2 emissions from cattle)
• Explore tha allocation of farm ponds to animal population with the GTAs
• Review calculations of pasture ET and the link to feed for cattle
• obtain better assessments of confined vs. unconfined cattle in Brazil

Benchmark analysis

The benchmark analysis is part of the first deliverable of objective 1 and aims to take a closer look at the changes in environmental performance of soybean and cattle sourced from municipalities. The analysis uses data from trase (land use, deforestation, GhG emissions) and add water use to check for any potential changes in performance (e.g. a municipality with little deforestation, might requite large amounts of water for production).

We can carry out a benchmark analysis on production (all Brazil), and then on consumption (focusing specifically on the municipalities linked to the consumer markets). The latter would be equivalent to some LCI analysis.

Important notes:

• This first analysis is done for soybean produced in all of Brazil, but for cattle produced only in Mato Grosso
• Trase currently has domestic market information for soybean, bust not for cattle

References

Arvor et al (2014) Spatial patterns of rainfall regimes related to levels of double cropping agriculture systems in Mato Grosso (Brazil). International Journal of Climatology 34(8): 2622-2633, doi: 10.1002/joc.3863.

Dias et al (2016) Patterns of land use, extensification, and intensification of Brazilian agriculture. Global Change Biology 22(8): 2887-2903, doi: 10.1111/gcb.13314

Flach et al (2018) The effects of cropping intensity and cropland expansion of Brazilian soybean production on green water flows. Environmental Research Communications 29(27): 071001, doi: 10.1088/2515-7620/ab9d04.

Lathuillière et al (2019) Cattle production in Southern Amazonia: implications for land and water management. Environmental Research Letters 14(11), 114025, doi: 10.1088/1748-9326/ab30a7.

Xavier et al (2016) Daily gridded meteorological variables in Brazil (1980-2013). International Journal of Climatology 36(6): 2644-2659, doi: 10.1002/joc.4518.