1. No category

Federal Department of Economic Affairs FDEA
Agroscope Reckenholz-Tänikon Research Station ART
How to establish life cycle
inventories of agricultural
products?
Thomas Nemecek
Agroscope Reckenholz-Tänikon Research Station ART
CH-8046 Zurich, Switzerland
http://www.agroscope.ch
[email protected]
2 March 2010
Overview
 Defining system boundaries: temporal and process related
 How to get the LCI data: data survey vs. modelling
 ecoinvent database:
 Version 2.1
 Future development to version 3.0
 Direct field and farm emissions: how to estimate?
 Variability and uncertainty:




Sources of variability
Examples and implications
Analysis of variability
Assessment of uncertainty
 How to deal with missing data: generalisation and extrapolation
 Towards an integrated framework: SALCA
 Specific aspects of tropical crops
 Some recommendations
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
2
Defining system boundaries:
Temporal system boundaries
 Annual crops:
 Starting after harvest of previous crop (including fallow period
or catch crop, if no product)
 Ending with harvest of the considered crop
 Permanent crops:
 Annual basis (1st January to 31st December) or
 Multiannual cropping cycle (distinguishing different phases:
planting, young plantation, main yielding phase, eradication)
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
3
Defining system boundaries:
Example of crop production
Animal production system
Animal excrements
Products:
System boundary
Resources
Infrastructure:
•Buildings
•Machinery
Manure storage
Inputs:
•Seed
•Fertilisers (min. & org.)
•Pesticides
•Energy carriers
•Irrigation water
Field production
Catch crops
Field work processes:
•Soil cultivation
•Fertilisation
•Sowing
•Chemical plant protection
•Mechanical treatment
•Harvest
•Transport
Silage maize
Sugar beets
Fodder beets
Beetroot
Carrots
Cabbage
Grain drying
Wheat
Barley
Rye
Oats
Grain maize
CCM
Faba beans
Soya beans
Protein peas
Sunflowers
Rape seed
Potato grading
Potatoes
Product treatment:
Direct and indirect emissions
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
Co-Product:
Straw
© T. Nemecek, ART 2010
4
Defining system boundaries:
Where to draw the line between animal and
plant production?
Animal production (incl. feedstuffs,
buildings, emissions, etc.)
Manure storage
and treatment
?
Manure application
(incl. machinery use and emissions)
Nutrient use in plant production
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© Gaillard & Nemecek, 2006
5
Single crop or cropping system?
2
3
4
5
6
2
7
8
Potatoes
Year 4
1
2
3
4
5
6
7
8
Winter wheat
2
3
4
5
6
Spring
barley
7
8
9 10 11 12 1
9 10 11 12 1
2
3
4
5
Forage catch crop
5
... Fallow
Month
9 10 11 12 1
3
6
7
8
9 10 11 12
Grain maize
Fallow ...
1
... Grassclover mixture
Month
1
Green manure
Year
6
2
3
4
5
6
7
8
9 10 11 12 1
2
3
4
5
6
7
8
9 10 11 12
Grass-clover mixture ...
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© T. Nemecek, ART 2001
6
How to get representative LCI data?
Two approaches



Structural, general production and economic data are
regularly recorded in most countries (statistics, FADN, FAO,
EUROSTAT)
Data on agricultural management are largely missing
(fertiliser use, pesticides, use of machinery, timing of
interventions, etc.)
Two possible solutions:
1. Make a large survey: pilot farm networks
 one single data source
 enables to assess the variability
 preferable, but very expensive!
2. Modelling LCI: based on statistics, FADN, recommendations, expert
knowledge, etc.
 combination of several different data sources
 difficult to assess the variability
 most frequently used alternative, much cheaper
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
7
How to get representative LCI data?
1. Example of Swiss farm LCA network
Project Life Cycle Assessment – Farm Accountancy Data
Network (LCA-FADN)
 Integrate environmental LCA into FADN
 Project supported by the Swiss Federal Office for Agriculture
 Time-frame: 2004 - 2010 with data acquisition from 2006 - 2008
 Establish an operating system with 110 farms (during 3 years
with 60 in the first year)
 Establish an information technology infrastructure
 Training life cycle management principles in practice
 Develop concepts for evaluation and communication and
practice them with farmers and extension services
 Sectoral monitoring and environmental management of
farms
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
8
Farm
How to get representative LCI data? 1. Example
of farm network / Project LCA-FADN: workflow
Farm management
software
(AGRO-TECH)
AccountancySoftware
Technical data
AccountancySoftware
(AGRO-TWIN)
LCA centre
Trust and
accounting office
Accountancy data
Accountancy
Data
LCA data
FADN
evaluation
centre
Plausibility tests
SALCAprep
data extraction
SALCAcalc
LCA calculation
SALCAcheck
LCA validation and
benchmarking
Export ÖB-Stelle
FADN database
Feedback to
farmers
Existing
FADN
How to establish life cycle inventories of agricultural
products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research
Station data
ART
accountancy
Synergies
FADN  LCA-FADN
New FADN
9
Life Cycle Assessment
© Agroscope ART 2010
How to get representative LCI data?
2. Example of modelling LCI
Data category
Data source(s)
Yields for main products
FADN ART (weighted means for 1996-2003)
Straw yields and crop residues
Fertilising recommendations (Walther et al. 2001)
Moisture content
Quantity of seed
Use of machinery (number of
passes)
Gross-margin catalogue from the extension service (LBL et
al. 2000)
Sowing and harvest dates
Work budget (planning tool, Näf 1996)
Quantity of fertilisers
Fertilising recommendations (Walther et al. 2001)
Types of fertilisers in integrated
systems
Import statistics (years 1996-98 from Rossier 2000) for
mineral fertilisers
Pilot farm network (years 1994-96 from BLW et al. 1998) for
farmyard manure
Types of fertilisers in organic
systems
Pilot farm network (years 1994-96 from BLW et al. 1998) for
farmyard manure
Pesticide applications
Pilot farm network (years 1994-96 from BLW et al. 1998)
Chemical seed dressing
Information provided by seed suppliers and experts (survey)
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
Source: Nemecek, Erzinger (2004). Modelling representative
10
life cycle inventories for Swiss arable crops. Int J LCA.
Sources of LCI data:
ecoinvent database v.2.1
 More than 4000 generic LCI process datasets on energy supply,
resource extraction, material supply, chemicals, metals, agriculture,
waste management services, and transport services
A joint initiative of the
 Used by over 1200 members in more than 40 countries
ETH domain and Swiss
Federal Offices
 Included in the leading LCA software and eco-design tools
 Online access to LCI and LCIA results for all datasets
 Based on industry data, compiled by independent experts
 Consistent, validated and transparent
 Continuously maintained
 International in scope, including e.g. data on US agriculture, worldwide
sourcing of raw materials and production of electronics in Asia
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
11
Datasets for the biomass
production in ecoinvent: Overview
1. Datasets on agricultural means of
production: infrastructure (buildings and
machinery) and its usage, fertilisers,
pesticides, seed and animal feed
2. Datasets on agricultural and biomass
products:
•
•
•
•
Swiss Centre
For Life Cycle
Inventories
A joint initiative of the
ETH domain and Swiss
Federal Offices
Arable crop products
Grass
Wood
Fibres
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
12
relevant datasests available
How to establish life cycle inventories partly
of agricultural
products?
available
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
not available
Products Asia
Products America
Products Europe
Products CH
Inputs
Work processes
Machinery
Production branches
Arable crops
Fodder crops
Horticulture (Field)
Horticulture (Greenhouse)
Fruit growing
Vineyards
Cattle production
Pig production
Poultry production
Sheep production
Buildings
Contents of ecoinvent Version 2.1
What is covered in agriculture?
Swiss Centre
For Life Cycle
Inventories
A joint initiative of the
ETH domain and Swiss
Federal Offices
© ecoinvent centre, 2007
13
Contents of ecoinvent version 2.1
Datasets for biomass production
Category
agricultural means of production
agricultural means of production
agricultural means of production
agricultural means of production
agricultural means of production
agricultural means of production
agricultural means of production
agricultural means of production
agricultural production
agricultural production
biomass
wooden materials
wood energy
Total
Subcategory
Number of datasets
buildings
23
machinery
6
work processes
39
mineral fertiliser
24
organic fertiliser
5
pesticides
68
seed
26
feed
10
plant production
120
animal production
4
production
4
extraction
123
fuels
13
465
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
Swiss Centre
For Life Cycle
Inventories
A joint initiative of the
ETH domain and Swiss
Federal Offices
© ecoinvent centre, 2007
14
Contents of ecoinvent version 2.1
Crops and countries
Crops
barley
cotton
faba beans
fodder beets
grain maize
grass
grass silage
green manure
hay
hemp
jute
kenaf
oil palm
potato
protein peas
ramie
rape seed
rice
rye
silage maize
soy beans
sugar beets
sugar cane
sunflower
sweet sorghum
wheat
Cereals
Oil crops
Protein crops
Fibre crops
Grass
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
Countries
Brazil
Cameroon
China
Europe
France
Germany
Global
India
Malaysia
Philippines
Scandinavia
Spain
Switzerland
Thailand
USA
Swiss Centre
For Life Cycle
Inventories
A joint initiative of the
ETH domain and Swiss
Federal Offices
© ecoinvent centre, 2007
15
ecoinvent database: online access
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
16
Example: Unit Process Inventory (extract from V1.0)
wheat straw IP,
at farm CH (kg)
SD
wheat grains IP,
at farm CH (kg)
Location/
Uncert
Type
Unit process inventory for: wheat IP, CH
Exchanges
Category
Unit Value
ammonium nitrate, as N, at regional storehouse
RER
kg
6.71E+01
1 1.07 (2,1,1,1,1,na) 92%
8%
pesticide unspecified, at regional storehouse
CH
kg
2.60E-01
1 1.13 (2,2,3,1,1,na) 92%
8%
wheat seed IP, at regional storehouse
CH
kg
1.80E+02
1 1.07 (2,1,1,1,1,na) 92%
8%
tillage, ploughing
CH
ha
1.00E+00
1 1.07 (2,1,1,1,1,na) 92%
8%
grain drying, low temperature
CH
kg
7.64E+01
1 1.07 (2,1,1,1,1,na) 100%
resource/land
m2a 7.94E+03
1 1.77 (2,1,1,1,1,na) 92%
8%
Transformation, from pasture and meadow, intensive resource/land
m2
2.90E+03
1 2.67 (2,1,1,1,1,na) 92%
8%
Carbon dioxide, in air
resource/in air
kg
1.39E+04
1 1.07 (2,2,1,1,1,na) 61%
39%
Energy, gross calorific value, in biomass
resource/biotic
MJ
1.67E+05
1 1.07 (2,2,1,1,1,na) 59%
41%
95% Uncert Scores
...............................
...............................
Occupation, arable, non-irrigated
...............................
air/low population
Ammonia
density
kg
9.06E+00
1 1.30 (2,2,1,1,1,na) 92%
8%
Phosphorus
water/river
kg
2.58E-01
1 1.77 (2,2,1,1,1,na) 92%
8%
Nitrate
water/ground-
kg
1.25E+02
1 1.77 (2,2,1,1,1,na) 92%
8%
Isoproturon
soil/agricultural
kg
1.27E+00
1 1.32 (2,2,3,1,1,na) 92%
8%
Cadmium
soil/agricultural
kg
3.91E-03
1 1.77 (2,2,1,1,1,na) 42%
58%
wheat grains IP, at farm
CH
kg
6.42E+03
wheat straw IP, at farm
CH
kg
3.91E+03
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
100%
100%
© ecoinvent centre, 2003
17
Plans for the ecoinvent database
v.3.0 – release 2011
 Co-operation with national database initiatives
 More detail, more technologies, more completeness:
 International editorial board and broader supplier base
 Parameterisation (geography, time, technologies, markets)
 New data structure based on supply-use framework, allowing easier
production of national versions
 New indicators
 Sponsor-funded Open Access to individual datasets
 More frequent updating
 Improved uncertainty estimation and calculation facilities
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
18
New developments for ecoinvent V3.0:
International editorial board and
broader supplier base
 International editorial board
 Activity editors, for each industry activity and for household
activities
 Cross-cutting editors, to ensure consistency and monitor
developments across the entire database, both for specific
(groups of) emissions, for geographical areas, scenarios, etc.,
and for the meta-data fields, e.g. uncertainty
 Broader supplier base
 Making it easier for experts and lay users to contribute with new
data or corrections to existing data
 All such contributions will still be subject to our strict quality
control, review, and validation procedures before entering into
the database
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
19
New developments for ecoinvent V3.0:
Parameterisation
 Geographical parameters:
 Core international datasets + national differences
 Using GIS coordinates, all other area parameters can be
expressed: Country codes, areas with different population densities,
habitat areas, watershed areas, etc. for site-dependent impact
assessment
 Temporal parameters (years)
 Scenario parameters (e.g. BaU, optimistic, pessimistic)
 Dataset-internal parameters
 Inheritance using parent child-relationships
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
20
New developments for ecoinvent V3.0:
Better support for alternative modelling
options
Attributional and consequential modelling:
 Average versus marginal market modelling
 Allocation versus substitution (system expansion)
 Several versions of attributional allocation
 The unallocated (multi-functional) unit processes are the
same for both models
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
21
Estimating direct field and farm
emissions
 Usually no measurement on site possible
Two options:
 1. Literature values, experiments: take a value for a given
situation




 Specific for the situation
 Difficult to find
 Not flexible
 Mitigation options usually cannot be considered
 2. Modelling




 More flexible
 Mitigation options can be considered, depending on the model
 Level of detail should be consistent across the models
 No globally usable emission models available
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
22
Estimating direct field and farm
emissions
Ideal emission models should
 Reflect the underlying environmental mechanisms
 Be site and time dependent
 Consider the effect of soil and climate
 Consider the effect of management
 Be applicable under a wide range of different situations
 The different models should have a similar level of detail
 But also be usable:
 Parameters are measurable
 Data can be collected in a reasonable time
 Calculation is feasible
A compromise is needed!
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
23
SALCA emission models
Ammonia (NH3)
4 Emissions paths are modelled:
1. Application of farm manure = f(fertiliser amount, NH3 and
NH4-concentration, covered area, saturation deficit in the air
in function of average monthly temperature)
2. Application of mineral fertiliser = emission factors according
to fertiliser type (2-15%, Asman 1992)
3. Emission from pasture = 5% of total N in excrements
emitted as NH3
4. Emission from stable = emission factors dependent on
animal category, housing system, farm manure type (liquid
or solid)
Source: Menzi et al. (1997)
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© Agroscope ART, 2010
24
SALCA emission models
Nitrous oxide (N2O)
N2O in air: adapted method according to IPCC 2006, under
consideration of induced N2O-Emissions:
 Fertilisers: Direct emissions: 1% of available N
 Symbiotic N-fixation in legumes: no emissions
 Crop residues: emission factor 1%
 Storage of farmyard manure: emission factors 0.1% for liquid
manure and 2% for dung
 Pasture: emission factor 2%
 Induced Emissions: 1% of NH3-N and 0.75% of NO3-N
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© Agroscope ART, 2010
25
SALCA emission models
SALCA-nitrate
N mineralisation of soil
organic
matter
N uptake
plants
Non
leached N
Leaching
Leaching
+
Input of
mineral N
through
fertilisers
(NH4, NO3,
Amid-N)
GRUDAF:
Temperature dependent Example:
60 dt yield
80 dt yield
158 kg N uptake N-Uptake functions
211 kg N uptake
(STICS)
Monthly N-uptake
Source: Richner et al. (2006)
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© Agroscope ART, 2010
26
SALCA emission models
Methane (CH4)
 IPCC method 2 (Houghton et al. 1995)
 currently under revision
 Animal breading:
 Emissions from digestion = f(animal category, feeding)
 Emissions from storage of farm manure = f(animal category,
housing system)
 Emission factors:
Liquid manure: 10%
Dung and pasture: 1%
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© Agroscope ART, 2010
27
SALCA emission models
Phosphorus (P)
4 kinds of P-emissions in water:
•
•
•
•
Surface run-off in rivers (solved PO43-)
Drainage losses in rivers (solved PO43-)
Erosion in rivers (P bound to soil particles)
Leaching in ground water (solved PO43-)
Emissions are dependent of:
•
•
•
•
•
•
Soil characteristics (granulation, bulk density, soil water
balance) and drainage
Quantity of P-fertiliser
Type of P-fertiliser (manure, compost, mineral)
Field slope and distance to rivers
Quantity of eroded soil
Plant available P in upper soil
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
Source: Prasuhn (2006)
© Agroscope ART, 2010
28
SALCA emission models
Heavy metals
 Input-Output-Balance (caused by farmer) per field for:
Cd, Cu, Zn, Pb, Ni, Cr, Hg
 Inputs:
-
Fertilisers (mineral and organic)
Seed
Pesticides
Feedstuff and auxiliary materials for animal breeding
 Outputs:
-
Exported primary products (e.g. grains, meat)
Exported co-products (e.g. straw, animal manure)
Leaching to groundwater and drainage to surface water
Erosion to surface water
 Allocation for inputs caused by the farmer
Source: Freiermuth (2006)
 The final balance can be negative!
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© Agroscope ART, 2010
29
Variability and uncertainty: Factors
influencing environmental impacts
Socio-economic
conditions
Crop
management
Pedo-climatic
conditions
Crop yield
Life cycle
inventory
Environmental
impacts
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
To understand the
variability of
environmental
impacts, we need to
look on the
variability of the
influencing factors
© T. Nemecek ART, 2010
30
median
q25%
q2.5%
Yield [t/ha]
Global variability of yields
Example: potato
50
45
40
35
30
25
20
15
10
5
0
Cumulated potato world production as a function of the yield
0.0
20.0
40.0
60.0
80.0
100.0
Cumulated world production [%]
Source: FAOSTAT
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
31
Variability of environmental impacts:
Wheat datasets in ecoinvent V2.01 (2007)
US
w heat grains, at farm
Saxony, DE
0.76
Barrois, FR
0.63
CH, org
0.59
0.20
0.40
0.60
6.42
Barrois, FR
0.67
CH, IP
3.49
Castilla, ES
0.59
CH, ext
4.63
Saxony, DE
0.55
Castilla, ES
0.00
US
0.60
0.80
How to establish life cycle inventories of agricultural products?
factors| of
crop LCI/LCA
variability: exampleResearch
of wheat Station ART
T. Key
Nemecek
© Agroscope
Reckenholz-Tänikon
100a,
kg CO2-eq./kg
T. Nemecek | © GWP
Agroscope
Reckenholz-Tänikon
Research Station ART
3.58
CH, org
2.31
CH, ext
3.45
CH, IP
3.30
0.00
2.00
4.00
6.00
© ecoinvent centre 2007
w heat grains, at farm
8.00
energy demand, MJ-eq./kg
32
Variability of environmental impacts:
Energy demand per ha UAA (62 Swiss farms)
M J -E q .
Energy demand per ha UAA
300000
280000
260000
240000
220000
200000
180000
160000
140000
120000
100000
80000
60000
40000
20000
0
31 22 21 22 23 11 21 21 15 11 11 14 11 22 21 11 11 21 21 51 13 21 53 51 51 21 21 51 55 51 51 11 11 21 14 21 22 16 53 52 53 21 53 21 55 23 21 21 11 51 53 53 56 51 11 53 53 53
Fa r m ty pe
11
13
14
15
16
21
22
D es c ri pti on
a rab le fa rm ing
v e ge ta b le c u ltiv at io n
f ruit c ultiv at io n
v itic ultu re
o th er c u ltu re s
d airy f arm
s u c k ler c o ws
Fa rm ty pe
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
23
31
51
52
53
55
56
D es c rip tio n
o th er c at tle
h ors e s /g o at s /s h e ep
d airy fa rm / a ra ble fa rm in g c om bin ed
s u c k ler c ow s / a rab le fa rm ing c o m b ine d
p ig s a n d po u ltry / a ra ble fa rm in g c om bine d
d airy fa rm s / o th e r c o m b in e d
c a ttle / ot he r c o m b ine d
© Agroscope ART, 2010
33
Variability of environmental impacts:
Example: Energy demand per ha UAA (dairy farms)
Energy demand of dairy farms
80000
Eutrophication of dairy farms
70000
other inputs
350
emissions of animals
purchase of foodstuff
300
purchase of animals
50000
PPP
250
40000
kg N-Eq./ha UAA
MJ-Eq./ha UAA
60000
30000
20000
10000
200
150
100
other inputs
emissions of animals
purchase of foodstuff
fertiliser / nutrients
purchase of animals
seeds
PPP
energy carriers
fertiliser / nutrients
machines
seeds
buildings / equipment
energy carriers
machines
0
50
1
2
3
4
5
6
buildings / equipment
7
8
9
1
2
3
10 11 12 13 14 15
0
4
5
6
7
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
8
9
10
11
12
13
14
15
© Agroscope ART, 2010
34
(i)
Energy
use (MJ eq./$)
6.0
Variance control as a
basis for environmental
management
y = 3.79x - 0.46
r = 0.73, P = 0.007
4.0
M
2.0
 An balanced use of energy and
fertilisers improves ecoefficiency.
 The best farms (1, 2) had the
lowest pesticide use per area
unit.
The orchards have high yields
(high labour input) and a good
physiological and ecological
equilibrium.
Farm
No. 1
Farm
No. 2
0.0
0%
20%
40%
60%
80%
100%
120%
140%
160%
Aq. (iii)
Eutrophication (PO4 eq./$)
0.15
y = 0.09x - 0.01
r = 0.77, P = 0.003
0.10
M
0.05
Farm
No. 1
Farm
No. 2
0.00
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
0%
20%
40%
60%
80%
100%
120%
Coefficent of Variance
140%
160%
35
Source: Mouron et al. (2006)
16
0.8
14
0.7
12
0.6
10
0.5
8
0.4
6
0.3
NH3 emission (kg NH3/ha)
4
0.2
Relative emission rate (kg NH3-N/kg TAN)
2
0.1
0
kg NH3-N/kg TAN
kg NH3/ha
Variability and non-linearity
Averages may lead to wrong results
1x40 m3 slurry
 13.5 kg NH3
2x20 m3 slurry
 17.4 kg NH3
0.0
0
10
20
30
m3 of slurry
40
50
Ammonia emission as a function of quantity of slurry applied.
TAN = total ammonia N in the slurry (after Menzi et al. 1997)
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© T. Nemecek ART, 2010
36
Uncertainty assessment in ecoinvent
V2.1: Pedigree matrix
Indicator score
Reliability
Completeness
Temporal
correlation
Geographical
correlation
Further
technological
correlation
Sample size
1
2
3
4
Qualified estimate (e.g.
Verified data partly
by industrial expert); data
Non-verified data partly
Verified data based on based on assumptions
derived from theoretical
based on qualified
measurements
OR non-verified data
information
estimates
based on measurements
(stoichiometry, enthalpy,
etc.)
Representative data
Representative data
Representative data
Representative data
from all sites relevant for from >50% of the sites from only some sites
from only one site
relevant for the market (<<50%) relevant for the relevant for the market
the market considered
market considered OR considered OR some
over an adequate period considered over an
adequate period to even >50% of sites but from
to even out normal
sites but from shorter
out normal fluctuations shorter periods
fluctuations
periods
5
Non-qualified estimate
Remarks
verified means: published in public
environmental reports of companies, official
statistics, etc
unverified means: personal information by
letter, fax or e-mail
Representativeness
unknown or data from a
Length of adequate period depends on
small number of sites
process/technology
AND from shorter
periods
less than 3 years means: data measured in
1997 or later;
Less than 15 years of
Less than 10 years of
Less than 6 years of
Less than 3 years of
score for processes with investment cycles
difference to our
difference to our
difference to our
difference to our
of <10 years;
reference year (2000)
reference year (2000)
reference year (2000)
reference year (2000)
for other cases, scoring adjustments can be
made accordingly
Similarity expressed in terms of
Data from unknown OR enviornmental legislation. Suggestion for
grouping:
distinctly different area
Average data from larger Data from smaller area
(north america instead of North America, Australia;
Data from area under
area in which the area
than area under study, or
European Union, Japan, South Africa;
middle east, OECDstudy
under study is included from similar area
South America, North and Central Africa
Europe instead of
and Middle East;
Russia)
Russia, China, Far East Asia
Examples for different technology:
Data on related
- steam turbine instead of motor propulsion
processes or materials Data on related
in ships
processes or materials Data on related
but same technology,
Data from enterprises,
- emission factor B(a)P for diesel train
but different technology, processes or materials
OR
processes and materials
based on lorry motor data
but on laboratory scale of
OR data on laboratory
Data from processes
under study (i.e. identical
Examples for related processes or
different technology
scale processes and
and materials under
technology)
materials:
study but from different same technology
- data for tyles instead of bricks production
technology
- data of refinery infrastructure for chemical
>100, continous
> 10, aggregated figure
sample size behind a figure reported in the
>=3
unknown
measurement, balance >20
in env. report
information source
of purchased products
Age of data unknown or
more than 15 years of
difference to our
reference year (2000)
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© ecoinvent centre, 2007
37
Uncertainty assessment for French
wheat
95% confidence interval
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© T. Nemecek ART, 2010
38
Potential use of multivariate statistics
in LCA  explain variability
 Multivariate statistics (like principal component analysis,
PCA) can be used to show similarities between
environmental impacts
 It can be also used to group environmental profiles, e.g.
of crops
 Analysis based on a set of midpoint LCIA indicators
 In the study applied to crop inventories from SALCA
(Switzerland) and ecoinvent (global)
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
39
Principal component analysis of
SALCA inventories
Eigenvalues of correlation matrix
Active variables only
5.0
4.5
52.73%
4.0
3.5
Eigenvalue
3.0
2.5
27.63%
2.0
1.5
1.0
6.18% 5.07%
4.51%
0.5
2.24%
.94%
.70%
0.0
-0.5
-1
0
1
2
3
4
5
6
7
8
9
10
Eigenvalue number
80% of variance explained by first two principal components
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© T. Nemecek ART, 2010
40
Principal component analysis of
SALCA inventories
Projection of the variables on the factor-plane ( 1 x 2)
1.0
Acidi
Eutro
0.5
Factor 2 : 27.63%
GWP
Ozone
0.0
AET_EDIP
Energy
HTP_CML
-0.5
TET_EDIP
-1.0
-1.0
-0.5
0.0
0.5
1.0
Factor 1 : 52.73%
Relationship between impact indicators and factors 1 and 2
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© T. Nemecek ART, 2010
41
Factor 1:
- can group crops
- related to the yield
5
Data for Swiss crops
from SALCA database:
grouping by crop group
(CER = cereals,
LEG = legumes,
MAI = maize,
OIL = oil crops,
ROOT = root crops,
VEG = vegetables).
4
3
Factor 2
2
1
0
-1
-2
-3
-4
-6
-4
-2
0
2
4
6
CER
LEG
MAI
OIL
ROOT
VEG
Factor 1
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© T. Nemecek ART, 2010
42
Factor 2:
- related to the farming system and the
intensity
5
4
Data for Swiss crops from
SALCA database: grouping
by farming system
(Conv=conventional,
IPint = integrated intensive,
IPext = integrated extensive,
Org = organic).
3
2
Factor 2
1
0
-1
-2
-3
-4
-6
-4
-2
0
2
4
6
Conv
Ipint
Ipext
Org
Factor 1
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© T. Nemecek ART, 2010
43
Principal component analysis of
SALCA inventories
Scatterplot (FALSR58_Res 14v*246c)
Factor 1 = -5.9426-2.1271*x
2
1
0
Factor 1
-1
-2
-3
-4
-5
-6
-7
-3.0
-2.8
-2.6
-2.4
-2.2
-2.0
LnInvYield:Factor 1: r2 = 0.4561
-1.8
-1.6
-1.4
-1.2
-1.0
-0.8
-0.6
-0.4
LnInvYield
Yield is a key factor
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© T. Nemecek ART, 2010
44
Principal component analysis of
ecoinvent inventories
3.0
2.5
2.0
1.5
1.0
0.5
0.0
-0.5
-1.0
-1.5
-2.0
-2.5
-8
-6
-4
-2
0
2
4
6
CER
FIB
LEG
MAI
OIL
ROOT
Factor 1
Effect of the crop group (factor 1)
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© T. Nemecek ART, 2010
45
Principal component analysis of
ecoinvent inventories
3.0
2.5
2.0
1.5
1.0
0.5
0.0
-0.5
-1.0
-1.5
-2.0
-2.5
-8
-6
-4
-2
0
2
4
6
Conv
IPint
IPext
Org
Factor 1
Effect of the farming system (factor 2)
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© T. Nemecek ART, 2010
46
Principal component analysis of
ecoinvent inventories
3
CHCH
CH
2
1
CH
CH
CH
CH
CH
CH
0
ESES
-1
FRFR
US
DE
DE
RER
-2
-8
-6
-4
-2
0
2
4
6
w heat
barley
rye
Factor 1
Cereals in different countries
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© T. Nemecek ART, 2010
47
Potential use of multivariate statistics
in LCA to explain variability
 Between 76 and 80% of the variability could be explained by the first
two principal components.
 Factor 1  crop (group) and yield
 Factor 2  farming system (conventional, integrated, extensive,
organic)
 More data are needed for more systematic analyses
 The analysis helps to
 show similarities and differences between environmental profiles
 to find suitable proxies
 to derive simplified methods for extrapolations and approximations
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
48
How to fill data gaps in agricultural
LCI?
The classical approach:
1. Establish detailed and specific inventories for each
situation
Currently used alternatives:
2. Use proxies: what you think is the closest LCI
(generalisation)
3. Streamlined LCA models
New approaches:
4. Extrapolation by yield correction
5. Modular extrapolation method
 geographical extrapolation
 product extrapolation
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
49
Extrapolation by yield correction
c'
c'
E
E
 Product extrapolation: E cp  e p  ac '  (1  e p )  ac
Y
Y
l'
l'
E
E
l
a
a
 Geographical extrapolation: E p  e p  l '  (1  e p )  l
Y
Y
Impacts related to the yield (constant per kg)
Impacts not related to the yield (constant per ha)
ep 
Fraction of the impacts related to the yield
Estimation of this fraction:
• 0.7 for cereals from the ecoinvent datasets
• 0.5 as default value
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© Roches & Nemecek ART, 2010
50
Modular EXtrapolation for
Agricultural LCA (MEXALCA)
Basic idea:
 It is possible to split an inventory into different independent
modules.
 This enables easier adaptation of an existing inventory to a new
situation.
Working procedure:
1.
2.
3.
4.
Establish a base inventory for one or several typical situations
Split the inventory into independent modules
Calculate unit inventories/impacts per module and input unit
Determine amount of inputs used in each country (using global
estimators derived from FAOSTAT)
5. Extrapolate inventory to any producing country
6. Estimate global/regional impacts (medians, means, distribution)
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
51
Extrapolation using MEXALCA
Impacts per input unit
Base crop inventory
Basic cropping operations
Soil tillage
Variable machinery operations
N fertilisation, including N-emissions
P fertilisation, including P-emissions
K fertilisation
Pesticide application
Irrigation
Product drying
Splitting
Calculation of unit impacts
Input parameters:
•yield per area unit
•Mechanisation index
•% of no-till area
•kg N, P2O5, K2O applied
•kg pesticide active ingredient
•m3 water used
•kg water evaporated
Global estimators
(based on FAOSTAT)
Good quality data available
for some or all inputs
Total impacts
for extrapolated
country x
GWP 100 a [kg CO2-eq/kg]
0.3
Global distribution of impacts
0.25
Extrapolation
0.2
Total impacts
for extrapolated
country y
0.15
0.1
Impacts for extrapolated situation

Total impacts
0.05
How to establish life cycle inventories of agricultural products?
extrapolated
T. Nemecek | © Agroscope Reckenholz-Tänikon Research for
Station
ART
0
0%
20%
40%
60%
80%
Percentage of the world potato production
100%
52
country z
© T. Nemecek ART, 2010
MEXALCA results:
impacts per input unit
Modules
Potatoes
Impacts
MachFix
MachTill MachVar Nfert
13604.50
1818.25
1074.68
photochemic O3 formation [kg ethylene-eq]
non-renewable Energy [MJ-eq]
Pfert
Kfert Pestic Irrigat Drying
4621.45 70.91 31.26 10.69 341.5 9.988
0
118.49
272.66
0.614 15.127 0.247
0
0.65
0.08
0.23
0.001 6E-04 2E-04 0.0092 2E-04
0
Nutrient enrichment [kg N-eq]
12.65
0.34
0.60
0.917 0.126 7E-04 0.023 2E-04
0
Acidification [kg SO2-eq]
9.38
0.95
1.80
0.282 0.039 0.003 0.099 9E-04
0
Aquatic ecotoxicity 100a [kg 1,4-DCB-eq]
56.92
0.13
0.45
0.015 0.404 0.007 114.99 4E-04
0
Terrestrial ecotoxicity 100a [kg 1,4-DCB-eq]
0.99
0.01
0.05
7E-04 0.009 3E-04 80.696 1E-04
0
460.52
38.32
209.11
1.216 0.97 0.337 337.68 0.181
0
GWP 100a [kg CO2-eq]
Human toxicity 100a [kg 1,4-DCB-eq]
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
13.45
2
© Roches & Nemecek ART, 2010
53
MEXALCA results:
impacts per kg of potato in the world
QUANTILES
Energy [MJ-eq]
GWP [kg CO2-eq]
O3 form. [kg ethylene-eq]
IMPACTS Nutr. enrich. [kg N-eq]
Acidific. [kg SO2-eq]
Aquat. Ecotox.[kg 1,4-DCB-eq]
Terr. Ecotox. [kg 1,4-DCB-eq]
Human tox.[kg 1,4-DCB-eq]
2.5%
9.11E-01
7.38E-02
2.84E-05
1.85E-03
9.44E-04
10.0%
9.77E-01
8.58E-02
3.13E-05
1.92E-03
1.14E-03
6.91E-02
6.96E-02
25.0%
median
75.0%
1.27E+00 1.72E+00 3.00E+00
1.11E-01 1.23E-01 1.91E-01
4.75E-05 6.59E-05 8.50E-05
2.41E-03 3.44E-03 5.54E-03
1.23E-03 1.49E-03 2.27E-03
1.18E-02 1.65E-02 2.30E-02
5.41E-03 9.15E-03 1.26E-02
7.26E-02 8.34E-02 1.01E-01
90.0%
3.05E+00
1.92E-01
8.53E-05
5.61E-03
2.30E-03
3.06E-02
1.89E-02
1.40E-01
97.5%
4.15E+00
2.05E-01
1.07E-04
7.52E-03
2.82E-03
5.24E-02
3.50E-02
2.00E-01
The modular inventory system enables us to calculate the inputs and
impacts in any producing country and to calculate median and
quantiles for the inputs and for the impacts for the global production
(per kg of product or per cultivated ha).
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© Roches & Nemecek ART, 2010
54
Results: estimated distribution of GWP
of the potato production
GWP 100 a [kg CO2-eq/kg]
0.3
0.25
0.2
0.15
0.1
0.05
0
0%
20%
40%
60%
80%
100%
Percentage of the world potato production
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
© Roches & Nemecek ART, 2010
55
First validation: impacts per kg
Global Warming Potential 100 years [kg CO2-eq]
Acidification [kg SO2-eq]
Colours
barley
wheat
ry e
potato
pea
2
4
6
8
10
0.2
0.4
0.6
0.8
1.0
1. 2
ecoinvent
ec oinvent
Photochemical ozone formation [kg ethylene-eq]
Nutrient enrichment [kg N-eq]
r2  0.43 5
0.006
0. 6
r2  0.493
y  1.1 03x 0.00 2
Colours
b arley
whea t
ry e
p otat o
p ea
0.002
r 2  0.493
modular inventory
1. 0
y  0.386x 0.198
0. 2
4
6
r2  0.796
modular inv entory
8
y  1.163 x -0.042
2
modular inventory
10
0.010
Non renewable energy demand [MJ-eq]
0.002
0.004
0.006
0.008
0.010
barley
wheat
ry e
pot ato
pea
0.00005
0.00015
0. 00025
0. 02
0. 03
y  - 0.158x 0.021
r 2  0.022
Colours
barley
wheat
ry e
potato
pea
0. 01
Colours
modular inventory
0.00020
y  0.973x 0
r2  0.939
0.00005
modular inventory
0. 04
ecoinvent
0.01
0. 02
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
0.03
0.04
56
© Roches & Nemecek ART, 2010
Sensitivity analysis
 Performed considering the median (=q50%), q10% and q90% of
each input (estimated variability of the inputs)
POT AT O
Q uantiles
IMPACTS
non-renewable energy [MJ-eq]
G WP 100a [k g CO 2-eq]
photo. oz one formation [kg ethylene-Eq]
nutrient enrichm ent [kg N -eq]
Acidification [kg SO2-Eq]
Aquatic ecotoxic it y, 100a [kg 1,4-DCB-Eq]
Terres trial ec otoxic it y, 100a [kg 1,4-DCB-Eq]
Human toxicity, 100a [k g 1, 4-DC B-Eq]
Variation: 5 to 10%
MachVar
Nfer t
q10%
q90% q10%
-1%
-1%
-1%
0%
0%
0%
0%
-1%
7%
5%
11%
0%
3%
0%
0%
7%
-11%
-28%
-5%
-64%
-47%
0%
0%
-4%
Variation: 10 to 50%
q90%
22%
55%
11%
125%
93%
1%
0%
8%
INPUTS
Pfert
K fert
Pestic
Irrig at
q10% q90% q10% q90% q10% q90% q10%
-2%
-2%
-1%
-4%
-3%
-3%
0%
-1%
3%
2%
1%
5%
3%
4%
0%
2%
-1%
-1%
-1%
0%
0%
0%
0%
-1%
Variation: 50 to 100%
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
4% -2%
7% -27%
3% -1%
4%
-9%
3% -2%
6% -15%
0%
0%
0%
0%
1% -1%
2%
-3%
0% -76% 288%
0%
0% -99% 377%
0%
3% -43% 165% -11%
Drying
q90% q10% q90%
62%
21%
34%
1%
6%
0%
0%
25%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
0%
Variation: > 100%
© Roches & Nemecek ART, 2010
57
Potentials of extrapolation
 Extrapolation cannot replace data collection and the establishment of
detailed and specific inventories
 Very important time saving possible
 Allows to create generic data sets on global and multinational level
 Assessment of global variability
 Fairly good estimates possible for energy demand, global warming and
ozone formation, land occupation
 Difficult for eutrophication and acidification (no site-specific parameters
considered) and toxicity (no detailed information on pesticide active
ingredients)
 Can be used as first approximation and where ingredients is not so
relevant
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
58
SALCA: An integrated concept for
agricultural environmental assessment
SALCA = Swiss Agricultural Life Cycle Assessment
SALCA consists of the following elements:
 Database for life cycle inventories for agriculture (in collaboration
with ecoinvent)
 Models for the calculation of direct emissions from field and farm
 A selection of impact assessment methods (midpoints)
 Methods for the assessment of impacts on biodiversity and soil
quality
 Calculation tools for agricultural systems (farm, crop)
 Interpretation scheme for agricultural LCA
 Communication concept for the environmental management of
farms
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
59
SALCA calculation tools
 Large variability  large number of calculations
 automation required
 Generic parametrised system modelling for farms and crops:
 SALCA-farm: generic LCA system for farms
 SALCA-crop: generic LCA system for arable crops and forage
production systems
 The templates are designed in order to cover all farms/crops
 All elements, which occur in at least one system must be
included
 Variables are defined, which can describe the different
quantities of inputs
 The variables that are not relevant for a particular system are
set to zero
 Modular structure
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
60
Modular architecture of the tool SALCA-crop V3.1
Produktionsinventar.xls
Input data
SALCA-field
SALCA-Nitrate
Calculations
SALCA-Field
(other direct
emissions)
Calculations
Input data
SALCA-nitrate
SALCA
(TEAM/SimaPro)
LCI Calculations
SALCA-Erosion
Calculations
Data transfer by macros
Input data
SALCA-erosion
Input data
SALCA-soil quality
Input data
SALCA
(TEAM/SimaPro)
SALCA-Heavy
metals
Calculations
SALCA-soil
quality
Calculations
SALCA
(TEAM/SimaPro)
LCIA Calculation
SALCA-biodiversity
Data entry
Calculations
(separate tool)
61
How
to establish
cycle inventories of agricultural products?
Transfer
LCIlife
data
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
6 separate tools in
EXCEL: data
entry can be done
through the
common
production
inventory or
directly in the tool
LIFE CYCLE
IMPACT
ASSESSMENT
(LCIA)
Input data SALCA
heavy metals
Production
inventory:
Common data entry of
all parameters for all
tools
LIFE CYCLE
INVENTORY (LCI)
Internal Links in EXCEL-sheet
Data entry
© R. Freiermuth, T. Nemecek, ART 2010
Specific aspects of tropical production
systems: relevant LCI aspects
 Less managed production
 higher variability
 more dependent on the environment
 Labour input instead of machinery
 how to consider manpower?
 Use of draught animals  how to consider?
 Reconsider the delimitation between plant and animal
production
 Adaptation of emission models to the conditions of the tropics
and subtropics (soil, climate)
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
62
Recommendations for agricultural LCI
 Large variability  many observations needed
 Collect detailed farm management data
 Standardised methodology
 Automated calculation
 Use of standard LCI formats (EcoSpold, ILCD)
 Need for a standardised format for agricultural
management data
 Regionalisation, use of GIS
 Variability and uncertainty should be assessed as
standard
 Infrastructure should be included
 Development of globally applicable emission models
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
63
Thanks to
 My colleagues: Gérard Gaillard, Ruth Freiermuth,
Martina Alig, Daniel, Baumgartner, Anne Roches,
Katharina Plassmann
 Ecoinvent centre
 Unilever: Llorenç Milà i Canals, Sarah Sim, Tirma
Garcia-Suarez
 You for your kind attention!
How to establish life cycle inventories of agricultural products?
T. Nemecek | © Agroscope Reckenholz-Tänikon Research Station ART
64